diff --git "a/Understanding/meta/val/grasppose_choice.json" "b/Understanding/meta/val/grasppose_choice.json" new file mode 100644--- /dev/null +++ "b/Understanding/meta/val/grasppose_choice.json" @@ -0,0 +1,362070 @@ +[ + { + "task_id": "0000007", + "id": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_750612070853#262", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Insert the tip of a medium pipette into the holder for medium pipette tips'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000007/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 423, + 147 + ], + [ + 393, + 121 + ], + [ + 405, + 180 + ], + [ + 440, + 116 + ], + [ + 434, + 205 + ], + [ + 467, + 141 + ] + ], + [ + [ + 414, + 120 + ], + [ + 410, + 79 + ], + [ + 375, + 115 + ], + [ + 456, + 125 + ], + [ + 381, + 149 + ], + [ + 458, + 163 + ] + ], + [ + [ + 419, + 165 + ], + [ + 411, + 200 + ], + [ + 397, + 138 + ], + [ + 439, + 190 + ], + [ + 406, + 105 + ], + [ + 445, + 158 + ] + ], + [ + [ + 452, + 131 + ], + [ + 450, + 160 + ], + [ + 457, + 98 + ], + [ + 448, + 160 + ], + [ + 459, + 72 + ], + [ + 451, + 134 + ] + ] + ] + }, + { + "task_id": "0000011", + "id": "RH20T_cfg3_task_0059_user_0010_scene_0004_cfg_0003_104122062823#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000011/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 246, + 93 + ], + [ + 250, + 65 + ], + [ + 274, + 99 + ], + [ + 218, + 86 + ], + [ + 269, + 124 + ], + [ + 215, + 112 + ] + ], + [ + [ + 268, + 73 + ], + [ + 261, + 100 + ], + [ + 291, + 80 + ], + [ + 247, + 67 + ], + [ + 298, + 52 + ], + [ + 253, + 40 + ] + ], + [ + [ + 247, + 82 + ], + [ + 237, + 77 + ], + [ + 272, + 67 + ], + [ + 222, + 96 + ], + [ + 279, + 72 + ], + [ + 232, + 99 + ] + ], + [ + [ + 231, + 61 + ], + [ + 240, + 42 + ], + [ + 245, + 48 + ], + [ + 217, + 72 + ], + [ + 236, + 69 + ], + [ + 208, + 92 + ] + ] + ] + }, + { + "task_id": "0000012", + "id": "RH20T_cfg3_task_0059_user_0010_scene_0004_cfg_0003_104122062823#204", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000012/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 245, + 109 + ], + [ + 242, + 117 + ], + [ + 236, + 84 + ], + [ + 255, + 134 + ], + [ + 239, + 74 + ], + [ + 259, + 126 + ] + ], + [ + [ + 260, + 112 + ], + [ + 270, + 105 + ], + [ + 236, + 95 + ], + [ + 283, + 128 + ], + [ + 228, + 103 + ], + [ + 273, + 133 + ] + ], + [ + [ + 245, + 89 + ], + [ + 250, + 63 + ], + [ + 270, + 89 + ], + [ + 218, + 90 + ], + [ + 266, + 112 + ], + [ + 215, + 114 + ] + ], + [ + [ + 227, + 100 + ], + [ + 233, + 111 + ], + [ + 208, + 81 + ], + [ + 245, + 119 + ], + [ + 203, + 72 + ], + [ + 238, + 109 + ] + ] + ] + }, + { + "task_id": "0000016", + "id": "RH20T_cfg1_task_0015_user_0008_scene_0008_cfg_0001_035622060973#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000016/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 184, + 134 + ], + [ + 189, + 147 + ], + [ + 195, + 146 + ], + [ + 172, + 120 + ], + [ + 189, + 134 + ], + [ + 167, + 108 + ] + ], + [ + [ + 202, + 139 + ], + [ + 186, + 136 + ], + [ + 190, + 136 + ], + [ + 215, + 143 + ], + [ + 206, + 139 + ], + [ + 231, + 146 + ] + ], + [ + [ + 183, + 136 + ], + [ + 181, + 124 + ], + [ + 162, + 138 + ], + [ + 204, + 135 + ], + [ + 165, + 149 + ], + [ + 206, + 146 + ] + ], + [ + [ + 175, + 133 + ], + [ + 174, + 118 + ], + [ + 173, + 145 + ], + [ + 178, + 120 + ], + [ + 174, + 161 + ], + [ + 180, + 137 + ] + ] + ] + }, + { + "task_id": "0000017", + "id": "RH20T_cfg1_task_0015_user_0008_scene_0008_cfg_0001_035622060973#217", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000017/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 192, + 184 + ], + [ + 196, + 166 + ], + [ + 171, + 180 + ], + [ + 212, + 188 + ], + [ + 167, + 199 + ], + [ + 207, + 206 + ] + ], + [ + [ + 153, + 197 + ], + [ + 162, + 177 + ], + [ + 149, + 196 + ], + [ + 157, + 199 + ], + [ + 140, + 216 + ], + [ + 147, + 220 + ] + ], + [ + [ + 166, + 191 + ], + [ + 145, + 185 + ], + [ + 161, + 212 + ], + [ + 171, + 171 + ], + [ + 181, + 218 + ], + [ + 191, + 178 + ] + ], + [ + [ + 179, + 195 + ], + [ + 176, + 182 + ], + [ + 176, + 212 + ], + [ + 181, + 178 + ], + [ + 179, + 223 + ], + [ + 184, + 189 + ] + ] + ] + }, + { + "task_id": "0000018", + "id": "RH20T_cfg6_task_0038_user_0020_scene_0001_cfg_0006_104122061018#531", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000018/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 308, + 142 + ], + [ + 295, + 116 + ], + [ + 301, + 159 + ], + [ + 316, + 122 + ], + [ + 315, + 186 + ], + [ + 331, + 151 + ] + ], + [ + [ + 337, + 142 + ], + [ + 368, + 139 + ], + [ + 336, + 111 + ], + [ + 338, + 174 + ], + [ + 307, + 114 + ], + [ + 308, + 176 + ] + ], + [ + [ + 317, + 120 + ], + [ + 307, + 95 + ], + [ + 292, + 111 + ], + [ + 345, + 131 + ], + [ + 302, + 133 + ], + [ + 353, + 154 + ] + ], + [ + [ + 310, + 126 + ], + [ + 307, + 103 + ], + [ + 327, + 147 + ], + [ + 293, + 107 + ], + [ + 328, + 168 + ], + [ + 296, + 128 + ] + ] + ] + }, + { + "task_id": "0000021", + "id": "RH20T_cfg1_task_0041_user_0001_scene_0006_cfg_0001_038522063145#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the tabletop with a sponge'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000021/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 401, + 125 + ], + [ + 384, + 136 + ], + [ + 390, + 127 + ], + [ + 414, + 124 + ], + [ + 406, + 116 + ], + [ + 431, + 112 + ] + ], + [ + [ + 389, + 145 + ], + [ + 404, + 129 + ], + [ + 388, + 152 + ], + [ + 391, + 137 + ], + [ + 373, + 167 + ], + [ + 375, + 153 + ] + ], + [ + [ + 412, + 135 + ], + [ + 394, + 135 + ], + [ + 425, + 135 + ], + [ + 398, + 135 + ], + [ + 444, + 135 + ], + [ + 417, + 136 + ] + ], + [ + [ + 410, + 140 + ], + [ + 414, + 123 + ], + [ + 426, + 153 + ], + [ + 394, + 128 + ], + [ + 422, + 168 + ], + [ + 391, + 143 + ] + ] + ] + }, + { + "task_id": "0000022", + "id": "RH20T_cfg1_task_0015_user_0001_scene_0007_cfg_0001_750612070851#266", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000022/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 325, + 209 + ], + [ + 364, + 213 + ], + [ + 330, + 197 + ], + [ + 319, + 223 + ], + [ + 295, + 193 + ], + [ + 278, + 219 + ] + ], + [ + [ + 330, + 230 + ], + [ + 332, + 213 + ], + [ + 336, + 267 + ], + [ + 324, + 197 + ], + [ + 333, + 280 + ], + [ + 323, + 213 + ] + ], + [ + [ + 350, + 207 + ], + [ + 338, + 239 + ], + [ + 374, + 209 + ], + [ + 324, + 206 + ], + [ + 386, + 177 + ], + [ + 336, + 170 + ] + ], + [ + [ + 333, + 199 + ], + [ + 350, + 165 + ], + [ + 369, + 216 + ], + [ + 298, + 183 + ], + [ + 352, + 253 + ], + [ + 280, + 218 + ] + ] + ] + }, + { + "task_id": "0000024", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122061330#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000024/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 311, + 138 + ], + [ + 290, + 125 + ], + [ + 310, + 123 + ], + [ + 313, + 151 + ], + [ + 333, + 136 + ], + [ + 333, + 163 + ] + ], + [ + [ + 319, + 142 + ], + [ + 328, + 149 + ], + [ + 329, + 165 + ], + [ + 309, + 121 + ], + [ + 320, + 158 + ], + [ + 299, + 112 + ] + ], + [ + [ + 328, + 156 + ], + [ + 329, + 138 + ], + [ + 317, + 174 + ], + [ + 339, + 139 + ], + [ + 316, + 190 + ], + [ + 338, + 155 + ] + ], + [ + [ + 316, + 145 + ], + [ + 290, + 144 + ], + [ + 312, + 158 + ], + [ + 319, + 133 + ], + [ + 339, + 160 + ], + [ + 344, + 134 + ] + ] + ] + }, + { + "task_id": "0000028", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_037522061512#98", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000028/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 280, + 168 + ], + [ + 274, + 154 + ], + [ + 291, + 150 + ], + [ + 269, + 187 + ], + [ + 296, + 164 + ], + [ + 275, + 200 + ] + ], + [ + [ + 293, + 160 + ], + [ + 314, + 167 + ], + [ + 288, + 145 + ], + [ + 297, + 173 + ], + [ + 267, + 138 + ], + [ + 278, + 166 + ] + ], + [ + [ + 280, + 155 + ], + [ + 294, + 173 + ], + [ + 287, + 151 + ], + [ + 274, + 159 + ], + [ + 272, + 132 + ], + [ + 260, + 141 + ] + ], + [ + [ + 279, + 165 + ], + [ + 294, + 166 + ], + [ + 272, + 141 + ], + [ + 286, + 188 + ], + [ + 258, + 142 + ], + [ + 272, + 186 + ] + ] + ] + }, + { + "task_id": "0000030", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_037522061512#294", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000030/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 199, + 145 + ], + [ + 211, + 127 + ], + [ + 180, + 133 + ], + [ + 216, + 157 + ], + [ + 167, + 152 + ], + [ + 204, + 176 + ] + ], + [ + [ + 208, + 152 + ], + [ + 200, + 136 + ], + [ + 221, + 136 + ], + [ + 195, + 169 + ], + [ + 228, + 151 + ], + [ + 204, + 183 + ] + ], + [ + [ + 222, + 149 + ], + [ + 236, + 153 + ], + [ + 239, + 143 + ], + [ + 205, + 156 + ], + [ + 226, + 140 + ], + [ + 192, + 152 + ] + ], + [ + [ + 217, + 183 + ], + [ + 199, + 183 + ], + [ + 232, + 193 + ], + [ + 202, + 174 + ], + [ + 248, + 193 + ], + [ + 219, + 174 + ] + ] + ] + }, + { + "task_id": "0000032", + "id": "RH20T_cfg3_task_0018_user_0011_scene_0002_cfg_0003_104422070011#196", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000032/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 325, + 169 + ], + [ + 348, + 157 + ], + [ + 339, + 190 + ], + [ + 311, + 149 + ], + [ + 315, + 203 + ], + [ + 288, + 162 + ] + ], + [ + [ + 310, + 189 + ], + [ + 318, + 178 + ], + [ + 322, + 212 + ], + [ + 297, + 166 + ], + [ + 314, + 220 + ], + [ + 291, + 177 + ] + ], + [ + [ + 300, + 162 + ], + [ + 303, + 135 + ], + [ + 292, + 161 + ], + [ + 309, + 165 + ], + [ + 288, + 186 + ], + [ + 305, + 193 + ] + ], + [ + [ + 346, + 190 + ], + [ + 368, + 181 + ], + [ + 363, + 200 + ], + [ + 331, + 180 + ], + [ + 339, + 210 + ], + [ + 309, + 189 + ] + ] + ] + }, + { + "task_id": "0000033", + "id": "RH20T_cfg1_task_0030_user_0015_scene_0001_cfg_0001_750612070853#251", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour out the test tube'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000033/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 598, + 329 + ], + [ + 558, + 270 + ], + [ + 523, + 360 + ], + [ + 640, + 292 + ], + [ + 561, + 360 + ], + [ + 640, + 351 + ] + ], + [ + [ + 640, + 296 + ], + [ + 640, + 211 + ], + [ + 640, + 315 + ], + [ + 549, + 282 + ], + [ + 640, + 360 + ], + [ + 540, + 360 + ] + ], + [ + [ + 622, + 275 + ], + [ + 577, + 242 + ], + [ + 640, + 323 + ], + [ + 579, + 240 + ], + [ + 640, + 356 + ], + [ + 617, + 268 + ] + ], + [ + [ + 598, + 343 + ], + [ + 549, + 280 + ], + [ + 561, + 360 + ], + [ + 640, + 268 + ], + [ + 617, + 360 + ], + [ + 640, + 353 + ] + ] + ] + }, + { + "task_id": "0000035", + "id": "RH20T_cfg1_task_0070_user_0015_scene_0002_cfg_0001_750612070851#178", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000035/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 382, + 275 + ], + [ + 385, + 261 + ], + [ + 345, + 277 + ], + [ + 420, + 273 + ], + [ + 345, + 290 + ], + [ + 415, + 286 + ] + ], + [ + [ + 368, + 315 + ], + [ + 371, + 353 + ], + [ + 406, + 310 + ], + [ + 332, + 320 + ], + [ + 401, + 275 + ], + [ + 330, + 285 + ] + ], + [ + [ + 399, + 299 + ], + [ + 396, + 264 + ], + [ + 393, + 288 + ], + [ + 404, + 308 + ], + [ + 397, + 328 + ], + [ + 407, + 343 + ] + ], + [ + [ + 382, + 286 + ], + [ + 378, + 251 + ], + [ + 398, + 271 + ], + [ + 368, + 299 + ], + [ + 403, + 311 + ], + [ + 372, + 335 + ] + ] + ] + }, + { + "task_id": "0000038", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_819112070044#106", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000038/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 384, + 170 + ], + [ + 372, + 160 + ], + [ + 386, + 156 + ], + [ + 383, + 183 + ], + [ + 400, + 168 + ], + [ + 395, + 194 + ] + ], + [ + [ + 394, + 149 + ], + [ + 410, + 155 + ], + [ + 385, + 151 + ], + [ + 404, + 147 + ], + [ + 369, + 145 + ], + [ + 387, + 141 + ] + ], + [ + [ + 387, + 173 + ], + [ + 390, + 159 + ], + [ + 400, + 165 + ], + [ + 373, + 181 + ], + [ + 396, + 179 + ], + [ + 370, + 194 + ] + ], + [ + [ + 391, + 165 + ], + [ + 399, + 175 + ], + [ + 404, + 170 + ], + [ + 377, + 159 + ], + [ + 396, + 161 + ], + [ + 369, + 149 + ] + ] + ] + }, + { + "task_id": "0000040", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_819112070044#372", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000040/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 340, + 231 + ], + [ + 343, + 217 + ], + [ + 347, + 213 + ], + [ + 332, + 249 + ], + [ + 343, + 226 + ], + [ + 329, + 261 + ] + ], + [ + [ + 356, + 211 + ], + [ + 375, + 215 + ], + [ + 343, + 224 + ], + [ + 367, + 199 + ], + [ + 324, + 221 + ], + [ + 349, + 196 + ] + ], + [ + [ + 322, + 220 + ], + [ + 341, + 226 + ], + [ + 326, + 198 + ], + [ + 318, + 241 + ], + [ + 308, + 193 + ], + [ + 300, + 234 + ] + ], + [ + [ + 345, + 257 + ], + [ + 341, + 265 + ], + [ + 368, + 253 + ], + [ + 322, + 261 + ], + [ + 373, + 244 + ], + [ + 326, + 252 + ] + ] + ] + }, + { + "task_id": "0000043", + "id": "RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_104122061330#179", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000043/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 364, + 61 + ], + [ + 345, + 81 + ], + [ + 380, + 78 + ], + [ + 350, + 46 + ], + [ + 401, + 57 + ], + [ + 369, + 26 + ] + ], + [ + [ + 381, + 106 + ], + [ + 385, + 113 + ], + [ + 399, + 84 + ], + [ + 363, + 127 + ], + [ + 397, + 75 + ], + [ + 359, + 120 + ] + ], + [ + [ + 377, + 81 + ], + [ + 368, + 107 + ], + [ + 384, + 88 + ], + [ + 369, + 75 + ], + [ + 393, + 63 + ], + [ + 378, + 48 + ] + ], + [ + [ + 387, + 96 + ], + [ + 412, + 102 + ], + [ + 384, + 109 + ], + [ + 390, + 82 + ], + [ + 360, + 104 + ], + [ + 364, + 77 + ] + ] + ] + }, + { + "task_id": "0000045", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0007_cfg_0004_104122062823#364", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000045/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 278, + 103 + ], + [ + 288, + 119 + ], + [ + 300, + 84 + ], + [ + 256, + 123 + ], + [ + 289, + 66 + ], + [ + 243, + 106 + ] + ], + [ + [ + 272, + 78 + ], + [ + 257, + 70 + ], + [ + 255, + 54 + ], + [ + 287, + 100 + ], + [ + 271, + 62 + ], + [ + 303, + 110 + ] + ], + [ + [ + 300, + 77 + ], + [ + 323, + 77 + ], + [ + 319, + 83 + ], + [ + 280, + 70 + ], + [ + 298, + 82 + ], + [ + 258, + 70 + ] + ], + [ + [ + 295, + 81 + ], + [ + 287, + 52 + ], + [ + 314, + 65 + ], + [ + 275, + 98 + ], + [ + 321, + 92 + ], + [ + 284, + 124 + ] + ] + ] + }, + { + "task_id": "0000046", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0007_cfg_0004_104122062823#1583", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000046/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 281, + 10 + ], + [ + 309, + 0 + ], + [ + 300, + 16 + ], + [ + 259, + 3 + ], + [ + 275, + 25 + ], + [ + 233, + 13 + ] + ], + [ + [ + 277, + 31 + ], + [ + 251, + 34 + ], + [ + 295, + 61 + ], + [ + 260, + 1 + ], + [ + 318, + 57 + ], + [ + 284, + 0 + ] + ], + [ + [ + 275, + 37 + ], + [ + 258, + 7 + ], + [ + 288, + 21 + ], + [ + 261, + 55 + ], + [ + 302, + 48 + ], + [ + 278, + 83 + ] + ], + [ + [ + 254, + 20 + ], + [ + 283, + 5 + ], + [ + 235, + 0 + ], + [ + 272, + 52 + ], + [ + 209, + 3 + ], + [ + 245, + 65 + ] + ] + ] + }, + { + "task_id": "0000053", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0002_cfg_0005_104122061850#301", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000053/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 92 + ], + [ + 403, + 88 + ], + [ + 386, + 118 + ], + [ + 399, + 65 + ], + [ + 377, + 120 + ], + [ + 389, + 70 + ] + ], + [ + [ + 393, + 70 + ], + [ + 396, + 49 + ], + [ + 368, + 72 + ], + [ + 420, + 68 + ], + [ + 365, + 94 + ], + [ + 417, + 91 + ] + ], + [ + [ + 405, + 96 + ], + [ + 410, + 69 + ], + [ + 398, + 92 + ], + [ + 413, + 100 + ], + [ + 393, + 117 + ], + [ + 408, + 128 + ] + ], + [ + [ + 380, + 77 + ], + [ + 381, + 58 + ], + [ + 357, + 89 + ], + [ + 401, + 66 + ], + [ + 357, + 106 + ], + [ + 399, + 84 + ] + ] + ] + }, + { + "task_id": "0000057", + "id": "RH20T_cfg4_task_0003_user_0010_scene_0010_cfg_0004_038522062288#238", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000057/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 277 + ], + [ + 276, + 283 + ], + [ + 287, + 284 + ], + [ + 333, + 269 + ], + [ + 316, + 279 + ], + [ + 363, + 264 + ] + ], + [ + [ + 296, + 281 + ], + [ + 292, + 254 + ], + [ + 262, + 276 + ], + [ + 332, + 286 + ], + [ + 267, + 300 + ], + [ + 334, + 310 + ] + ], + [ + [ + 289, + 319 + ], + [ + 277, + 354 + ], + [ + 255, + 309 + ], + [ + 325, + 329 + ], + [ + 267, + 278 + ], + [ + 334, + 296 + ] + ], + [ + [ + 316, + 270 + ], + [ + 297, + 300 + ], + [ + 343, + 289 + ], + [ + 291, + 252 + ], + [ + 362, + 257 + ], + [ + 310, + 223 + ] + ] + ] + }, + { + "task_id": "0000059", + "id": "RH20T_cfg5_task_0123_user_0007_scene_0007_cfg_0005_037522062165#186", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the knife on the knife rack'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000059/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 312 + ], + [ + 251, + 333 + ], + [ + 291, + 337 + ], + [ + 256, + 288 + ], + [ + 312, + 318 + ], + [ + 278, + 270 + ] + ], + [ + [ + 271, + 304 + ], + [ + 274, + 290 + ], + [ + 251, + 282 + ], + [ + 292, + 327 + ], + [ + 249, + 294 + ], + [ + 289, + 338 + ] + ], + [ + [ + 265, + 284 + ], + [ + 242, + 284 + ], + [ + 246, + 291 + ], + [ + 286, + 277 + ], + [ + 267, + 290 + ], + [ + 307, + 278 + ] + ], + [ + [ + 263, + 301 + ], + [ + 237, + 321 + ], + [ + 282, + 327 + ], + [ + 245, + 276 + ], + [ + 307, + 308 + ], + [ + 270, + 258 + ] + ] + ] + }, + { + "task_id": "0000061", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0002_cfg_0006_104122064161#255", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000061/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 380, + 101 + ], + [ + 399, + 106 + ], + [ + 376, + 121 + ], + [ + 383, + 80 + ], + [ + 356, + 116 + ], + [ + 363, + 75 + ] + ], + [ + [ + 393, + 112 + ], + [ + 407, + 115 + ], + [ + 410, + 119 + ], + [ + 378, + 106 + ], + [ + 396, + 115 + ], + [ + 364, + 102 + ] + ], + [ + [ + 381, + 121 + ], + [ + 382, + 134 + ], + [ + 372, + 136 + ], + [ + 391, + 105 + ], + [ + 371, + 123 + ], + [ + 390, + 93 + ] + ], + [ + [ + 400, + 113 + ], + [ + 413, + 127 + ], + [ + 387, + 112 + ], + [ + 413, + 113 + ], + [ + 375, + 98 + ], + [ + 401, + 99 + ] + ] + ] + }, + { + "task_id": "0000062", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0002_cfg_0006_104122064161#506", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000062/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 386, + 176 + ], + [ + 374, + 160 + ], + [ + 401, + 179 + ], + [ + 373, + 173 + ], + [ + 413, + 195 + ], + [ + 384, + 188 + ] + ], + [ + [ + 382, + 150 + ], + [ + 382, + 159 + ], + [ + 361, + 143 + ], + [ + 404, + 156 + ], + [ + 362, + 135 + ], + [ + 403, + 148 + ] + ], + [ + [ + 385, + 146 + ], + [ + 364, + 150 + ], + [ + 377, + 154 + ], + [ + 394, + 137 + ], + [ + 396, + 150 + ], + [ + 415, + 134 + ] + ], + [ + [ + 399, + 172 + ], + [ + 399, + 189 + ], + [ + 395, + 187 + ], + [ + 405, + 156 + ], + [ + 395, + 171 + ], + [ + 404, + 140 + ] + ] + ] + }, + { + "task_id": "0000064", + "id": "RH20T_cfg7_task_0050_user_0020_scene_0008_cfg_0007_104122061602#130", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the dish on the dish rack'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000064/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 462, + 194 + ], + [ + 436, + 190 + ], + [ + 467, + 170 + ], + [ + 457, + 217 + ], + [ + 493, + 174 + ], + [ + 482, + 221 + ] + ], + [ + [ + 470, + 183 + ], + [ + 469, + 162 + ], + [ + 495, + 190 + ], + [ + 447, + 177 + ], + [ + 494, + 210 + ], + [ + 448, + 197 + ] + ], + [ + [ + 463, + 167 + ], + [ + 481, + 151 + ], + [ + 444, + 158 + ], + [ + 481, + 175 + ], + [ + 428, + 173 + ], + [ + 464, + 189 + ] + ], + [ + [ + 491, + 191 + ], + [ + 479, + 214 + ], + [ + 466, + 180 + ], + [ + 517, + 203 + ], + [ + 478, + 157 + ], + [ + 529, + 180 + ] + ] + ] + }, + { + "task_id": "0000070", + "id": "RH20T_cfg7_task_0042_user_0020_scene_0003_cfg_0007_104122061330#802", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000070/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 154 + ], + [ + 317, + 142 + ], + [ + 327, + 175 + ], + [ + 350, + 133 + ], + [ + 349, + 188 + ], + [ + 372, + 146 + ] + ], + [ + [ + 367, + 134 + ], + [ + 353, + 148 + ], + [ + 388, + 129 + ], + [ + 348, + 138 + ], + [ + 401, + 115 + ], + [ + 361, + 125 + ] + ], + [ + [ + 356, + 146 + ], + [ + 351, + 131 + ], + [ + 344, + 168 + ], + [ + 367, + 125 + ], + [ + 349, + 181 + ], + [ + 370, + 139 + ] + ], + [ + [ + 373, + 156 + ], + [ + 368, + 182 + ], + [ + 347, + 153 + ], + [ + 399, + 160 + ], + [ + 352, + 128 + ], + [ + 404, + 135 + ] + ] + ] + }, + { + "task_id": "0000071", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_038522063145#135", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000071/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 435, + 126 + ], + [ + 451, + 115 + ], + [ + 424, + 106 + ], + [ + 446, + 145 + ], + [ + 408, + 117 + ], + [ + 430, + 156 + ] + ], + [ + [ + 455, + 134 + ], + [ + 463, + 117 + ], + [ + 474, + 146 + ], + [ + 436, + 121 + ], + [ + 466, + 162 + ], + [ + 430, + 137 + ] + ], + [ + [ + 461, + 140 + ], + [ + 440, + 148 + ], + [ + 470, + 161 + ], + [ + 452, + 118 + ], + [ + 491, + 152 + ], + [ + 472, + 110 + ] + ], + [ + [ + 445, + 121 + ], + [ + 455, + 134 + ], + [ + 429, + 116 + ], + [ + 462, + 126 + ], + [ + 418, + 102 + ], + [ + 451, + 112 + ] + ] + ] + }, + { + "task_id": "0000076", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0006_cfg_0004_104122062295#360", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000076/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 310, + 169 + ], + [ + 286, + 142 + ], + [ + 318, + 181 + ], + [ + 303, + 159 + ], + [ + 343, + 208 + ], + [ + 325, + 182 + ] + ], + [ + [ + 384, + 182 + ], + [ + 410, + 215 + ], + [ + 389, + 185 + ], + [ + 378, + 179 + ], + [ + 367, + 157 + ], + [ + 351, + 145 + ] + ], + [ + [ + 366, + 140 + ], + [ + 369, + 111 + ], + [ + 406, + 146 + ], + [ + 326, + 133 + ], + [ + 405, + 178 + ], + [ + 321, + 164 + ] + ], + [ + [ + 343, + 182 + ], + [ + 378, + 171 + ], + [ + 334, + 208 + ], + [ + 351, + 160 + ], + [ + 299, + 218 + ], + [ + 321, + 171 + ] + ] + ] + }, + { + "task_id": "0000077", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0006_cfg_0004_104122062295#3411", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000077/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 200, + 85 + ], + [ + 200, + 87 + ], + [ + 159, + 96 + ], + [ + 239, + 75 + ], + [ + 155, + 94 + ], + [ + 243, + 72 + ] + ], + [ + [ + 212, + 112 + ], + [ + 229, + 100 + ], + [ + 188, + 78 + ], + [ + 234, + 144 + ], + [ + 174, + 91 + ], + [ + 218, + 152 + ] + ], + [ + [ + 197, + 159 + ], + [ + 220, + 129 + ], + [ + 177, + 147 + ], + [ + 219, + 172 + ], + [ + 154, + 176 + ], + [ + 194, + 205 + ] + ], + [ + [ + 230, + 107 + ], + [ + 219, + 69 + ], + [ + 193, + 108 + ], + [ + 263, + 105 + ], + [ + 205, + 149 + ], + [ + 274, + 141 + ] + ] + ] + }, + { + "task_id": "0000078", + "id": "RH20T_cfg6_task_0029_user_0014_scene_0005_cfg_0006_104122064161#101", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000078/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 161 + ], + [ + 309, + 147 + ], + [ + 307, + 186 + ], + [ + 311, + 137 + ], + [ + 308, + 198 + ], + [ + 311, + 151 + ] + ], + [ + [ + 281, + 161 + ], + [ + 252, + 158 + ], + [ + 277, + 176 + ], + [ + 284, + 148 + ], + [ + 307, + 180 + ], + [ + 311, + 151 + ] + ], + [ + [ + 333, + 136 + ], + [ + 318, + 124 + ], + [ + 331, + 113 + ], + [ + 336, + 158 + ], + [ + 348, + 126 + ], + [ + 352, + 171 + ] + ], + [ + [ + 309, + 154 + ], + [ + 303, + 131 + ], + [ + 283, + 158 + ], + [ + 337, + 150 + ], + [ + 290, + 180 + ], + [ + 342, + 172 + ] + ] + ] + }, + { + "task_id": "0000079", + "id": "RH20T_cfg7_task_0039_user_0014_scene_0008_cfg_0007_104122061330#110", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000079/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 297, + 218 + ], + [ + 272, + 213 + ], + [ + 306, + 190 + ], + [ + 287, + 246 + ], + [ + 332, + 195 + ], + [ + 313, + 252 + ] + ], + [ + [ + 280, + 229 + ], + [ + 275, + 214 + ], + [ + 266, + 254 + ], + [ + 293, + 205 + ], + [ + 271, + 267 + ], + [ + 297, + 220 + ] + ], + [ + [ + 260, + 235 + ], + [ + 272, + 209 + ], + [ + 241, + 232 + ], + [ + 280, + 237 + ], + [ + 229, + 258 + ], + [ + 267, + 266 + ] + ], + [ + [ + 292, + 272 + ], + [ + 323, + 272 + ], + [ + 294, + 257 + ], + [ + 291, + 289 + ], + [ + 265, + 257 + ], + [ + 258, + 289 + ] + ] + ] + }, + { + "task_id": "0000080", + "id": "RH20T_cfg3_task_0038_user_0010_scene_0005_cfg_0003_104122062295#196", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000080/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 387, + 239 + ], + [ + 347, + 248 + ], + [ + 394, + 287 + ], + [ + 380, + 195 + ], + [ + 441, + 279 + ], + [ + 422, + 183 + ] + ], + [ + [ + 388, + 241 + ], + [ + 394, + 202 + ], + [ + 341, + 222 + ], + [ + 439, + 262 + ], + [ + 339, + 255 + ], + [ + 430, + 295 + ] + ], + [ + [ + 419, + 253 + ], + [ + 411, + 276 + ], + [ + 397, + 213 + ], + [ + 444, + 299 + ], + [ + 404, + 185 + ], + [ + 457, + 275 + ] + ], + [ + [ + 323, + 255 + ], + [ + 304, + 305 + ], + [ + 275, + 234 + ], + [ + 371, + 276 + ], + [ + 295, + 190 + ], + [ + 386, + 231 + ] + ] + ] + }, + { + "task_id": "0000081", + "id": "RH20T_cfg7_task_0001_user_0014_scene_0007_cfg_0007_104122063633#101", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000081/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 183, + 145 + ], + [ + 179, + 130 + ], + [ + 187, + 127 + ], + [ + 179, + 163 + ], + [ + 190, + 142 + ], + [ + 182, + 177 + ] + ], + [ + [ + 165, + 126 + ], + [ + 177, + 111 + ], + [ + 145, + 124 + ], + [ + 186, + 128 + ], + [ + 131, + 139 + ], + [ + 173, + 144 + ] + ], + [ + [ + 190, + 170 + ], + [ + 176, + 153 + ], + [ + 182, + 170 + ], + [ + 198, + 171 + ], + [ + 197, + 188 + ], + [ + 212, + 188 + ] + ], + [ + [ + 190, + 150 + ], + [ + 213, + 149 + ], + [ + 185, + 161 + ], + [ + 194, + 140 + ], + [ + 161, + 161 + ], + [ + 172, + 141 + ] + ] + ] + }, + { + "task_id": "0000082", + "id": "RH20T_cfg3_task_0071_user_0010_scene_0008_cfg_0003_036422060909#155", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000082/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 366, + 216 + ], + [ + 355, + 191 + ], + [ + 383, + 204 + ], + [ + 350, + 226 + ], + [ + 396, + 231 + ], + [ + 361, + 251 + ] + ], + [ + [ + 309, + 201 + ], + [ + 320, + 227 + ], + [ + 305, + 206 + ], + [ + 312, + 197 + ], + [ + 294, + 182 + ], + [ + 300, + 169 + ] + ], + [ + [ + 334, + 210 + ], + [ + 344, + 198 + ], + [ + 355, + 200 + ], + [ + 311, + 220 + ], + [ + 345, + 210 + ], + [ + 303, + 230 + ] + ], + [ + [ + 309, + 193 + ], + [ + 306, + 210 + ], + [ + 317, + 216 + ], + [ + 301, + 170 + ], + [ + 319, + 200 + ], + [ + 303, + 155 + ] + ] + ] + }, + { + "task_id": "0000084", + "id": "RH20T_cfg3_task_0071_user_0010_scene_0008_cfg_0003_036422060909#217", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000084/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 349, + 216 + ], + [ + 354, + 199 + ], + [ + 370, + 205 + ], + [ + 327, + 227 + ], + [ + 364, + 220 + ], + [ + 323, + 243 + ] + ], + [ + [ + 349, + 207 + ], + [ + 361, + 184 + ], + [ + 334, + 210 + ], + [ + 365, + 204 + ], + [ + 322, + 232 + ], + [ + 353, + 229 + ] + ], + [ + [ + 330, + 238 + ], + [ + 310, + 219 + ], + [ + 316, + 246 + ], + [ + 345, + 229 + ], + [ + 335, + 264 + ], + [ + 365, + 249 + ] + ], + [ + [ + 369, + 191 + ], + [ + 381, + 204 + ], + [ + 344, + 189 + ], + [ + 393, + 192 + ], + [ + 335, + 176 + ], + [ + 381, + 181 + ] + ] + ] + }, + { + "task_id": "0000085", + "id": "RH20T_cfg3_task_0071_user_0010_scene_0008_cfg_0003_036422060909#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000085/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 331, + 188 + ], + [ + 354, + 203 + ], + [ + 316, + 202 + ], + [ + 346, + 175 + ], + [ + 293, + 186 + ], + [ + 324, + 161 + ] + ], + [ + [ + 350, + 191 + ], + [ + 328, + 179 + ], + [ + 335, + 192 + ], + [ + 368, + 189 + ], + [ + 355, + 203 + ], + [ + 389, + 201 + ] + ], + [ + [ + 330, + 197 + ], + [ + 339, + 184 + ], + [ + 349, + 186 + ], + [ + 309, + 209 + ], + [ + 340, + 198 + ], + [ + 302, + 221 + ] + ], + [ + [ + 323, + 178 + ], + [ + 334, + 200 + ], + [ + 300, + 176 + ], + [ + 346, + 180 + ], + [ + 290, + 156 + ], + [ + 336, + 160 + ] + ] + ] + }, + { + "task_id": "0000087", + "id": "RH20T_cfg2_task_0065_user_0007_scene_0003_cfg_0002_037522062165#203", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000087/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 167, + 330 + ], + [ + 165, + 302 + ], + [ + 138, + 331 + ], + [ + 195, + 328 + ], + [ + 140, + 360 + ], + [ + 198, + 356 + ] + ], + [ + [ + 179, + 306 + ], + [ + 179, + 309 + ], + [ + 149, + 301 + ], + [ + 207, + 310 + ], + [ + 150, + 298 + ], + [ + 205, + 307 + ] + ], + [ + [ + 167, + 303 + ], + [ + 174, + 277 + ], + [ + 162, + 287 + ], + [ + 172, + 320 + ], + [ + 155, + 311 + ], + [ + 164, + 346 + ] + ], + [ + [ + 172, + 330 + ], + [ + 167, + 318 + ], + [ + 188, + 305 + ], + [ + 154, + 358 + ], + [ + 191, + 317 + ], + [ + 160, + 360 + ] + ] + ] + }, + { + "task_id": "0000089", + "id": "RH20T_cfg6_task_0062_user_0014_scene_0003_cfg_0006_104122061330#107", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the toilet paper on its holder'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000089/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 286, + 278 + ], + [ + 317, + 270 + ], + [ + 286, + 299 + ], + [ + 287, + 260 + ], + [ + 253, + 308 + ], + [ + 258, + 267 + ] + ], + [ + [ + 301, + 252 + ], + [ + 289, + 246 + ], + [ + 291, + 280 + ], + [ + 311, + 224 + ], + [ + 302, + 284 + ], + [ + 321, + 230 + ] + ], + [ + [ + 280, + 269 + ], + [ + 305, + 274 + ], + [ + 292, + 246 + ], + [ + 268, + 295 + ], + [ + 269, + 242 + ], + [ + 244, + 289 + ] + ], + [ + [ + 303, + 232 + ], + [ + 291, + 257 + ], + [ + 298, + 216 + ], + [ + 308, + 249 + ], + [ + 309, + 191 + ], + [ + 321, + 221 + ] + ] + ] + }, + { + "task_id": "0000090", + "id": "RH20T_cfg5_task_0101_user_0007_scene_0007_cfg_0005_104122063678#227", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the table with a cloth'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000090/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 339, + 151 + ], + [ + 340, + 142 + ], + [ + 348, + 133 + ], + [ + 330, + 169 + ], + [ + 346, + 142 + ], + [ + 328, + 180 + ] + ], + [ + [ + 323, + 159 + ], + [ + 340, + 151 + ], + [ + 332, + 178 + ], + [ + 315, + 140 + ], + [ + 315, + 185 + ], + [ + 298, + 148 + ] + ], + [ + [ + 328, + 153 + ], + [ + 331, + 138 + ], + [ + 326, + 170 + ], + [ + 329, + 138 + ], + [ + 323, + 184 + ], + [ + 326, + 152 + ] + ], + [ + [ + 343, + 161 + ], + [ + 350, + 142 + ], + [ + 322, + 152 + ], + [ + 364, + 170 + ], + [ + 315, + 169 + ], + [ + 356, + 187 + ] + ] + ] + }, + { + "task_id": "0000091", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0008_cfg_0005_104122063678#440", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000091/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 407, + 181 + ], + [ + 430, + 186 + ], + [ + 411, + 165 + ], + [ + 403, + 196 + ], + [ + 387, + 161 + ], + [ + 381, + 192 + ] + ], + [ + [ + 409, + 198 + ], + [ + 391, + 185 + ], + [ + 420, + 183 + ], + [ + 397, + 213 + ], + [ + 439, + 196 + ], + [ + 416, + 227 + ] + ], + [ + [ + 397, + 195 + ], + [ + 405, + 180 + ], + [ + 399, + 213 + ], + [ + 395, + 179 + ], + [ + 391, + 228 + ], + [ + 388, + 193 + ] + ], + [ + [ + 374, + 199 + ], + [ + 384, + 220 + ], + [ + 369, + 204 + ], + [ + 380, + 194 + ], + [ + 359, + 185 + ], + [ + 369, + 173 + ] + ] + ] + }, + { + "task_id": "0000093", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0008_cfg_0005_104122063678#1017", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000093/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 422, + 244 + ], + [ + 430, + 229 + ], + [ + 414, + 262 + ], + [ + 429, + 227 + ], + [ + 406, + 276 + ], + [ + 422, + 241 + ] + ], + [ + [ + 413, + 220 + ], + [ + 440, + 216 + ], + [ + 411, + 220 + ], + [ + 416, + 220 + ], + [ + 384, + 224 + ], + [ + 391, + 224 + ] + ], + [ + [ + 421, + 260 + ], + [ + 414, + 283 + ], + [ + 439, + 271 + ], + [ + 405, + 249 + ], + [ + 446, + 246 + ], + [ + 412, + 226 + ] + ], + [ + [ + 418, + 254 + ], + [ + 425, + 270 + ], + [ + 417, + 272 + ], + [ + 419, + 234 + ], + [ + 410, + 258 + ], + [ + 412, + 220 + ] + ] + ] + }, + { + "task_id": "0000094", + "id": "RH20T_cfg2_task_0057_user_0016_scene_0001_cfg_0002_105422061350#262", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000094/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 69 + ], + [ + 325, + 68 + ], + [ + 305, + 48 + ], + [ + 324, + 89 + ], + [ + 296, + 49 + ], + [ + 314, + 89 + ] + ], + [ + [ + 275, + 61 + ], + [ + 258, + 64 + ], + [ + 278, + 84 + ], + [ + 273, + 37 + ], + [ + 297, + 82 + ], + [ + 291, + 33 + ] + ], + [ + [ + 278, + 33 + ], + [ + 282, + 33 + ], + [ + 292, + 13 + ], + [ + 265, + 53 + ], + [ + 288, + 12 + ], + [ + 260, + 54 + ] + ], + [ + [ + 292, + 57 + ], + [ + 297, + 39 + ], + [ + 301, + 42 + ], + [ + 283, + 72 + ], + [ + 296, + 58 + ], + [ + 279, + 88 + ] + ] + ] + }, + { + "task_id": "0000095", + "id": "RH20T_cfg2_task_0057_user_0016_scene_0001_cfg_0002_105422061350#330", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000095/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 71 + ], + [ + 306, + 80 + ], + [ + 291, + 52 + ], + [ + 279, + 88 + ], + [ + 268, + 42 + ], + [ + 258, + 79 + ] + ], + [ + [ + 279, + 37 + ], + [ + 263, + 41 + ], + [ + 295, + 50 + ], + [ + 263, + 25 + ], + [ + 310, + 46 + ], + [ + 278, + 22 + ] + ], + [ + [ + 299, + 46 + ], + [ + 303, + 29 + ], + [ + 306, + 30 + ], + [ + 292, + 62 + ], + [ + 302, + 46 + ], + [ + 288, + 78 + ] + ], + [ + [ + 323, + 42 + ], + [ + 315, + 54 + ], + [ + 340, + 30 + ], + [ + 306, + 53 + ], + [ + 347, + 19 + ], + [ + 314, + 43 + ] + ] + ] + }, + { + "task_id": "0000096", + "id": "RH20T_cfg2_task_0057_user_0016_scene_0001_cfg_0002_105422061350#420", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000096/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 281, + 47 + ], + [ + 285, + 30 + ], + [ + 289, + 32 + ], + [ + 273, + 64 + ], + [ + 285, + 47 + ], + [ + 269, + 79 + ] + ], + [ + [ + 273, + 20 + ], + [ + 274, + 30 + ], + [ + 289, + 3 + ], + [ + 257, + 36 + ], + [ + 287, + 0 + ], + [ + 256, + 26 + ] + ], + [ + [ + 291, + 77 + ], + [ + 289, + 100 + ], + [ + 278, + 79 + ], + [ + 305, + 75 + ], + [ + 281, + 57 + ], + [ + 307, + 51 + ] + ], + [ + [ + 284, + 40 + ], + [ + 302, + 25 + ], + [ + 296, + 60 + ], + [ + 273, + 21 + ], + [ + 278, + 75 + ], + [ + 256, + 37 + ] + ] + ] + }, + { + "task_id": "0000099", + "id": "RH20T_cfg2_task_0057_user_0016_scene_0001_cfg_0002_105422061350#748", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000099/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 258, + 46 + ], + [ + 258, + 28 + ], + [ + 264, + 29 + ], + [ + 251, + 63 + ], + [ + 264, + 45 + ], + [ + 251, + 79 + ] + ], + [ + [ + 275, + 71 + ], + [ + 288, + 89 + ], + [ + 260, + 87 + ], + [ + 290, + 54 + ], + [ + 246, + 69 + ], + [ + 277, + 35 + ] + ], + [ + [ + 256, + 42 + ], + [ + 243, + 31 + ], + [ + 266, + 52 + ], + [ + 245, + 32 + ], + [ + 279, + 64 + ], + [ + 259, + 44 + ] + ], + [ + [ + 258, + 53 + ], + [ + 266, + 36 + ], + [ + 243, + 62 + ], + [ + 271, + 44 + ], + [ + 236, + 78 + ], + [ + 263, + 60 + ] + ] + ] + }, + { + "task_id": "0000101", + "id": "RH20T_cfg5_task_0082_user_0007_scene_0008_cfg_0005_036422060215#187", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000101/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 256, + 128 + ], + [ + 260, + 110 + ], + [ + 277, + 125 + ], + [ + 234, + 132 + ], + [ + 273, + 142 + ], + [ + 232, + 150 + ] + ], + [ + [ + 239, + 148 + ], + [ + 231, + 155 + ], + [ + 215, + 148 + ], + [ + 261, + 149 + ], + [ + 223, + 140 + ], + [ + 271, + 142 + ] + ], + [ + [ + 257, + 124 + ], + [ + 278, + 119 + ], + [ + 253, + 138 + ], + [ + 263, + 109 + ], + [ + 232, + 144 + ], + [ + 240, + 114 + ] + ], + [ + [ + 256, + 109 + ], + [ + 279, + 110 + ], + [ + 250, + 109 + ], + [ + 261, + 110 + ], + [ + 226, + 108 + ], + [ + 240, + 108 + ] + ] + ] + }, + { + "task_id": "0000104", + "id": "RH20T_cfg1_task_0076_user_0010_scene_0001_cfg_0001_038522063145#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) one on top of the other every time'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000104/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 507, + 160 + ], + [ + 516, + 169 + ], + [ + 510, + 140 + ], + [ + 504, + 181 + ], + [ + 500, + 130 + ], + [ + 494, + 172 + ] + ], + [ + [ + 516, + 152 + ], + [ + 511, + 175 + ], + [ + 493, + 147 + ], + [ + 539, + 158 + ], + [ + 498, + 125 + ], + [ + 543, + 135 + ] + ], + [ + [ + 478, + 165 + ], + [ + 474, + 186 + ], + [ + 497, + 170 + ], + [ + 459, + 161 + ], + [ + 501, + 149 + ], + [ + 463, + 140 + ] + ], + [ + [ + 503, + 159 + ], + [ + 510, + 143 + ], + [ + 524, + 171 + ], + [ + 483, + 149 + ], + [ + 517, + 186 + ], + [ + 478, + 164 + ] + ] + ] + }, + { + "task_id": "0000109", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0002_cfg_0001_038522062547#748", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000109/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 416, + 70 + ], + [ + 398, + 77 + ], + [ + 432, + 68 + ], + [ + 401, + 72 + ], + [ + 450, + 62 + ], + [ + 418, + 66 + ] + ], + [ + [ + 419, + 49 + ], + [ + 422, + 34 + ], + [ + 425, + 69 + ], + [ + 414, + 30 + ], + [ + 422, + 82 + ], + [ + 411, + 45 + ] + ], + [ + [ + 396, + 40 + ], + [ + 373, + 44 + ], + [ + 399, + 63 + ], + [ + 393, + 16 + ], + [ + 422, + 59 + ], + [ + 416, + 11 + ] + ], + [ + [ + 437, + 68 + ], + [ + 451, + 61 + ], + [ + 457, + 74 + ], + [ + 418, + 62 + ], + [ + 443, + 80 + ], + [ + 404, + 68 + ] + ] + ] + }, + { + "task_id": "0000110", + "id": "RH20T_cfg7_task_0047_user_0020_scene_0004_cfg_0007_104122060811#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000110/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 362, + 110 + ], + [ + 377, + 106 + ], + [ + 362, + 112 + ], + [ + 363, + 109 + ], + [ + 347, + 116 + ], + [ + 348, + 113 + ] + ], + [ + [ + 355, + 119 + ], + [ + 360, + 104 + ], + [ + 341, + 115 + ], + [ + 368, + 124 + ], + [ + 336, + 129 + ], + [ + 362, + 138 + ] + ], + [ + [ + 349, + 112 + ], + [ + 350, + 103 + ], + [ + 351, + 124 + ], + [ + 347, + 99 + ], + [ + 350, + 133 + ], + [ + 345, + 108 + ] + ], + [ + [ + 369, + 120 + ], + [ + 373, + 135 + ], + [ + 356, + 125 + ], + [ + 383, + 116 + ], + [ + 352, + 110 + ], + [ + 379, + 100 + ] + ] + ] + }, + { + "task_id": "0000112", + "id": "RH20T_cfg1_task_0120_user_0014_scene_0002_cfg_0001_819112071037#300", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000112/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 421, + 127 + ], + [ + 428, + 47 + ], + [ + 452, + 153 + ], + [ + 398, + 108 + ], + [ + 443, + 233 + ], + [ + 394, + 170 + ] + ], + [ + [ + 388, + 160 + ], + [ + 383, + 206 + ], + [ + 323, + 121 + ], + [ + 449, + 196 + ], + [ + 335, + 89 + ], + [ + 446, + 157 + ] + ], + [ + [ + 505, + 87 + ], + [ + 422, + 92 + ], + [ + 560, + 1 + ], + [ + 465, + 150 + ], + [ + 640, + 3 + ], + [ + 530, + 143 + ] + ], + [ + [ + 476, + 140 + ], + [ + 427, + 98 + ], + [ + 510, + 75 + ], + [ + 449, + 192 + ], + [ + 580, + 125 + ], + [ + 499, + 241 + ] + ] + ] + }, + { + "task_id": "0000116", + "id": "RH20T_cfg4_task_0067_user_0007_scene_0007_cfg_0004_104122063550#83", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button from top to bottom with obstacles'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000116/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 200, + 254 + ], + [ + 193, + 227 + ], + [ + 220, + 235 + ], + [ + 182, + 272 + ], + [ + 229, + 266 + ], + [ + 189, + 302 + ] + ], + [ + [ + 200, + 260 + ], + [ + 166, + 264 + ], + [ + 199, + 287 + ], + [ + 201, + 231 + ], + [ + 231, + 282 + ], + [ + 234, + 227 + ] + ], + [ + [ + 219, + 248 + ], + [ + 215, + 241 + ], + [ + 222, + 280 + ], + [ + 217, + 217 + ], + [ + 226, + 284 + ], + [ + 221, + 225 + ] + ], + [ + [ + 203, + 229 + ], + [ + 201, + 211 + ], + [ + 228, + 244 + ], + [ + 175, + 213 + ], + [ + 232, + 265 + ], + [ + 175, + 233 + ] + ] + ] + }, + { + "task_id": "0000117", + "id": "RH20T_cfg4_task_0069_user_0007_scene_0006_cfg_0004_104122062295#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Assemble one piece of a puzzle'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000117/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 97, + 269 + ], + [ + 68, + 261 + ], + [ + 135, + 227 + ], + [ + 52, + 318 + ], + [ + 169, + 233 + ], + [ + 84, + 332 + ] + ], + [ + [ + 85, + 234 + ], + [ + 49, + 216 + ], + [ + 40, + 233 + ], + [ + 121, + 235 + ], + [ + 81, + 255 + ], + [ + 157, + 252 + ] + ], + [ + [ + 143, + 271 + ], + [ + 140, + 231 + ], + [ + 91, + 292 + ], + [ + 188, + 253 + ], + [ + 97, + 329 + ], + [ + 189, + 287 + ] + ], + [ + [ + 116, + 342 + ], + [ + 84, + 323 + ], + [ + 139, + 291 + ], + [ + 92, + 360 + ], + [ + 175, + 309 + ], + [ + 127, + 360 + ] + ] + ] + }, + { + "task_id": "0000118", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122063633#94", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000118/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 214 + ], + [ + 244, + 204 + ], + [ + 259, + 231 + ], + [ + 273, + 196 + ], + [ + 280, + 240 + ], + [ + 296, + 206 + ] + ], + [ + [ + 267, + 191 + ], + [ + 279, + 171 + ], + [ + 247, + 180 + ], + [ + 287, + 203 + ], + [ + 234, + 200 + ], + [ + 274, + 224 + ] + ], + [ + [ + 244, + 200 + ], + [ + 243, + 185 + ], + [ + 245, + 181 + ], + [ + 242, + 219 + ], + [ + 246, + 194 + ], + [ + 243, + 232 + ] + ], + [ + [ + 251, + 191 + ], + [ + 227, + 184 + ], + [ + 246, + 189 + ], + [ + 256, + 193 + ], + [ + 267, + 194 + ], + [ + 279, + 199 + ] + ] + ] + }, + { + "task_id": "0000121", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122063633#424", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000121/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 153, + 204 + ], + [ + 140, + 185 + ], + [ + 166, + 206 + ], + [ + 138, + 201 + ], + [ + 179, + 225 + ], + [ + 153, + 222 + ] + ], + [ + [ + 151, + 183 + ], + [ + 155, + 196 + ], + [ + 124, + 178 + ], + [ + 176, + 187 + ], + [ + 122, + 166 + ], + [ + 172, + 175 + ] + ], + [ + [ + 177, + 195 + ], + [ + 158, + 206 + ], + [ + 192, + 200 + ], + [ + 161, + 189 + ], + [ + 212, + 190 + ], + [ + 182, + 178 + ] + ], + [ + [ + 169, + 199 + ], + [ + 165, + 182 + ], + [ + 175, + 181 + ], + [ + 163, + 218 + ], + [ + 179, + 196 + ], + [ + 168, + 233 + ] + ] + ] + }, + { + "task_id": "0000122", + "id": "RH20T_cfg4_task_0001_user_0010_scene_0004_cfg_0004_038522062288#141", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000122/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 135, + 162 + ], + [ + 147, + 137 + ], + [ + 143, + 179 + ], + [ + 125, + 144 + ], + [ + 132, + 205 + ], + [ + 111, + 172 + ] + ], + [ + [ + 150, + 158 + ], + [ + 164, + 135 + ], + [ + 119, + 139 + ], + [ + 180, + 176 + ], + [ + 103, + 164 + ], + [ + 167, + 202 + ] + ], + [ + [ + 148, + 175 + ], + [ + 138, + 155 + ], + [ + 118, + 171 + ], + [ + 179, + 179 + ], + [ + 129, + 189 + ], + [ + 186, + 196 + ] + ], + [ + [ + 106, + 173 + ], + [ + 87, + 150 + ], + [ + 99, + 200 + ], + [ + 112, + 148 + ], + [ + 117, + 219 + ], + [ + 128, + 169 + ] + ] + ] + }, + { + "task_id": "0000124", + "id": "RH20T_cfg4_task_0071_user_0007_scene_0010_cfg_0004_104122062295#195", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000124/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 353, + 261 + ], + [ + 315, + 258 + ], + [ + 364, + 284 + ], + [ + 343, + 241 + ], + [ + 403, + 286 + ], + [ + 376, + 244 + ] + ], + [ + [ + 328, + 210 + ], + [ + 343, + 179 + ], + [ + 293, + 211 + ], + [ + 360, + 209 + ], + [ + 281, + 240 + ], + [ + 346, + 236 + ] + ], + [ + [ + 306, + 212 + ], + [ + 319, + 176 + ], + [ + 270, + 200 + ], + [ + 343, + 224 + ], + [ + 258, + 235 + ], + [ + 330, + 259 + ] + ], + [ + [ + 317, + 206 + ], + [ + 319, + 200 + ], + [ + 283, + 226 + ], + [ + 351, + 186 + ], + [ + 278, + 235 + ], + [ + 352, + 191 + ] + ] + ] + }, + { + "task_id": "0000125", + "id": "RH20T_cfg1_task_0009_user_0014_scene_0007_cfg_0001_750612070851#321", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000125/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 474, + 243 + ], + [ + 446, + 226 + ], + [ + 470, + 259 + ], + [ + 478, + 225 + ], + [ + 499, + 276 + ], + [ + 510, + 243 + ] + ], + [ + [ + 483, + 251 + ], + [ + 448, + 243 + ], + [ + 489, + 270 + ], + [ + 477, + 234 + ], + [ + 527, + 278 + ], + [ + 510, + 241 + ] + ], + [ + [ + 470, + 239 + ], + [ + 480, + 217 + ], + [ + 449, + 257 + ], + [ + 489, + 222 + ], + [ + 441, + 276 + ], + [ + 480, + 242 + ] + ], + [ + [ + 513, + 239 + ], + [ + 476, + 236 + ], + [ + 514, + 209 + ], + [ + 511, + 267 + ], + [ + 555, + 211 + ], + [ + 549, + 270 + ] + ] + ] + }, + { + "task_id": "0000127", + "id": "RH20T_cfg1_task_0014_user_0014_scene_0001_cfg_0001_750612070853#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the object from the scale'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000127/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 435, + 277 + ], + [ + 426, + 222 + ], + [ + 494, + 273 + ], + [ + 381, + 280 + ], + [ + 503, + 331 + ], + [ + 390, + 333 + ] + ], + [ + [ + 467, + 245 + ], + [ + 415, + 253 + ], + [ + 457, + 195 + ], + [ + 477, + 298 + ], + [ + 509, + 187 + ], + [ + 532, + 291 + ] + ], + [ + [ + 466, + 168 + ], + [ + 417, + 192 + ], + [ + 453, + 192 + ], + [ + 482, + 139 + ], + [ + 495, + 171 + ], + [ + 534, + 115 + ] + ], + [ + [ + 446, + 238 + ], + [ + 445, + 194 + ], + [ + 402, + 216 + ], + [ + 499, + 262 + ], + [ + 406, + 253 + ], + [ + 496, + 302 + ] + ] + ] + }, + { + "task_id": "0000129", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0005_cfg_0003_036422060909#254", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000129/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 360, + 90 + ], + [ + 363, + 114 + ], + [ + 338, + 93 + ], + [ + 381, + 88 + ], + [ + 335, + 68 + ], + [ + 378, + 64 + ] + ], + [ + [ + 358, + 80 + ], + [ + 363, + 62 + ], + [ + 371, + 68 + ], + [ + 345, + 93 + ], + [ + 366, + 84 + ], + [ + 341, + 109 + ] + ], + [ + [ + 356, + 78 + ], + [ + 341, + 70 + ], + [ + 357, + 54 + ], + [ + 356, + 100 + ], + [ + 374, + 61 + ], + [ + 372, + 109 + ] + ], + [ + [ + 361, + 87 + ], + [ + 357, + 101 + ], + [ + 339, + 78 + ], + [ + 384, + 97 + ], + [ + 342, + 62 + ], + [ + 389, + 82 + ] + ] + ] + }, + { + "task_id": "0000130", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0005_cfg_0003_036422060909#296", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000130/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 302, + 129 + ], + [ + 286, + 116 + ], + [ + 283, + 144 + ], + [ + 321, + 116 + ], + [ + 300, + 159 + ], + [ + 338, + 129 + ] + ], + [ + [ + 287, + 134 + ], + [ + 269, + 127 + ], + [ + 304, + 135 + ], + [ + 269, + 134 + ], + [ + 323, + 141 + ], + [ + 289, + 141 + ] + ], + [ + [ + 308, + 123 + ], + [ + 308, + 105 + ], + [ + 328, + 111 + ], + [ + 288, + 136 + ], + [ + 327, + 127 + ], + [ + 289, + 152 + ] + ], + [ + [ + 300, + 145 + ], + [ + 293, + 161 + ], + [ + 277, + 131 + ], + [ + 323, + 158 + ], + [ + 285, + 113 + ], + [ + 333, + 142 + ] + ] + ] + }, + { + "task_id": "0000131", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0005_cfg_0003_036422060909#350", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000131/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 223, + 166 + ], + [ + 246, + 150 + ], + [ + 233, + 185 + ], + [ + 212, + 146 + ], + [ + 211, + 201 + ], + [ + 187, + 163 + ] + ], + [ + [ + 258, + 149 + ], + [ + 264, + 173 + ], + [ + 229, + 158 + ], + [ + 286, + 141 + ], + [ + 221, + 132 + ], + [ + 280, + 115 + ] + ], + [ + [ + 250, + 147 + ], + [ + 274, + 157 + ], + [ + 250, + 163 + ], + [ + 249, + 132 + ], + [ + 224, + 153 + ], + [ + 225, + 122 + ] + ], + [ + [ + 260, + 163 + ], + [ + 260, + 146 + ], + [ + 279, + 148 + ], + [ + 239, + 179 + ], + [ + 278, + 164 + ], + [ + 240, + 195 + ] + ] + ] + }, + { + "task_id": "0000133", + "id": "RH20T_cfg1_task_0001_user_0014_scene_0006_cfg_0001_819112070044#222", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000133/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 231 + ], + [ + 332, + 237 + ], + [ + 331, + 248 + ], + [ + 307, + 216 + ], + [ + 316, + 242 + ], + [ + 293, + 209 + ] + ], + [ + [ + 310, + 227 + ], + [ + 312, + 212 + ], + [ + 329, + 218 + ], + [ + 290, + 238 + ], + [ + 326, + 233 + ], + [ + 289, + 252 + ] + ], + [ + [ + 305, + 235 + ], + [ + 319, + 223 + ], + [ + 311, + 221 + ], + [ + 300, + 248 + ], + [ + 296, + 234 + ], + [ + 285, + 261 + ] + ], + [ + [ + 310, + 206 + ], + [ + 315, + 185 + ], + [ + 299, + 211 + ], + [ + 321, + 202 + ], + [ + 294, + 233 + ], + [ + 316, + 222 + ] + ] + ] + }, + { + "task_id": "0000139", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_819112071037#198", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000139/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 57, + 265 + ], + [ + 53, + 157 + ], + [ + 0, + 273 + ], + [ + 162, + 256 + ], + [ + 0, + 360 + ], + [ + 168, + 360 + ] + ], + [ + [ + 117, + 222 + ], + [ + 134, + 169 + ], + [ + 204, + 178 + ], + [ + 34, + 263 + ], + [ + 199, + 246 + ], + [ + 0, + 345 + ] + ], + [ + [ + 91, + 240 + ], + [ + 166, + 245 + ], + [ + 166, + 168 + ], + [ + 22, + 306 + ], + [ + 66, + 140 + ], + [ + 0, + 317 + ] + ], + [ + [ + 54, + 225 + ], + [ + 40, + 136 + ], + [ + 0, + 222 + ], + [ + 149, + 228 + ], + [ + 0, + 301 + ], + [ + 154, + 304 + ] + ] + ] + }, + { + "task_id": "0000141", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_819112071037#445", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000141/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 293, + 196 + ], + [ + 291, + 193 + ], + [ + 367, + 214 + ], + [ + 218, + 176 + ], + [ + 384, + 221 + ], + [ + 205, + 175 + ] + ], + [ + [ + 218, + 292 + ], + [ + 258, + 307 + ], + [ + 264, + 222 + ], + [ + 169, + 360 + ], + [ + 220, + 191 + ], + [ + 104, + 360 + ] + ], + [ + [ + 284, + 274 + ], + [ + 242, + 252 + ], + [ + 207, + 260 + ], + [ + 345, + 285 + ], + [ + 258, + 291 + ], + [ + 404, + 313 + ] + ], + [ + [ + 265, + 250 + ], + [ + 268, + 169 + ], + [ + 179, + 244 + ], + [ + 350, + 257 + ], + [ + 182, + 317 + ], + [ + 343, + 328 + ] + ] + ] + }, + { + "task_id": "0000142", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_819112071037#591", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000142/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 353, + 262 + ], + [ + 361, + 178 + ], + [ + 266, + 257 + ], + [ + 439, + 266 + ], + [ + 263, + 333 + ], + [ + 428, + 340 + ] + ], + [ + [ + 263, + 273 + ], + [ + 324, + 340 + ], + [ + 323, + 299 + ], + [ + 180, + 238 + ], + [ + 276, + 253 + ], + [ + 142, + 188 + ] + ], + [ + [ + 308, + 233 + ], + [ + 356, + 239 + ], + [ + 395, + 163 + ], + [ + 233, + 294 + ], + [ + 340, + 134 + ], + [ + 148, + 301 + ] + ], + [ + [ + 306, + 211 + ], + [ + 271, + 290 + ], + [ + 378, + 235 + ], + [ + 247, + 191 + ], + [ + 412, + 153 + ], + [ + 279, + 125 + ] + ] + ] + }, + { + "task_id": "0000143", + "id": "RH20T_cfg1_task_0120_user_0014_scene_0002_cfg_0001_750612070851#117", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000143/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 517, + 224 + ], + [ + 492, + 218 + ], + [ + 522, + 183 + ], + [ + 513, + 264 + ], + [ + 552, + 186 + ], + [ + 542, + 274 + ] + ], + [ + [ + 424, + 220 + ], + [ + 443, + 188 + ], + [ + 422, + 234 + ], + [ + 426, + 204 + ], + [ + 403, + 266 + ], + [ + 404, + 241 + ] + ], + [ + [ + 460, + 187 + ], + [ + 434, + 209 + ], + [ + 485, + 180 + ], + [ + 440, + 193 + ], + [ + 511, + 158 + ], + [ + 463, + 173 + ] + ], + [ + [ + 455, + 228 + ], + [ + 473, + 198 + ], + [ + 432, + 240 + ], + [ + 476, + 218 + ], + [ + 416, + 268 + ], + [ + 459, + 244 + ] + ] + ] + }, + { + "task_id": "0000147", + "id": "RH20T_cfg1_task_0210_user_0015_scene_0008_cfg_0001_039422060546#325", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Chop the onions'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000147/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 102 + ], + [ + 336, + 106 + ], + [ + 301, + 78 + ], + [ + 309, + 123 + ], + [ + 270, + 73 + ], + [ + 281, + 118 + ] + ], + [ + [ + 276, + 110 + ], + [ + 273, + 85 + ], + [ + 264, + 135 + ], + [ + 286, + 87 + ], + [ + 268, + 156 + ], + [ + 288, + 110 + ] + ], + [ + [ + 253, + 113 + ], + [ + 277, + 139 + ], + [ + 226, + 124 + ], + [ + 277, + 103 + ], + [ + 202, + 98 + ], + [ + 255, + 79 + ] + ], + [ + [ + 260, + 90 + ], + [ + 251, + 60 + ], + [ + 228, + 101 + ], + [ + 292, + 79 + ], + [ + 237, + 131 + ], + [ + 302, + 109 + ] + ] + ] + }, + { + "task_id": "0000151", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122060811#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000151/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 177 + ], + [ + 317, + 162 + ], + [ + 334, + 175 + ], + [ + 304, + 178 + ], + [ + 336, + 190 + ], + [ + 305, + 193 + ] + ], + [ + [ + 334, + 191 + ], + [ + 323, + 199 + ], + [ + 324, + 186 + ], + [ + 345, + 196 + ], + [ + 335, + 178 + ], + [ + 356, + 188 + ] + ], + [ + [ + 335, + 184 + ], + [ + 347, + 177 + ], + [ + 345, + 185 + ], + [ + 326, + 184 + ], + [ + 334, + 192 + ], + [ + 315, + 190 + ] + ], + [ + [ + 357, + 192 + ], + [ + 353, + 201 + ], + [ + 343, + 195 + ], + [ + 369, + 188 + ], + [ + 347, + 185 + ], + [ + 373, + 177 + ] + ] + ] + }, + { + "task_id": "0000152", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122060811#305", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000152/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 345, + 139 + ], + [ + 352, + 140 + ], + [ + 338, + 152 + ], + [ + 351, + 126 + ], + [ + 331, + 151 + ], + [ + 344, + 124 + ] + ], + [ + [ + 345, + 144 + ], + [ + 355, + 135 + ], + [ + 334, + 133 + ], + [ + 355, + 155 + ], + [ + 324, + 143 + ], + [ + 345, + 165 + ] + ], + [ + [ + 344, + 150 + ], + [ + 357, + 150 + ], + [ + 339, + 155 + ], + [ + 350, + 145 + ], + [ + 325, + 155 + ], + [ + 336, + 145 + ] + ], + [ + [ + 328, + 140 + ], + [ + 332, + 154 + ], + [ + 316, + 146 + ], + [ + 342, + 135 + ], + [ + 312, + 133 + ], + [ + 338, + 122 + ] + ] + ] + }, + { + "task_id": "0000153", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122060811#479", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000153/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 349, + 174 + ], + [ + 336, + 169 + ], + [ + 347, + 165 + ], + [ + 351, + 184 + ], + [ + 360, + 170 + ], + [ + 363, + 189 + ] + ], + [ + [ + 344, + 171 + ], + [ + 347, + 184 + ], + [ + 339, + 165 + ], + [ + 349, + 177 + ], + [ + 336, + 152 + ], + [ + 346, + 164 + ] + ], + [ + [ + 353, + 161 + ], + [ + 365, + 154 + ], + [ + 361, + 164 + ], + [ + 345, + 158 + ], + [ + 350, + 172 + ], + [ + 334, + 166 + ] + ], + [ + [ + 343, + 158 + ], + [ + 334, + 152 + ], + [ + 331, + 162 + ], + [ + 354, + 155 + ], + [ + 340, + 168 + ], + [ + 364, + 161 + ] + ] + ] + }, + { + "task_id": "0000154", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122060811#519", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000154/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 358, + 161 + ], + [ + 362, + 173 + ], + [ + 345, + 160 + ], + [ + 371, + 162 + ], + [ + 342, + 148 + ], + [ + 368, + 150 + ] + ], + [ + [ + 355, + 179 + ], + [ + 353, + 166 + ], + [ + 350, + 175 + ], + [ + 361, + 184 + ], + [ + 352, + 189 + ], + [ + 363, + 197 + ] + ], + [ + [ + 361, + 173 + ], + [ + 369, + 167 + ], + [ + 372, + 170 + ], + [ + 350, + 177 + ], + [ + 365, + 176 + ], + [ + 343, + 183 + ] + ], + [ + [ + 357, + 174 + ], + [ + 366, + 180 + ], + [ + 352, + 164 + ], + [ + 362, + 184 + ], + [ + 343, + 157 + ], + [ + 353, + 178 + ] + ] + ] + }, + { + "task_id": "0000155", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122060811#580", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000155/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 376, + 181 + ], + [ + 382, + 194 + ], + [ + 362, + 187 + ], + [ + 390, + 176 + ], + [ + 356, + 174 + ], + [ + 384, + 164 + ] + ], + [ + [ + 366, + 159 + ], + [ + 364, + 174 + ], + [ + 365, + 157 + ], + [ + 366, + 161 + ], + [ + 367, + 143 + ], + [ + 369, + 145 + ] + ], + [ + [ + 385, + 168 + ], + [ + 379, + 179 + ], + [ + 390, + 178 + ], + [ + 379, + 157 + ], + [ + 395, + 166 + ], + [ + 384, + 146 + ] + ], + [ + [ + 370, + 172 + ], + [ + 378, + 166 + ], + [ + 382, + 171 + ], + [ + 357, + 173 + ], + [ + 374, + 178 + ], + [ + 350, + 179 + ] + ] + ] + }, + { + "task_id": "0000156", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122060811#810", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000156/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 145 + ], + [ + 315, + 136 + ], + [ + 327, + 135 + ], + [ + 311, + 155 + ], + [ + 331, + 145 + ], + [ + 315, + 165 + ] + ], + [ + [ + 308, + 146 + ], + [ + 319, + 141 + ], + [ + 299, + 138 + ], + [ + 316, + 154 + ], + [ + 288, + 143 + ], + [ + 305, + 159 + ] + ], + [ + [ + 317, + 150 + ], + [ + 324, + 143 + ], + [ + 328, + 149 + ], + [ + 305, + 152 + ], + [ + 322, + 156 + ], + [ + 298, + 159 + ] + ], + [ + [ + 319, + 151 + ], + [ + 316, + 138 + ], + [ + 309, + 157 + ], + [ + 330, + 145 + ], + [ + 312, + 170 + ], + [ + 333, + 159 + ] + ] + ] + }, + { + "task_id": "0000157", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0007_cfg_0006_104122061018#155", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000157/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 62 + ], + [ + 321, + 62 + ], + [ + 281, + 66 + ], + [ + 337, + 57 + ], + [ + 271, + 65 + ], + [ + 325, + 57 + ] + ], + [ + [ + 332, + 72 + ], + [ + 323, + 42 + ], + [ + 321, + 78 + ], + [ + 344, + 64 + ], + [ + 329, + 106 + ], + [ + 354, + 95 + ] + ], + [ + [ + 311, + 92 + ], + [ + 315, + 64 + ], + [ + 339, + 96 + ], + [ + 283, + 87 + ], + [ + 335, + 123 + ], + [ + 280, + 114 + ] + ], + [ + [ + 346, + 73 + ], + [ + 322, + 85 + ], + [ + 326, + 71 + ], + [ + 368, + 76 + ], + [ + 348, + 60 + ], + [ + 391, + 64 + ] + ] + ] + }, + { + "task_id": "0000158", + "id": "RH20T_cfg1_task_0038_user_0016_scene_0009_cfg_0001_038522062547#129", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000158/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 434, + 128 + ], + [ + 450, + 144 + ], + [ + 410, + 135 + ], + [ + 461, + 119 + ], + [ + 393, + 118 + ], + [ + 444, + 100 + ] + ], + [ + [ + 427, + 159 + ], + [ + 428, + 145 + ], + [ + 435, + 184 + ], + [ + 418, + 134 + ], + [ + 434, + 195 + ], + [ + 418, + 147 + ] + ], + [ + [ + 428, + 136 + ], + [ + 428, + 115 + ], + [ + 448, + 122 + ], + [ + 408, + 151 + ], + [ + 447, + 141 + ], + [ + 409, + 170 + ] + ], + [ + [ + 402, + 155 + ], + [ + 427, + 166 + ], + [ + 399, + 139 + ], + [ + 407, + 173 + ], + [ + 374, + 128 + ], + [ + 379, + 162 + ] + ] + ] + }, + { + "task_id": "0000162", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0004_cfg_0006_037522061512#514", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000162/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 239, + 116 + ], + [ + 224, + 129 + ], + [ + 220, + 106 + ], + [ + 256, + 124 + ], + [ + 236, + 91 + ], + [ + 272, + 110 + ] + ], + [ + [ + 214, + 146 + ], + [ + 194, + 153 + ], + [ + 207, + 124 + ], + [ + 221, + 168 + ], + [ + 228, + 116 + ], + [ + 242, + 161 + ] + ], + [ + [ + 226, + 125 + ], + [ + 240, + 109 + ], + [ + 234, + 141 + ], + [ + 217, + 106 + ], + [ + 221, + 158 + ], + [ + 202, + 123 + ] + ], + [ + [ + 198, + 138 + ], + [ + 219, + 146 + ], + [ + 191, + 129 + ], + [ + 205, + 146 + ], + [ + 169, + 121 + ], + [ + 184, + 138 + ] + ] + ] + }, + { + "task_id": "0000163", + "id": "RH20T_cfg1_task_0073_user_0001_scene_0003_cfg_0001_038522062547#181", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000163/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 381, + 187 + ], + [ + 384, + 172 + ], + [ + 385, + 223 + ], + [ + 377, + 154 + ], + [ + 383, + 235 + ], + [ + 376, + 169 + ] + ], + [ + [ + 371, + 193 + ], + [ + 346, + 213 + ], + [ + 382, + 177 + ], + [ + 362, + 206 + ], + [ + 407, + 157 + ], + [ + 385, + 188 + ] + ], + [ + [ + 347, + 207 + ], + [ + 356, + 234 + ], + [ + 372, + 186 + ], + [ + 320, + 229 + ], + [ + 364, + 157 + ], + [ + 309, + 199 + ] + ], + [ + [ + 363, + 184 + ], + [ + 328, + 181 + ], + [ + 374, + 166 + ], + [ + 354, + 199 + ], + [ + 411, + 169 + ], + [ + 387, + 202 + ] + ] + ] + }, + { + "task_id": "0000166", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_104122062295#169", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000166/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 0, + 109 + ], + [ + 0, + 65 + ], + [ + 33, + 122 + ], + [ + 0, + 93 + ], + [ + 25, + 165 + ], + [ + 0, + 143 + ] + ], + [ + [ + 0, + 177 + ], + [ + 0, + 134 + ], + [ + 0, + 162 + ], + [ + 46, + 192 + ], + [ + 0, + 203 + ], + [ + 41, + 231 + ] + ], + [ + [ + 0, + 195 + ], + [ + 0, + 236 + ], + [ + 0, + 199 + ], + [ + 0, + 190 + ], + [ + 0, + 157 + ], + [ + 0, + 141 + ] + ], + [ + [ + 9, + 169 + ], + [ + 14, + 159 + ], + [ + 0, + 190 + ], + [ + 54, + 148 + ], + [ + 0, + 204 + ], + [ + 53, + 157 + ] + ] + ] + }, + { + "task_id": "0000167", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_104122062295#317", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000167/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 119, + 220 + ], + [ + 131, + 177 + ], + [ + 89, + 193 + ], + [ + 151, + 250 + ], + [ + 80, + 232 + ], + [ + 139, + 290 + ] + ], + [ + [ + 123, + 287 + ], + [ + 73, + 311 + ], + [ + 159, + 281 + ], + [ + 91, + 292 + ], + [ + 198, + 260 + ], + [ + 132, + 273 + ] + ], + [ + [ + 133, + 194 + ], + [ + 128, + 246 + ], + [ + 149, + 198 + ], + [ + 114, + 190 + ], + [ + 153, + 152 + ], + [ + 120, + 134 + ] + ], + [ + [ + 93, + 254 + ], + [ + 39, + 231 + ], + [ + 63, + 308 + ], + [ + 121, + 205 + ], + [ + 113, + 325 + ], + [ + 163, + 228 + ] + ] + ] + }, + { + "task_id": "0000171", + "id": "RH20T_cfg1_task_0012_user_0014_scene_0004_cfg_0001_750612070851#103", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push the soccer ball into the goal'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000171/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 372, + 303 + ], + [ + 372, + 293 + ], + [ + 333, + 302 + ], + [ + 411, + 304 + ], + [ + 336, + 310 + ], + [ + 408, + 312 + ] + ], + [ + [ + 336, + 338 + ], + [ + 311, + 315 + ], + [ + 353, + 348 + ], + [ + 317, + 327 + ], + [ + 380, + 360 + ], + [ + 346, + 354 + ] + ], + [ + [ + 358, + 320 + ], + [ + 379, + 348 + ], + [ + 324, + 341 + ], + [ + 392, + 300 + ], + [ + 308, + 314 + ], + [ + 372, + 277 + ] + ], + [ + [ + 347, + 292 + ], + [ + 313, + 312 + ], + [ + 356, + 275 + ], + [ + 340, + 307 + ], + [ + 389, + 255 + ], + [ + 370, + 289 + ] + ] + ] + }, + { + "task_id": "0000173", + "id": "RH20T_cfg1_task_0048_user_0001_scene_0003_cfg_0001_750612070851#172", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000173/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 474, + 50 + ], + [ + 488, + 8 + ], + [ + 443, + 62 + ], + [ + 503, + 39 + ], + [ + 433, + 99 + ], + [ + 490, + 76 + ] + ], + [ + [ + 467, + 38 + ], + [ + 439, + 31 + ], + [ + 468, + 0 + ], + [ + 466, + 76 + ], + [ + 501, + 3 + ], + [ + 496, + 87 + ] + ], + [ + [ + 479, + 80 + ], + [ + 505, + 54 + ], + [ + 451, + 44 + ], + [ + 506, + 115 + ], + [ + 422, + 70 + ], + [ + 479, + 143 + ] + ], + [ + [ + 471, + 92 + ], + [ + 434, + 101 + ], + [ + 453, + 59 + ], + [ + 490, + 127 + ], + [ + 488, + 51 + ], + [ + 526, + 118 + ] + ] + ] + }, + { + "task_id": "0000175", + "id": "RH20T_cfg7_task_0006_user_0020_scene_0006_cfg_0007_104122061330#184", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000175/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 296, + 282 + ], + [ + 281, + 308 + ], + [ + 299, + 291 + ], + [ + 292, + 272 + ], + [ + 312, + 267 + ], + [ + 307, + 246 + ] + ], + [ + [ + 306, + 304 + ], + [ + 280, + 296 + ], + [ + 307, + 333 + ], + [ + 305, + 279 + ], + [ + 332, + 340 + ], + [ + 328, + 286 + ] + ], + [ + [ + 283, + 279 + ], + [ + 279, + 304 + ], + [ + 291, + 264 + ], + [ + 274, + 296 + ], + [ + 296, + 239 + ], + [ + 278, + 269 + ] + ], + [ + [ + 316, + 285 + ], + [ + 290, + 271 + ], + [ + 327, + 259 + ], + [ + 304, + 312 + ], + [ + 352, + 271 + ], + [ + 330, + 325 + ] + ] + ] + }, + { + "task_id": "0000177", + "id": "RH20T_cfg1_task_0080_user_0015_scene_0005_cfg_0001_039422060546#126", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Flip over and spread out the paper that is laid flat on the table'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000177/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 135 + ], + [ + 291, + 130 + ], + [ + 282, + 130 + ], + [ + 327, + 139 + ], + [ + 297, + 134 + ], + [ + 344, + 144 + ] + ], + [ + [ + 361, + 101 + ], + [ + 346, + 79 + ], + [ + 364, + 84 + ], + [ + 357, + 120 + ], + [ + 377, + 104 + ], + [ + 372, + 140 + ] + ], + [ + [ + 338, + 111 + ], + [ + 312, + 101 + ], + [ + 330, + 125 + ], + [ + 345, + 98 + ], + [ + 357, + 136 + ], + [ + 370, + 108 + ] + ], + [ + [ + 334, + 124 + ], + [ + 337, + 101 + ], + [ + 329, + 143 + ], + [ + 340, + 107 + ], + [ + 327, + 164 + ], + [ + 337, + 128 + ] + ] + ] + }, + { + "task_id": "0000180", + "id": "RH20T_cfg1_task_0013_user_0001_scene_0005_cfg_0001_750612070853#125", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000180/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 380, + 360 + ], + [ + 378, + 360 + ], + [ + 323, + 360 + ], + [ + 446, + 360 + ], + [ + 329, + 360 + ], + [ + 443, + 360 + ] + ], + [ + [ + 355, + 360 + ], + [ + 314, + 360 + ], + [ + 337, + 338 + ], + [ + 372, + 360 + ], + [ + 373, + 310 + ], + [ + 405, + 360 + ] + ], + [ + [ + 347, + 326 + ], + [ + 342, + 265 + ], + [ + 319, + 335 + ], + [ + 382, + 316 + ], + [ + 322, + 360 + ], + [ + 391, + 360 + ] + ], + [ + [ + 330, + 360 + ], + [ + 279, + 360 + ], + [ + 295, + 360 + ], + [ + 356, + 360 + ], + [ + 359, + 360 + ], + [ + 404, + 360 + ] + ] + ] + }, + { + "task_id": "0000187", + "id": "RH20T_cfg4_task_0034_user_0016_scene_0009_cfg_0004_104122062823#388", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000187/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 289, + 106 + ], + [ + 292, + 75 + ], + [ + 280, + 117 + ], + [ + 297, + 97 + ], + [ + 277, + 148 + ], + [ + 294, + 124 + ] + ], + [ + [ + 267, + 123 + ], + [ + 284, + 148 + ], + [ + 293, + 101 + ], + [ + 240, + 145 + ], + [ + 275, + 72 + ], + [ + 219, + 118 + ] + ], + [ + [ + 249, + 98 + ], + [ + 266, + 74 + ], + [ + 280, + 103 + ], + [ + 220, + 94 + ], + [ + 262, + 129 + ], + [ + 201, + 118 + ] + ], + [ + [ + 283, + 107 + ], + [ + 284, + 78 + ], + [ + 315, + 116 + ], + [ + 252, + 98 + ], + [ + 312, + 142 + ], + [ + 252, + 124 + ] + ] + ] + }, + { + "task_id": "0000188", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0005_cfg_0001_750612070851#168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000188/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 281, + 245 + ], + [ + 307, + 248 + ], + [ + 306, + 225 + ], + [ + 253, + 267 + ], + [ + 282, + 223 + ], + [ + 229, + 262 + ] + ], + [ + [ + 222, + 233 + ], + [ + 190, + 256 + ], + [ + 196, + 205 + ], + [ + 246, + 258 + ], + [ + 230, + 181 + ], + [ + 277, + 235 + ] + ], + [ + [ + 215, + 203 + ], + [ + 220, + 181 + ], + [ + 216, + 233 + ], + [ + 213, + 170 + ], + [ + 210, + 259 + ], + [ + 205, + 194 + ] + ], + [ + [ + 244, + 239 + ], + [ + 240, + 215 + ], + [ + 205, + 245 + ], + [ + 283, + 232 + ], + [ + 212, + 266 + ], + [ + 284, + 252 + ] + ] + ] + }, + { + "task_id": "0000189", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0005_cfg_0001_750612070851#499", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000189/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 270, + 289 + ], + [ + 269, + 268 + ], + [ + 231, + 289 + ], + [ + 310, + 290 + ], + [ + 235, + 308 + ], + [ + 309, + 308 + ] + ], + [ + [ + 290, + 255 + ], + [ + 255, + 263 + ], + [ + 285, + 290 + ], + [ + 295, + 219 + ], + [ + 316, + 281 + ], + [ + 326, + 213 + ] + ], + [ + [ + 237, + 287 + ], + [ + 241, + 252 + ], + [ + 246, + 264 + ], + [ + 229, + 307 + ], + [ + 241, + 305 + ], + [ + 223, + 345 + ] + ], + [ + [ + 303, + 242 + ], + [ + 307, + 271 + ], + [ + 340, + 229 + ], + [ + 266, + 255 + ], + [ + 334, + 204 + ], + [ + 265, + 230 + ] + ] + ] + }, + { + "task_id": "0000190", + "id": "RH20T_cfg6_task_0082_user_0020_scene_0001_cfg_0006_037522061512#263", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000190/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 232, + 68 + ], + [ + 253, + 82 + ], + [ + 229, + 67 + ], + [ + 236, + 70 + ], + [ + 208, + 54 + ], + [ + 212, + 55 + ] + ], + [ + [ + 258, + 111 + ], + [ + 273, + 96 + ], + [ + 255, + 117 + ], + [ + 261, + 105 + ], + [ + 241, + 133 + ], + [ + 245, + 122 + ] + ], + [ + [ + 258, + 113 + ], + [ + 262, + 127 + ], + [ + 280, + 107 + ], + [ + 235, + 119 + ], + [ + 277, + 92 + ], + [ + 230, + 105 + ] + ], + [ + [ + 240, + 90 + ], + [ + 254, + 87 + ], + [ + 236, + 112 + ], + [ + 244, + 66 + ], + [ + 221, + 116 + ], + [ + 229, + 69 + ] + ] + ] + }, + { + "task_id": "0000191", + "id": "RH20T_cfg5_task_0081_user_0007_scene_0010_cfg_0005_104122063678#233", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000191/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 368, + 184 + ], + [ + 391, + 185 + ], + [ + 366, + 207 + ], + [ + 369, + 160 + ], + [ + 343, + 206 + ], + [ + 345, + 159 + ] + ], + [ + [ + 372, + 194 + ], + [ + 366, + 170 + ], + [ + 396, + 188 + ], + [ + 349, + 200 + ], + [ + 402, + 212 + ], + [ + 355, + 224 + ] + ], + [ + [ + 367, + 195 + ], + [ + 371, + 181 + ], + [ + 346, + 203 + ], + [ + 387, + 187 + ], + [ + 344, + 216 + ], + [ + 383, + 200 + ] + ], + [ + [ + 396, + 187 + ], + [ + 383, + 173 + ], + [ + 376, + 192 + ], + [ + 416, + 182 + ], + [ + 390, + 208 + ], + [ + 431, + 197 + ] + ] + ] + }, + { + "task_id": "0000195", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0008_cfg_0006_104122061018#403", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000195/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 351, + 129 + ], + [ + 327, + 143 + ], + [ + 357, + 110 + ], + [ + 346, + 145 + ], + [ + 382, + 97 + ], + [ + 368, + 132 + ] + ], + [ + [ + 339, + 156 + ], + [ + 346, + 132 + ], + [ + 323, + 136 + ], + [ + 356, + 177 + ], + [ + 318, + 158 + ], + [ + 349, + 199 + ] + ], + [ + [ + 373, + 172 + ], + [ + 364, + 162 + ], + [ + 370, + 142 + ], + [ + 376, + 201 + ], + [ + 380, + 152 + ], + [ + 386, + 214 + ] + ], + [ + [ + 339, + 165 + ], + [ + 354, + 150 + ], + [ + 359, + 158 + ], + [ + 320, + 172 + ], + [ + 342, + 176 + ], + [ + 303, + 189 + ] + ] + ] + }, + { + "task_id": "0000197", + "id": "RH20T_cfg7_task_0028_user_0020_scene_0003_cfg_0007_037522061512#187", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and close the drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000197/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 187, + 174 + ], + [ + 195, + 152 + ], + [ + 203, + 180 + ], + [ + 169, + 167 + ], + [ + 196, + 202 + ], + [ + 161, + 191 + ] + ], + [ + [ + 199, + 185 + ], + [ + 192, + 181 + ], + [ + 184, + 205 + ], + [ + 213, + 166 + ], + [ + 191, + 211 + ], + [ + 222, + 170 + ] + ], + [ + [ + 198, + 192 + ], + [ + 196, + 215 + ], + [ + 221, + 190 + ], + [ + 173, + 195 + ], + [ + 223, + 168 + ], + [ + 175, + 171 + ] + ], + [ + [ + 196, + 155 + ], + [ + 202, + 171 + ], + [ + 215, + 166 + ], + [ + 175, + 144 + ], + [ + 209, + 152 + ], + [ + 170, + 130 + ] + ] + ] + }, + { + "task_id": "0000198", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0002_cfg_0004_038522062288#173", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000198/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 227, + 186 + ], + [ + 240, + 210 + ], + [ + 222, + 201 + ], + [ + 233, + 170 + ], + [ + 210, + 178 + ], + [ + 219, + 146 + ] + ], + [ + [ + 257, + 181 + ], + [ + 255, + 165 + ], + [ + 235, + 198 + ], + [ + 278, + 164 + ], + [ + 238, + 212 + ], + [ + 279, + 179 + ] + ], + [ + [ + 248, + 202 + ], + [ + 271, + 211 + ], + [ + 257, + 223 + ], + [ + 239, + 180 + ], + [ + 236, + 214 + ], + [ + 217, + 171 + ] + ], + [ + [ + 235, + 165 + ], + [ + 228, + 188 + ], + [ + 218, + 148 + ], + [ + 253, + 182 + ], + [ + 225, + 124 + ], + [ + 261, + 157 + ] + ] + ] + }, + { + "task_id": "0000200", + "id": "RH20T_cfg6_task_0022_user_0014_scene_0006_cfg_0006_104122061602#175", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Wave the flag'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000200/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 435, + 196 + ], + [ + 415, + 199 + ], + [ + 427, + 173 + ], + [ + 443, + 220 + ], + [ + 446, + 172 + ], + [ + 462, + 217 + ] + ], + [ + [ + 458, + 188 + ], + [ + 475, + 181 + ], + [ + 469, + 211 + ], + [ + 448, + 165 + ], + [ + 453, + 217 + ], + [ + 433, + 173 + ] + ], + [ + [ + 462, + 199 + ], + [ + 449, + 214 + ], + [ + 439, + 191 + ], + [ + 487, + 208 + ], + [ + 452, + 178 + ], + [ + 500, + 194 + ] + ], + [ + [ + 444, + 203 + ], + [ + 466, + 191 + ], + [ + 436, + 190 + ], + [ + 452, + 215 + ], + [ + 414, + 203 + ], + [ + 431, + 226 + ] + ] + ] + }, + { + "task_id": "0000202", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_038522062547#195", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000202/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 417, + 99 + ], + [ + 440, + 94 + ], + [ + 411, + 104 + ], + [ + 423, + 95 + ], + [ + 389, + 109 + ], + [ + 399, + 101 + ] + ], + [ + [ + 431, + 110 + ], + [ + 438, + 98 + ], + [ + 435, + 133 + ], + [ + 428, + 88 + ], + [ + 429, + 143 + ], + [ + 423, + 100 + ] + ], + [ + [ + 451, + 80 + ], + [ + 425, + 82 + ], + [ + 452, + 57 + ], + [ + 450, + 102 + ], + [ + 479, + 54 + ], + [ + 476, + 100 + ] + ], + [ + [ + 412, + 135 + ], + [ + 391, + 145 + ], + [ + 423, + 156 + ], + [ + 400, + 114 + ], + [ + 445, + 145 + ], + [ + 421, + 102 + ] + ] + ] + }, + { + "task_id": "0000203", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_038522062547#285", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000203/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 444, + 80 + ], + [ + 449, + 66 + ], + [ + 450, + 100 + ], + [ + 438, + 60 + ], + [ + 445, + 112 + ], + [ + 434, + 73 + ] + ], + [ + [ + 457, + 108 + ], + [ + 452, + 119 + ], + [ + 434, + 105 + ], + [ + 481, + 112 + ], + [ + 439, + 93 + ], + [ + 488, + 100 + ] + ], + [ + [ + 462, + 92 + ], + [ + 447, + 89 + ], + [ + 461, + 69 + ], + [ + 462, + 116 + ], + [ + 475, + 72 + ], + [ + 476, + 118 + ] + ], + [ + [ + 444, + 95 + ], + [ + 420, + 101 + ], + [ + 437, + 72 + ], + [ + 449, + 117 + ], + [ + 463, + 65 + ], + [ + 474, + 111 + ] + ] + ] + }, + { + "task_id": "0000207", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0005_cfg_0005_104122061850#748", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000207/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 256, + 187 + ], + [ + 249, + 173 + ], + [ + 232, + 199 + ], + [ + 280, + 175 + ], + [ + 240, + 212 + ], + [ + 286, + 188 + ] + ], + [ + [ + 265, + 183 + ], + [ + 254, + 164 + ], + [ + 287, + 178 + ], + [ + 241, + 189 + ], + [ + 299, + 198 + ], + [ + 251, + 210 + ] + ], + [ + [ + 271, + 167 + ], + [ + 247, + 173 + ], + [ + 280, + 159 + ], + [ + 260, + 176 + ], + [ + 304, + 153 + ], + [ + 287, + 170 + ] + ], + [ + [ + 255, + 187 + ], + [ + 233, + 195 + ], + [ + 248, + 200 + ], + [ + 262, + 173 + ], + [ + 268, + 192 + ], + [ + 283, + 165 + ] + ] + ] + }, + { + "task_id": "0000208", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_104122063678#279", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000208/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 225 + ], + [ + 371, + 222 + ], + [ + 409, + 245 + ], + [ + 379, + 207 + ], + [ + 431, + 247 + ], + [ + 399, + 210 + ] + ], + [ + [ + 390, + 238 + ], + [ + 401, + 220 + ], + [ + 404, + 261 + ], + [ + 378, + 217 + ], + [ + 393, + 277 + ], + [ + 368, + 233 + ] + ], + [ + [ + 398, + 237 + ], + [ + 377, + 240 + ], + [ + 419, + 246 + ], + [ + 379, + 229 + ], + [ + 439, + 244 + ], + [ + 399, + 228 + ] + ], + [ + [ + 370, + 232 + ], + [ + 373, + 253 + ], + [ + 396, + 226 + ], + [ + 344, + 238 + ], + [ + 394, + 204 + ], + [ + 340, + 216 + ] + ] + ] + }, + { + "task_id": "0000209", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_104122063678#401", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000209/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 367, + 241 + ], + [ + 371, + 266 + ], + [ + 393, + 233 + ], + [ + 341, + 251 + ], + [ + 389, + 207 + ], + [ + 336, + 225 + ] + ], + [ + [ + 343, + 230 + ], + [ + 339, + 205 + ], + [ + 316, + 236 + ], + [ + 370, + 224 + ], + [ + 321, + 259 + ], + [ + 373, + 247 + ] + ], + [ + [ + 345, + 221 + ], + [ + 324, + 206 + ], + [ + 355, + 208 + ], + [ + 336, + 232 + ], + [ + 377, + 223 + ], + [ + 356, + 246 + ] + ], + [ + [ + 364, + 232 + ], + [ + 375, + 217 + ], + [ + 355, + 252 + ], + [ + 373, + 213 + ], + [ + 345, + 265 + ], + [ + 363, + 227 + ] + ] + ] + }, + { + "task_id": "0000212", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0008_cfg_0006_104122063633#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000212/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 232, + 184 + ], + [ + 234, + 160 + ], + [ + 215, + 180 + ], + [ + 250, + 189 + ], + [ + 213, + 203 + ], + [ + 248, + 213 + ] + ], + [ + [ + 234, + 189 + ], + [ + 230, + 166 + ], + [ + 256, + 185 + ], + [ + 213, + 192 + ], + [ + 260, + 208 + ], + [ + 217, + 215 + ] + ], + [ + [ + 243, + 198 + ], + [ + 244, + 174 + ], + [ + 249, + 201 + ], + [ + 236, + 194 + ], + [ + 247, + 223 + ], + [ + 234, + 219 + ] + ], + [ + [ + 251, + 187 + ], + [ + 233, + 189 + ], + [ + 256, + 210 + ], + [ + 247, + 165 + ], + [ + 273, + 207 + ], + [ + 264, + 164 + ] + ] + ] + }, + { + "task_id": "0000216", + "id": "RH20T_cfg6_task_0007_user_0014_scene_0007_cfg_0006_104122063633#208", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000216/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 236, + 112 + ], + [ + 248, + 112 + ], + [ + 219, + 111 + ], + [ + 254, + 114 + ], + [ + 207, + 111 + ], + [ + 242, + 115 + ] + ], + [ + [ + 242, + 118 + ], + [ + 252, + 119 + ], + [ + 259, + 120 + ], + [ + 224, + 116 + ], + [ + 249, + 120 + ], + [ + 215, + 116 + ] + ], + [ + [ + 205, + 101 + ], + [ + 185, + 102 + ], + [ + 208, + 122 + ], + [ + 202, + 80 + ], + [ + 226, + 121 + ], + [ + 221, + 80 + ] + ], + [ + [ + 222, + 123 + ], + [ + 225, + 103 + ], + [ + 223, + 120 + ], + [ + 220, + 126 + ], + [ + 220, + 140 + ], + [ + 217, + 147 + ] + ] + ] + }, + { + "task_id": "0000218", + "id": "RH20T_cfg7_task_0046_user_0014_scene_0001_cfg_0007_104122061602#108", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000218/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 464, + 146 + ], + [ + 466, + 123 + ], + [ + 464, + 157 + ], + [ + 463, + 135 + ], + [ + 462, + 180 + ], + [ + 460, + 160 + ] + ], + [ + [ + 459, + 150 + ], + [ + 437, + 140 + ], + [ + 464, + 128 + ], + [ + 453, + 173 + ], + [ + 486, + 138 + ], + [ + 475, + 182 + ] + ], + [ + [ + 445, + 130 + ], + [ + 450, + 107 + ], + [ + 463, + 142 + ], + [ + 429, + 119 + ], + [ + 458, + 165 + ], + [ + 425, + 141 + ] + ], + [ + [ + 448, + 140 + ], + [ + 454, + 151 + ], + [ + 472, + 129 + ], + [ + 424, + 151 + ], + [ + 465, + 121 + ], + [ + 420, + 142 + ] + ] + ] + }, + { + "task_id": "0000219", + "id": "RH20T_cfg5_task_0013_user_0007_scene_0001_cfg_0005_104122061850#244", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000219/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 433, + 89 + ], + [ + 434, + 72 + ], + [ + 424, + 108 + ], + [ + 442, + 71 + ], + [ + 424, + 124 + ], + [ + 441, + 88 + ] + ], + [ + [ + 453, + 116 + ], + [ + 435, + 101 + ], + [ + 436, + 124 + ], + [ + 472, + 106 + ], + [ + 454, + 138 + ], + [ + 491, + 121 + ] + ], + [ + [ + 419, + 108 + ], + [ + 401, + 111 + ], + [ + 405, + 121 + ], + [ + 435, + 94 + ], + [ + 422, + 118 + ], + [ + 453, + 91 + ] + ], + [ + [ + 414, + 74 + ], + [ + 390, + 80 + ], + [ + 421, + 66 + ], + [ + 408, + 81 + ], + [ + 446, + 59 + ], + [ + 430, + 75 + ] + ] + ] + }, + { + "task_id": "0000220", + "id": "RH20T_cfg1_task_0029_user_0001_scene_0004_cfg_0001_819112070044#194", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000220/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 311, + 194 + ], + [ + 309, + 175 + ], + [ + 329, + 187 + ], + [ + 292, + 201 + ], + [ + 331, + 204 + ], + [ + 295, + 219 + ] + ], + [ + [ + 311, + 187 + ], + [ + 296, + 175 + ], + [ + 324, + 184 + ], + [ + 300, + 189 + ], + [ + 338, + 196 + ], + [ + 314, + 201 + ] + ], + [ + [ + 319, + 186 + ], + [ + 299, + 186 + ], + [ + 312, + 203 + ], + [ + 325, + 170 + ], + [ + 333, + 204 + ], + [ + 344, + 171 + ] + ], + [ + [ + 307, + 196 + ], + [ + 305, + 177 + ], + [ + 316, + 185 + ], + [ + 298, + 205 + ], + [ + 318, + 205 + ], + [ + 299, + 224 + ] + ] + ] + }, + { + "task_id": "0000224", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0005_cfg_0005_104122063678#339", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000224/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 201 + ], + [ + 342, + 184 + ], + [ + 339, + 223 + ], + [ + 330, + 181 + ], + [ + 331, + 238 + ], + [ + 323, + 197 + ] + ], + [ + [ + 339, + 211 + ], + [ + 327, + 191 + ], + [ + 362, + 200 + ], + [ + 315, + 222 + ], + [ + 373, + 219 + ], + [ + 327, + 240 + ] + ], + [ + [ + 318, + 216 + ], + [ + 298, + 210 + ], + [ + 313, + 195 + ], + [ + 323, + 238 + ], + [ + 332, + 201 + ], + [ + 343, + 244 + ] + ], + [ + [ + 353, + 181 + ], + [ + 355, + 179 + ], + [ + 379, + 189 + ], + [ + 328, + 173 + ], + [ + 377, + 191 + ], + [ + 328, + 176 + ] + ] + ] + }, + { + "task_id": "0000228", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0007_cfg_0004_104122062295#116", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000228/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 218, + 298 + ], + [ + 226, + 259 + ], + [ + 170, + 279 + ], + [ + 267, + 317 + ], + [ + 165, + 314 + ], + [ + 257, + 351 + ] + ], + [ + [ + 240, + 337 + ], + [ + 214, + 307 + ], + [ + 223, + 291 + ], + [ + 259, + 360 + ], + [ + 244, + 317 + ], + [ + 281, + 360 + ] + ], + [ + [ + 183, + 283 + ], + [ + 154, + 334 + ], + [ + 159, + 243 + ], + [ + 204, + 318 + ], + [ + 187, + 202 + ], + [ + 225, + 276 + ] + ], + [ + [ + 273, + 269 + ], + [ + 294, + 245 + ], + [ + 251, + 305 + ], + [ + 298, + 230 + ], + [ + 224, + 336 + ], + [ + 273, + 256 + ] + ] + ] + }, + { + "task_id": "0000229", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0007_cfg_0004_104122062295#250", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000229/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 250, + 144 + ], + [ + 242, + 174 + ], + [ + 279, + 156 + ], + [ + 221, + 133 + ], + [ + 288, + 125 + ], + [ + 228, + 101 + ] + ], + [ + [ + 287, + 159 + ], + [ + 285, + 190 + ], + [ + 322, + 152 + ], + [ + 254, + 166 + ], + [ + 322, + 124 + ], + [ + 257, + 138 + ] + ], + [ + [ + 291, + 164 + ], + [ + 290, + 149 + ], + [ + 309, + 189 + ], + [ + 270, + 139 + ], + [ + 311, + 207 + ], + [ + 270, + 155 + ] + ], + [ + [ + 286, + 148 + ], + [ + 295, + 116 + ], + [ + 250, + 139 + ], + [ + 320, + 156 + ], + [ + 243, + 170 + ], + [ + 310, + 185 + ] + ] + ] + }, + { + "task_id": "0000232", + "id": "RH20T_cfg2_task_0054_user_0007_scene_0009_cfg_0002_104422070044#64", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the moving object'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000232/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 343, + 185 + ], + [ + 330, + 176 + ], + [ + 350, + 188 + ], + [ + 335, + 183 + ], + [ + 363, + 197 + ], + [ + 348, + 193 + ] + ], + [ + [ + 360, + 187 + ], + [ + 366, + 174 + ], + [ + 350, + 190 + ], + [ + 370, + 183 + ], + [ + 343, + 203 + ], + [ + 364, + 197 + ] + ], + [ + [ + 345, + 190 + ], + [ + 335, + 202 + ], + [ + 358, + 194 + ], + [ + 332, + 185 + ], + [ + 367, + 182 + ], + [ + 341, + 173 + ] + ], + [ + [ + 357, + 180 + ], + [ + 356, + 170 + ], + [ + 362, + 194 + ], + [ + 353, + 166 + ], + [ + 363, + 203 + ], + [ + 355, + 176 + ] + ] + ] + }, + { + "task_id": "0000233", + "id": "RH20T_cfg5_task_0047_user_0007_scene_0009_cfg_0005_036422060215#278", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000233/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 187, + 149 + ], + [ + 187, + 135 + ], + [ + 208, + 158 + ], + [ + 164, + 140 + ], + [ + 208, + 173 + ], + [ + 163, + 155 + ] + ], + [ + [ + 184, + 123 + ], + [ + 173, + 135 + ], + [ + 168, + 104 + ], + [ + 199, + 141 + ], + [ + 178, + 93 + ], + [ + 208, + 129 + ] + ], + [ + [ + 163, + 142 + ], + [ + 166, + 165 + ], + [ + 163, + 153 + ], + [ + 163, + 130 + ], + [ + 161, + 132 + ], + [ + 160, + 107 + ] + ], + [ + [ + 167, + 127 + ], + [ + 171, + 107 + ], + [ + 192, + 132 + ], + [ + 141, + 122 + ], + [ + 188, + 151 + ], + [ + 138, + 142 + ] + ] + ] + }, + { + "task_id": "0000237", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122060811#821", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000237/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 333, + 153 + ], + [ + 323, + 146 + ], + [ + 324, + 156 + ], + [ + 341, + 151 + ], + [ + 335, + 164 + ], + [ + 352, + 158 + ] + ], + [ + [ + 335, + 155 + ], + [ + 333, + 169 + ], + [ + 349, + 158 + ], + [ + 321, + 152 + ], + [ + 351, + 144 + ], + [ + 324, + 139 + ] + ], + [ + [ + 356, + 147 + ], + [ + 370, + 143 + ], + [ + 353, + 147 + ], + [ + 358, + 147 + ], + [ + 339, + 151 + ], + [ + 344, + 150 + ] + ], + [ + [ + 341, + 142 + ], + [ + 353, + 135 + ], + [ + 347, + 146 + ], + [ + 334, + 138 + ], + [ + 335, + 152 + ], + [ + 322, + 145 + ] + ] + ] + }, + { + "task_id": "0000238", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122060811#1066", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000238/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 375, + 133 + ], + [ + 379, + 126 + ], + [ + 388, + 130 + ], + [ + 362, + 136 + ], + [ + 384, + 137 + ], + [ + 359, + 143 + ] + ], + [ + [ + 372, + 128 + ], + [ + 379, + 134 + ], + [ + 383, + 125 + ], + [ + 360, + 131 + ], + [ + 375, + 119 + ], + [ + 353, + 125 + ] + ], + [ + [ + 371, + 153 + ], + [ + 371, + 166 + ], + [ + 361, + 157 + ], + [ + 382, + 149 + ], + [ + 361, + 144 + ], + [ + 382, + 136 + ] + ], + [ + [ + 374, + 119 + ], + [ + 379, + 115 + ], + [ + 377, + 106 + ], + [ + 371, + 133 + ], + [ + 372, + 111 + ], + [ + 366, + 137 + ] + ] + ] + }, + { + "task_id": "0000240", + "id": "RH20T_cfg5_task_0072_user_0007_scene_0005_cfg_0005_037522062165#360", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drop coins into a piggy bank'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000240/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 195, + 269 + ], + [ + 192, + 253 + ], + [ + 170, + 254 + ], + [ + 222, + 286 + ], + [ + 174, + 270 + ], + [ + 223, + 300 + ] + ], + [ + [ + 182, + 300 + ], + [ + 186, + 267 + ], + [ + 166, + 297 + ], + [ + 200, + 304 + ], + [ + 163, + 330 + ], + [ + 196, + 340 + ] + ], + [ + [ + 185, + 280 + ], + [ + 204, + 255 + ], + [ + 211, + 297 + ], + [ + 160, + 264 + ], + [ + 192, + 324 + ], + [ + 139, + 290 + ] + ], + [ + [ + 191, + 310 + ], + [ + 175, + 341 + ], + [ + 199, + 304 + ], + [ + 181, + 318 + ], + [ + 213, + 276 + ], + [ + 197, + 285 + ] + ] + ] + }, + { + "task_id": "0000243", + "id": "RH20T_cfg1_task_0004_user_0005_scene_0003_cfg_0001_038522063145#125", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000243/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 463, + 186 + ], + [ + 473, + 167 + ], + [ + 485, + 201 + ], + [ + 443, + 172 + ], + [ + 475, + 219 + ], + [ + 435, + 190 + ] + ], + [ + [ + 461, + 213 + ], + [ + 443, + 225 + ], + [ + 466, + 200 + ], + [ + 456, + 226 + ], + [ + 485, + 187 + ], + [ + 477, + 213 + ] + ], + [ + [ + 448, + 201 + ], + [ + 432, + 217 + ], + [ + 460, + 220 + ], + [ + 436, + 181 + ], + [ + 475, + 204 + ], + [ + 452, + 165 + ] + ], + [ + [ + 467, + 197 + ], + [ + 480, + 187 + ], + [ + 449, + 179 + ], + [ + 486, + 216 + ], + [ + 434, + 189 + ], + [ + 472, + 227 + ] + ] + ] + }, + { + "task_id": "0000244", + "id": "RH20T_cfg4_task_0033_user_0016_scene_0008_cfg_0004_038522062288#143", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000244/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 131, + 121 + ], + [ + 150, + 104 + ], + [ + 150, + 124 + ], + [ + 109, + 118 + ], + [ + 131, + 140 + ], + [ + 90, + 135 + ] + ], + [ + [ + 149, + 115 + ], + [ + 136, + 95 + ], + [ + 123, + 118 + ], + [ + 175, + 113 + ], + [ + 136, + 136 + ], + [ + 185, + 131 + ] + ], + [ + [ + 163, + 106 + ], + [ + 159, + 126 + ], + [ + 139, + 114 + ], + [ + 186, + 98 + ], + [ + 142, + 91 + ], + [ + 191, + 77 + ] + ], + [ + [ + 175, + 148 + ], + [ + 183, + 127 + ], + [ + 163, + 157 + ], + [ + 188, + 138 + ], + [ + 155, + 179 + ], + [ + 180, + 161 + ] + ] + ] + }, + { + "task_id": "0000246", + "id": "RH20T_cfg1_task_0043_user_0015_scene_0004_cfg_0001_819112071037#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000246/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 122 + ], + [ + 437, + 95 + ], + [ + 381, + 145 + ], + [ + 437, + 94 + ], + [ + 348, + 174 + ], + [ + 400, + 127 + ] + ], + [ + [ + 421, + 89 + ], + [ + 426, + 22 + ], + [ + 383, + 101 + ], + [ + 452, + 80 + ], + [ + 379, + 166 + ], + [ + 446, + 135 + ] + ], + [ + [ + 360, + 117 + ], + [ + 413, + 145 + ], + [ + 364, + 121 + ], + [ + 354, + 111 + ], + [ + 318, + 97 + ], + [ + 296, + 79 + ] + ], + [ + [ + 428, + 140 + ], + [ + 407, + 163 + ], + [ + 437, + 84 + ], + [ + 420, + 189 + ], + [ + 455, + 69 + ], + [ + 437, + 167 + ] + ] + ] + }, + { + "task_id": "0000249", + "id": "RH20T_cfg7_task_0047_user_0020_scene_0004_cfg_0007_104122061330#161", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000249/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 324, + 152 + ], + [ + 295, + 147 + ], + [ + 319, + 183 + ], + [ + 329, + 121 + ], + [ + 349, + 188 + ], + [ + 360, + 125 + ] + ], + [ + [ + 347, + 148 + ], + [ + 330, + 166 + ], + [ + 357, + 173 + ], + [ + 335, + 122 + ], + [ + 372, + 156 + ], + [ + 352, + 106 + ] + ], + [ + [ + 351, + 156 + ], + [ + 371, + 166 + ], + [ + 371, + 147 + ], + [ + 330, + 165 + ], + [ + 352, + 138 + ], + [ + 312, + 155 + ] + ], + [ + [ + 356, + 115 + ], + [ + 347, + 92 + ], + [ + 366, + 134 + ], + [ + 347, + 98 + ], + [ + 374, + 155 + ], + [ + 356, + 119 + ] + ] + ] + }, + { + "task_id": "0000251", + "id": "RH20T_cfg6_task_0078_user_0014_scene_0008_cfg_0006_104122060811#111", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close the microwave door'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000251/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 391, + 89 + ], + [ + 396, + 77 + ], + [ + 405, + 85 + ], + [ + 377, + 93 + ], + [ + 401, + 96 + ], + [ + 374, + 104 + ] + ], + [ + [ + 394, + 79 + ], + [ + 401, + 83 + ], + [ + 403, + 64 + ], + [ + 386, + 93 + ], + [ + 396, + 60 + ], + [ + 380, + 89 + ] + ], + [ + [ + 377, + 98 + ], + [ + 378, + 82 + ], + [ + 382, + 102 + ], + [ + 371, + 95 + ], + [ + 381, + 119 + ], + [ + 370, + 110 + ] + ], + [ + [ + 376, + 77 + ], + [ + 375, + 78 + ], + [ + 367, + 63 + ], + [ + 386, + 91 + ], + [ + 367, + 62 + ], + [ + 387, + 90 + ] + ] + ] + }, + { + "task_id": "0000252", + "id": "RH20T_cfg5_task_0047_user_0007_scene_0009_cfg_0005_104422070044#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000252/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 117 + ], + [ + 346, + 103 + ], + [ + 361, + 110 + ], + [ + 335, + 124 + ], + [ + 363, + 125 + ], + [ + 337, + 139 + ] + ], + [ + [ + 361, + 110 + ], + [ + 361, + 97 + ], + [ + 369, + 121 + ], + [ + 353, + 100 + ], + [ + 369, + 134 + ], + [ + 354, + 112 + ] + ], + [ + [ + 346, + 114 + ], + [ + 330, + 118 + ], + [ + 352, + 130 + ], + [ + 340, + 98 + ], + [ + 368, + 125 + ], + [ + 356, + 94 + ] + ], + [ + [ + 359, + 107 + ], + [ + 375, + 106 + ], + [ + 366, + 105 + ], + [ + 353, + 108 + ], + [ + 349, + 106 + ], + [ + 337, + 109 + ] + ] + ] + }, + { + "task_id": "0000255", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122061330#220", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000255/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 368, + 183 + ], + [ + 363, + 204 + ], + [ + 344, + 177 + ], + [ + 393, + 188 + ], + [ + 349, + 154 + ], + [ + 399, + 166 + ] + ], + [ + [ + 372, + 184 + ], + [ + 368, + 171 + ], + [ + 361, + 207 + ], + [ + 382, + 163 + ], + [ + 365, + 218 + ], + [ + 386, + 176 + ] + ], + [ + [ + 373, + 177 + ], + [ + 389, + 161 + ], + [ + 372, + 191 + ], + [ + 374, + 161 + ], + [ + 356, + 207 + ], + [ + 357, + 177 + ] + ], + [ + [ + 348, + 166 + ], + [ + 333, + 169 + ], + [ + 341, + 140 + ], + [ + 356, + 192 + ], + [ + 355, + 138 + ], + [ + 370, + 188 + ] + ] + ] + }, + { + "task_id": "0000256", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122061330#555", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000256/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 235 + ], + [ + 324, + 255 + ], + [ + 287, + 249 + ], + [ + 331, + 221 + ], + [ + 273, + 230 + ], + [ + 316, + 203 + ] + ], + [ + [ + 319, + 223 + ], + [ + 304, + 232 + ], + [ + 312, + 246 + ], + [ + 326, + 199 + ], + [ + 325, + 237 + ], + [ + 339, + 192 + ] + ], + [ + [ + 323, + 237 + ], + [ + 303, + 217 + ], + [ + 327, + 230 + ], + [ + 319, + 245 + ], + [ + 347, + 249 + ], + [ + 341, + 267 + ] + ], + [ + [ + 308, + 231 + ], + [ + 289, + 211 + ], + [ + 321, + 212 + ], + [ + 296, + 248 + ], + [ + 341, + 233 + ], + [ + 315, + 268 + ] + ] + ] + }, + { + "task_id": "0000257", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122061330#828", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000257/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 355, + 251 + ], + [ + 331, + 238 + ], + [ + 350, + 275 + ], + [ + 361, + 225 + ], + [ + 376, + 289 + ], + [ + 387, + 238 + ] + ], + [ + [ + 358, + 245 + ], + [ + 358, + 223 + ], + [ + 341, + 264 + ], + [ + 373, + 229 + ], + [ + 342, + 285 + ], + [ + 372, + 249 + ] + ], + [ + [ + 356, + 234 + ], + [ + 339, + 226 + ], + [ + 335, + 252 + ], + [ + 375, + 218 + ], + [ + 353, + 262 + ], + [ + 394, + 226 + ] + ], + [ + [ + 367, + 291 + ], + [ + 352, + 318 + ], + [ + 394, + 307 + ], + [ + 342, + 276 + ], + [ + 410, + 279 + ], + [ + 357, + 250 + ] + ] + ] + }, + { + "task_id": "0000258", + "id": "RH20T_cfg7_task_0055_user_0020_scene_0007_cfg_0007_104122063633#149", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000258/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 188, + 179 + ], + [ + 190, + 160 + ], + [ + 194, + 162 + ], + [ + 182, + 197 + ], + [ + 192, + 180 + ], + [ + 181, + 215 + ] + ], + [ + [ + 164, + 204 + ], + [ + 137, + 206 + ], + [ + 159, + 212 + ], + [ + 169, + 197 + ], + [ + 187, + 211 + ], + [ + 195, + 196 + ] + ], + [ + [ + 207, + 180 + ], + [ + 202, + 163 + ], + [ + 225, + 189 + ], + [ + 189, + 171 + ], + [ + 229, + 206 + ], + [ + 194, + 188 + ] + ], + [ + [ + 219, + 160 + ], + [ + 202, + 174 + ], + [ + 209, + 139 + ], + [ + 230, + 182 + ], + [ + 227, + 125 + ], + [ + 249, + 168 + ] + ] + ] + }, + { + "task_id": "0000260", + "id": "RH20T_cfg7_task_0055_user_0020_scene_0007_cfg_0007_104122063633#395", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000260/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 219, + 139 + ], + [ + 226, + 139 + ], + [ + 239, + 133 + ], + [ + 198, + 145 + ], + [ + 232, + 132 + ], + [ + 189, + 145 + ] + ], + [ + [ + 217, + 130 + ], + [ + 198, + 141 + ], + [ + 226, + 150 + ], + [ + 208, + 110 + ], + [ + 246, + 139 + ], + [ + 227, + 99 + ] + ], + [ + [ + 196, + 144 + ], + [ + 198, + 122 + ], + [ + 200, + 135 + ], + [ + 191, + 152 + ], + [ + 198, + 156 + ], + [ + 189, + 174 + ] + ], + [ + [ + 180, + 138 + ], + [ + 158, + 125 + ], + [ + 166, + 157 + ], + [ + 193, + 119 + ], + [ + 187, + 168 + ], + [ + 214, + 132 + ] + ] + ] + }, + { + "task_id": "0000262", + "id": "RH20T_cfg1_task_0070_user_0014_scene_0008_cfg_0001_035622060973#155", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000262/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 215, + 174 + ], + [ + 200, + 183 + ], + [ + 230, + 179 + ], + [ + 201, + 169 + ], + [ + 244, + 171 + ], + [ + 215, + 162 + ] + ], + [ + [ + 219, + 185 + ], + [ + 217, + 176 + ], + [ + 209, + 169 + ], + [ + 229, + 201 + ], + [ + 212, + 178 + ], + [ + 231, + 209 + ] + ], + [ + [ + 200, + 184 + ], + [ + 214, + 174 + ], + [ + 193, + 192 + ], + [ + 209, + 175 + ], + [ + 179, + 202 + ], + [ + 194, + 185 + ] + ], + [ + [ + 232, + 189 + ], + [ + 252, + 186 + ], + [ + 221, + 190 + ], + [ + 243, + 188 + ], + [ + 201, + 193 + ], + [ + 225, + 191 + ] + ] + ] + }, + { + "task_id": "0000267", + "id": "RH20T_cfg5_task_0058_user_0007_scene_0004_cfg_0005_104122061850#340", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000267/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 304, + 123 + ], + [ + 316, + 139 + ], + [ + 322, + 104 + ], + [ + 286, + 140 + ], + [ + 310, + 89 + ], + [ + 275, + 124 + ] + ], + [ + [ + 348, + 146 + ], + [ + 355, + 124 + ], + [ + 331, + 135 + ], + [ + 364, + 156 + ], + [ + 323, + 158 + ], + [ + 358, + 178 + ] + ], + [ + [ + 331, + 133 + ], + [ + 333, + 112 + ], + [ + 318, + 145 + ], + [ + 343, + 122 + ], + [ + 317, + 166 + ], + [ + 340, + 142 + ] + ], + [ + [ + 345, + 138 + ], + [ + 330, + 121 + ], + [ + 362, + 128 + ], + [ + 326, + 149 + ], + [ + 377, + 145 + ], + [ + 341, + 167 + ] + ] + ] + }, + { + "task_id": "0000269", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0004_cfg_0006_104122061330#213", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000269/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 421, + 123 + ], + [ + 395, + 123 + ], + [ + 421, + 150 + ], + [ + 422, + 96 + ], + [ + 448, + 150 + ], + [ + 449, + 95 + ] + ], + [ + [ + 433, + 144 + ], + [ + 410, + 130 + ], + [ + 416, + 167 + ], + [ + 450, + 121 + ], + [ + 439, + 183 + ], + [ + 474, + 136 + ] + ], + [ + [ + 391, + 122 + ], + [ + 373, + 117 + ], + [ + 377, + 106 + ], + [ + 407, + 140 + ], + [ + 394, + 111 + ], + [ + 424, + 144 + ] + ], + [ + [ + 418, + 127 + ], + [ + 411, + 109 + ], + [ + 408, + 148 + ], + [ + 428, + 107 + ], + [ + 415, + 164 + ], + [ + 434, + 124 + ] + ] + ] + }, + { + "task_id": "0000271", + "id": "RH20T_cfg1_task_0033_user_0014_scene_0006_cfg_0001_750612070851#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000271/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 333, + 200 + ], + [ + 321, + 185 + ], + [ + 341, + 170 + ], + [ + 325, + 232 + ], + [ + 351, + 185 + ], + [ + 337, + 244 + ] + ], + [ + [ + 316, + 208 + ], + [ + 313, + 190 + ], + [ + 282, + 211 + ], + [ + 351, + 205 + ], + [ + 286, + 227 + ], + [ + 351, + 221 + ] + ], + [ + [ + 346, + 236 + ], + [ + 345, + 235 + ], + [ + 383, + 239 + ], + [ + 310, + 233 + ], + [ + 380, + 240 + ], + [ + 313, + 234 + ] + ], + [ + [ + 309, + 195 + ], + [ + 277, + 184 + ], + [ + 301, + 230 + ], + [ + 316, + 162 + ], + [ + 332, + 240 + ], + [ + 345, + 173 + ] + ] + ] + }, + { + "task_id": "0000272", + "id": "RH20T_cfg6_task_0015_user_0014_scene_0005_cfg_0006_104122061602#175", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000272/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 487, + 192 + ], + [ + 504, + 193 + ], + [ + 507, + 201 + ], + [ + 468, + 185 + ], + [ + 488, + 201 + ], + [ + 451, + 184 + ] + ], + [ + [ + 486, + 195 + ], + [ + 507, + 189 + ], + [ + 493, + 219 + ], + [ + 480, + 171 + ], + [ + 474, + 224 + ], + [ + 462, + 177 + ] + ], + [ + [ + 460, + 178 + ], + [ + 463, + 159 + ], + [ + 486, + 181 + ], + [ + 434, + 175 + ], + [ + 483, + 199 + ], + [ + 432, + 193 + ] + ], + [ + [ + 489, + 200 + ], + [ + 471, + 219 + ], + [ + 486, + 197 + ], + [ + 493, + 202 + ], + [ + 504, + 178 + ], + [ + 512, + 181 + ] + ] + ] + }, + { + "task_id": "0000275", + "id": "RH20T_cfg6_task_0106_user_0020_scene_0004_cfg_0006_104122061602#203", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the object down from the shelf'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000275/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 597, + 102 + ], + [ + 586, + 94 + ], + [ + 639, + 80 + ], + [ + 559, + 123 + ], + [ + 640, + 87 + ], + [ + 568, + 133 + ] + ], + [ + [ + 582, + 145 + ], + [ + 602, + 116 + ], + [ + 613, + 172 + ], + [ + 554, + 122 + ], + [ + 593, + 197 + ], + [ + 539, + 148 + ] + ], + [ + [ + 578, + 147 + ], + [ + 568, + 169 + ], + [ + 604, + 122 + ], + [ + 551, + 172 + ], + [ + 618, + 95 + ], + [ + 560, + 147 + ] + ], + [ + [ + 615, + 121 + ], + [ + 630, + 157 + ], + [ + 573, + 136 + ], + [ + 640, + 105 + ], + [ + 561, + 104 + ], + [ + 640, + 71 + ] + ] + ] + }, + { + "task_id": "0000278", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122064161#454", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000278/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 313, + 142 + ], + [ + 292, + 133 + ], + [ + 307, + 160 + ], + [ + 319, + 124 + ], + [ + 328, + 170 + ], + [ + 341, + 134 + ] + ], + [ + [ + 334, + 156 + ], + [ + 335, + 146 + ], + [ + 332, + 177 + ], + [ + 335, + 136 + ], + [ + 331, + 186 + ], + [ + 335, + 147 + ] + ], + [ + [ + 325, + 165 + ], + [ + 340, + 170 + ], + [ + 318, + 187 + ], + [ + 331, + 142 + ], + [ + 301, + 182 + ], + [ + 315, + 135 + ] + ], + [ + [ + 355, + 154 + ], + [ + 334, + 145 + ], + [ + 363, + 132 + ], + [ + 347, + 175 + ], + [ + 385, + 140 + ], + [ + 368, + 183 + ] + ] + ] + }, + { + "task_id": "0000281", + "id": "RH20T_cfg4_task_0016_user_0016_scene_0006_cfg_0004_036422060909#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000281/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 268, + 139 + ], + [ + 282, + 163 + ], + [ + 245, + 153 + ], + [ + 293, + 126 + ], + [ + 231, + 129 + ], + [ + 279, + 102 + ] + ], + [ + [ + 274, + 106 + ], + [ + 268, + 97 + ], + [ + 272, + 133 + ], + [ + 275, + 78 + ], + [ + 278, + 144 + ], + [ + 282, + 87 + ] + ], + [ + [ + 270, + 164 + ], + [ + 279, + 163 + ], + [ + 273, + 191 + ], + [ + 267, + 136 + ], + [ + 265, + 190 + ], + [ + 260, + 138 + ] + ], + [ + [ + 247, + 138 + ], + [ + 246, + 118 + ], + [ + 254, + 118 + ], + [ + 239, + 159 + ], + [ + 254, + 136 + ], + [ + 239, + 177 + ] + ] + ] + }, + { + "task_id": "0000290", + "id": "RH20T_cfg4_task_0002_user_0010_scene_0007_cfg_0004_104122062823#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000290/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 127, + 26 + ], + [ + 156, + 13 + ], + [ + 106, + 0 + ], + [ + 147, + 56 + ], + [ + 78, + 7 + ], + [ + 121, + 67 + ] + ], + [ + [ + 120, + 50 + ], + [ + 116, + 32 + ], + [ + 108, + 22 + ], + [ + 133, + 80 + ], + [ + 113, + 39 + ], + [ + 136, + 94 + ] + ], + [ + [ + 131, + 48 + ], + [ + 139, + 19 + ], + [ + 163, + 64 + ], + [ + 97, + 31 + ], + [ + 155, + 89 + ], + [ + 91, + 58 + ] + ], + [ + [ + 141, + 74 + ], + [ + 120, + 85 + ], + [ + 148, + 108 + ], + [ + 132, + 38 + ], + [ + 165, + 97 + ], + [ + 151, + 30 + ] + ] + ] + }, + { + "task_id": "0000292", + "id": "RH20T_cfg4_task_0013_user_0010_scene_0002_cfg_0004_104122062823#124", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000292/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 153, + 23 + ], + [ + 142, + 35 + ], + [ + 130, + 0 + ], + [ + 175, + 46 + ], + [ + 142, + 0 + ], + [ + 189, + 35 + ] + ], + [ + [ + 174, + 19 + ], + [ + 195, + 7 + ], + [ + 191, + 42 + ], + [ + 155, + 0 + ], + [ + 170, + 55 + ], + [ + 132, + 8 + ] + ], + [ + [ + 163, + 38 + ], + [ + 158, + 14 + ], + [ + 178, + 53 + ], + [ + 148, + 23 + ], + [ + 182, + 75 + ], + [ + 153, + 45 + ] + ], + [ + [ + 158, + 27 + ], + [ + 149, + 4 + ], + [ + 150, + 10 + ], + [ + 165, + 43 + ], + [ + 160, + 35 + ], + [ + 174, + 66 + ] + ] + ] + }, + { + "task_id": "0000293", + "id": "RH20T_cfg1_task_0064_user_0015_scene_0010_cfg_0001_750612070851#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000293/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 425, + 252 + ], + [ + 451, + 269 + ], + [ + 435, + 284 + ], + [ + 415, + 217 + ], + [ + 412, + 268 + ], + [ + 393, + 205 + ] + ], + [ + [ + 439, + 241 + ], + [ + 438, + 231 + ], + [ + 412, + 272 + ], + [ + 467, + 211 + ], + [ + 414, + 279 + ], + [ + 465, + 223 + ] + ], + [ + [ + 426, + 240 + ], + [ + 408, + 244 + ], + [ + 434, + 202 + ], + [ + 418, + 277 + ], + [ + 449, + 202 + ], + [ + 434, + 272 + ] + ], + [ + [ + 484, + 246 + ], + [ + 444, + 223 + ], + [ + 491, + 248 + ], + [ + 476, + 243 + ], + [ + 532, + 272 + ], + [ + 525, + 270 + ] + ] + ] + }, + { + "task_id": "0000295", + "id": "RH20T_cfg1_task_0103_user_0010_scene_0005_cfg_0001_750612070851#86", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000295/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 322, + 197 + ], + [ + 352, + 193 + ], + [ + 308, + 190 + ], + [ + 338, + 204 + ], + [ + 277, + 194 + ], + [ + 305, + 209 + ] + ], + [ + [ + 333, + 208 + ], + [ + 355, + 184 + ], + [ + 318, + 183 + ], + [ + 349, + 234 + ], + [ + 298, + 205 + ], + [ + 326, + 256 + ] + ], + [ + [ + 338, + 222 + ], + [ + 345, + 205 + ], + [ + 305, + 218 + ], + [ + 372, + 227 + ], + [ + 301, + 233 + ], + [ + 364, + 242 + ] + ], + [ + [ + 326, + 210 + ], + [ + 353, + 212 + ], + [ + 334, + 235 + ], + [ + 318, + 187 + ], + [ + 304, + 233 + ], + [ + 290, + 185 + ] + ] + ] + }, + { + "task_id": "0000297", + "id": "RH20T_cfg5_task_0060_user_0007_scene_0010_cfg_0005_104122063678#492", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to decrease the volume of the speaker'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000297/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 322, + 229 + ], + [ + 333, + 250 + ], + [ + 320, + 215 + ], + [ + 324, + 241 + ], + [ + 309, + 195 + ], + [ + 314, + 221 + ] + ], + [ + [ + 314, + 218 + ], + [ + 319, + 204 + ], + [ + 289, + 223 + ], + [ + 338, + 214 + ], + [ + 285, + 237 + ], + [ + 333, + 227 + ] + ], + [ + [ + 322, + 238 + ], + [ + 324, + 214 + ], + [ + 320, + 229 + ], + [ + 324, + 245 + ], + [ + 319, + 255 + ], + [ + 322, + 269 + ] + ], + [ + [ + 316, + 191 + ], + [ + 321, + 167 + ], + [ + 335, + 188 + ], + [ + 299, + 193 + ], + [ + 331, + 214 + ], + [ + 293, + 217 + ] + ] + ] + }, + { + "task_id": "0000298", + "id": "RH20T_cfg1_task_0016_user_0001_scene_0004_cfg_0001_819112070044#191", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000298/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 448, + 145 + ], + [ + 428, + 140 + ], + [ + 447, + 131 + ], + [ + 450, + 157 + ], + [ + 469, + 136 + ], + [ + 471, + 163 + ] + ], + [ + [ + 451, + 126 + ], + [ + 461, + 115 + ], + [ + 430, + 126 + ], + [ + 471, + 125 + ], + [ + 421, + 136 + ], + [ + 460, + 136 + ] + ], + [ + [ + 444, + 168 + ], + [ + 464, + 178 + ], + [ + 434, + 182 + ], + [ + 455, + 153 + ], + [ + 415, + 172 + ], + [ + 435, + 142 + ] + ], + [ + [ + 438, + 152 + ], + [ + 448, + 137 + ], + [ + 426, + 135 + ], + [ + 452, + 169 + ], + [ + 417, + 149 + ], + [ + 442, + 182 + ] + ] + ] + }, + { + "task_id": "0000301", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122061018#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000301/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 443, + 197 + ], + [ + 418, + 202 + ], + [ + 426, + 206 + ], + [ + 461, + 186 + ], + [ + 449, + 201 + ], + [ + 485, + 181 + ] + ], + [ + [ + 420, + 180 + ], + [ + 418, + 150 + ], + [ + 401, + 175 + ], + [ + 443, + 186 + ], + [ + 403, + 203 + ], + [ + 445, + 217 + ] + ], + [ + [ + 386, + 208 + ], + [ + 391, + 232 + ], + [ + 410, + 216 + ], + [ + 363, + 201 + ], + [ + 406, + 190 + ], + [ + 358, + 177 + ] + ], + [ + [ + 413, + 164 + ], + [ + 438, + 144 + ], + [ + 399, + 141 + ], + [ + 426, + 185 + ], + [ + 374, + 161 + ], + [ + 402, + 203 + ] + ] + ] + }, + { + "task_id": "0000305", + "id": "RH20T_cfg6_task_0029_user_0014_scene_0005_cfg_0006_104122061018#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000305/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 375, + 42 + ], + [ + 377, + 70 + ], + [ + 407, + 40 + ], + [ + 344, + 44 + ], + [ + 406, + 14 + ], + [ + 343, + 17 + ] + ], + [ + [ + 359, + 24 + ], + [ + 361, + 0 + ], + [ + 341, + 16 + ], + [ + 378, + 34 + ], + [ + 340, + 45 + ], + [ + 376, + 65 + ] + ], + [ + [ + 373, + 49 + ], + [ + 397, + 46 + ], + [ + 358, + 66 + ], + [ + 389, + 30 + ], + [ + 333, + 70 + ], + [ + 362, + 32 + ] + ], + [ + [ + 347, + 24 + ], + [ + 377, + 18 + ], + [ + 346, + 52 + ], + [ + 349, + 0 + ], + [ + 316, + 59 + ], + [ + 316, + 0 + ] + ] + ] + }, + { + "task_id": "0000309", + "id": "RH20T_cfg1_task_0061_user_0014_scene_0008_cfg_0001_038522062547#331", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000309/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 332, + 110 + ], + [ + 312, + 96 + ], + [ + 328, + 109 + ], + [ + 335, + 111 + ], + [ + 349, + 123 + ], + [ + 354, + 124 + ] + ], + [ + [ + 347, + 89 + ], + [ + 350, + 75 + ], + [ + 355, + 111 + ], + [ + 339, + 67 + ], + [ + 352, + 121 + ], + [ + 337, + 80 + ] + ], + [ + [ + 373, + 76 + ], + [ + 357, + 56 + ], + [ + 394, + 59 + ], + [ + 353, + 93 + ], + [ + 409, + 79 + ], + [ + 368, + 112 + ] + ], + [ + [ + 344, + 103 + ], + [ + 368, + 98 + ], + [ + 343, + 115 + ], + [ + 345, + 90 + ], + [ + 320, + 120 + ], + [ + 320, + 95 + ] + ] + ] + }, + { + "task_id": "0000314", + "id": "RH20T_cfg6_task_0014_user_0014_scene_0008_cfg_0006_104122063633#121", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the object from the scale'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000314/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 235, + 200 + ], + [ + 219, + 200 + ], + [ + 246, + 181 + ], + [ + 224, + 218 + ], + [ + 261, + 182 + ], + [ + 240, + 217 + ] + ], + [ + [ + 194, + 184 + ], + [ + 206, + 168 + ], + [ + 176, + 168 + ], + [ + 211, + 200 + ], + [ + 162, + 186 + ], + [ + 198, + 217 + ] + ], + [ + [ + 200, + 194 + ], + [ + 179, + 203 + ], + [ + 187, + 188 + ], + [ + 213, + 201 + ], + [ + 206, + 180 + ], + [ + 233, + 192 + ] + ], + [ + [ + 213, + 195 + ], + [ + 211, + 174 + ], + [ + 216, + 181 + ], + [ + 211, + 209 + ], + [ + 217, + 201 + ], + [ + 213, + 230 + ] + ] + ] + }, + { + "task_id": "0000315", + "id": "RH20T_cfg7_task_0044_user_0014_scene_0005_cfg_0007_104122064161#78", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000315/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 215 + ], + [ + 345, + 215 + ], + [ + 322, + 204 + ], + [ + 372, + 225 + ], + [ + 325, + 204 + ], + [ + 372, + 224 + ] + ], + [ + [ + 365, + 227 + ], + [ + 366, + 205 + ], + [ + 370, + 246 + ], + [ + 360, + 210 + ], + [ + 369, + 266 + ], + [ + 359, + 229 + ] + ], + [ + [ + 342, + 202 + ], + [ + 366, + 188 + ], + [ + 339, + 193 + ], + [ + 346, + 213 + ], + [ + 317, + 207 + ], + [ + 321, + 228 + ] + ], + [ + [ + 348, + 218 + ], + [ + 374, + 218 + ], + [ + 346, + 243 + ], + [ + 349, + 193 + ], + [ + 320, + 243 + ], + [ + 323, + 192 + ] + ] + ] + }, + { + "task_id": "0000317", + "id": "RH20T_cfg6_task_0038_user_0020_scene_0001_cfg_0006_037522061512#506", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000317/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 248, + 144 + ], + [ + 243, + 131 + ], + [ + 265, + 154 + ], + [ + 231, + 133 + ], + [ + 271, + 168 + ], + [ + 236, + 147 + ] + ], + [ + [ + 205, + 156 + ], + [ + 220, + 175 + ], + [ + 216, + 149 + ], + [ + 195, + 162 + ], + [ + 200, + 129 + ], + [ + 181, + 143 + ] + ], + [ + [ + 215, + 142 + ], + [ + 229, + 125 + ], + [ + 222, + 157 + ], + [ + 208, + 126 + ], + [ + 209, + 174 + ], + [ + 193, + 144 + ] + ], + [ + [ + 203, + 151 + ], + [ + 195, + 160 + ], + [ + 205, + 130 + ], + [ + 202, + 172 + ], + [ + 213, + 122 + ], + [ + 210, + 163 + ] + ] + ] + }, + { + "task_id": "0000320", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_035622060973#261", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000320/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 222, + 92 + ], + [ + 221, + 79 + ], + [ + 203, + 89 + ], + [ + 241, + 95 + ], + [ + 204, + 102 + ], + [ + 241, + 107 + ] + ], + [ + [ + 210, + 99 + ], + [ + 193, + 108 + ], + [ + 219, + 111 + ], + [ + 199, + 86 + ], + [ + 236, + 103 + ], + [ + 216, + 78 + ] + ], + [ + [ + 222, + 94 + ], + [ + 219, + 104 + ], + [ + 240, + 101 + ], + [ + 204, + 87 + ], + [ + 242, + 92 + ], + [ + 207, + 78 + ] + ], + [ + [ + 219, + 73 + ], + [ + 215, + 61 + ], + [ + 201, + 81 + ], + [ + 237, + 64 + ], + [ + 207, + 93 + ], + [ + 241, + 76 + ] + ] + ] + }, + { + "task_id": "0000322", + "id": "RH20T_cfg7_task_0033_user_0020_scene_0008_cfg_0007_104122061602#195", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000322/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 501, + 198 + ], + [ + 498, + 172 + ], + [ + 494, + 199 + ], + [ + 508, + 197 + ], + [ + 497, + 226 + ], + [ + 511, + 222 + ] + ], + [ + [ + 522, + 218 + ], + [ + 543, + 234 + ], + [ + 515, + 238 + ], + [ + 530, + 196 + ], + [ + 496, + 223 + ], + [ + 509, + 182 + ] + ], + [ + [ + 534, + 212 + ], + [ + 517, + 218 + ], + [ + 534, + 187 + ], + [ + 534, + 236 + ], + [ + 551, + 182 + ], + [ + 549, + 229 + ] + ], + [ + [ + 512, + 244 + ], + [ + 528, + 241 + ], + [ + 536, + 245 + ], + [ + 487, + 244 + ], + [ + 521, + 248 + ], + [ + 474, + 247 + ] + ] + ] + }, + { + "task_id": "0000324", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_037522062165#382", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000324/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 213, + 209 + ], + [ + 238, + 217 + ], + [ + 201, + 225 + ], + [ + 225, + 191 + ], + [ + 175, + 217 + ], + [ + 198, + 181 + ] + ], + [ + [ + 242, + 226 + ], + [ + 242, + 209 + ], + [ + 239, + 204 + ], + [ + 245, + 250 + ], + [ + 239, + 221 + ], + [ + 244, + 265 + ] + ], + [ + [ + 226, + 241 + ], + [ + 234, + 263 + ], + [ + 201, + 240 + ], + [ + 252, + 241 + ], + [ + 192, + 218 + ], + [ + 244, + 217 + ] + ], + [ + [ + 254, + 225 + ], + [ + 236, + 206 + ], + [ + 248, + 221 + ], + [ + 258, + 229 + ], + [ + 267, + 241 + ], + [ + 275, + 248 + ] + ] + ] + }, + { + "task_id": "0000331", + "id": "RH20T_cfg4_task_0091_user_0014_scene_0007_cfg_0004_038522062288#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000331/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 265, + 97 + ], + [ + 259, + 79 + ], + [ + 238, + 102 + ], + [ + 291, + 92 + ], + [ + 244, + 118 + ], + [ + 295, + 109 + ] + ], + [ + [ + 249, + 100 + ], + [ + 276, + 97 + ], + [ + 237, + 108 + ], + [ + 260, + 93 + ], + [ + 210, + 111 + ], + [ + 235, + 96 + ] + ], + [ + [ + 241, + 106 + ], + [ + 260, + 99 + ], + [ + 221, + 91 + ], + [ + 261, + 122 + ], + [ + 199, + 99 + ], + [ + 241, + 131 + ] + ], + [ + [ + 288, + 81 + ], + [ + 311, + 78 + ], + [ + 276, + 88 + ], + [ + 298, + 74 + ], + [ + 252, + 90 + ], + [ + 276, + 76 + ] + ] + ] + }, + { + "task_id": "0000333", + "id": "RH20T_cfg5_task_0048_user_0010_scene_0010_cfg_0005_105422061350#213", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000333/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 210, + 21 + ], + [ + 178, + 4 + ], + [ + 198, + 49 + ], + [ + 223, + 0 + ], + [ + 227, + 64 + ], + [ + 254, + 9 + ] + ], + [ + [ + 230, + 5 + ], + [ + 236, + 35 + ], + [ + 222, + 0 + ], + [ + 236, + 16 + ], + [ + 216, + 0 + ], + [ + 231, + 0 + ] + ], + [ + [ + 239, + 14 + ], + [ + 228, + 46 + ], + [ + 258, + 15 + ], + [ + 221, + 14 + ], + [ + 270, + 0 + ], + [ + 232, + 0 + ] + ], + [ + [ + 244, + 23 + ], + [ + 254, + 3 + ], + [ + 255, + 1 + ], + [ + 233, + 47 + ], + [ + 246, + 20 + ], + [ + 225, + 65 + ] + ] + ] + }, + { + "task_id": "0000335", + "id": "RH20T_cfg5_task_0048_user_0010_scene_0010_cfg_0005_105422061350#463", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000335/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 121 + ], + [ + 270, + 94 + ], + [ + 284, + 131 + ], + [ + 284, + 109 + ], + [ + 297, + 156 + ], + [ + 299, + 138 + ] + ], + [ + [ + 298, + 139 + ], + [ + 269, + 127 + ], + [ + 305, + 119 + ], + [ + 292, + 157 + ], + [ + 335, + 131 + ], + [ + 319, + 168 + ] + ], + [ + [ + 268, + 109 + ], + [ + 246, + 120 + ], + [ + 293, + 123 + ], + [ + 241, + 94 + ], + [ + 317, + 113 + ], + [ + 264, + 80 + ] + ], + [ + [ + 268, + 118 + ], + [ + 282, + 93 + ], + [ + 291, + 116 + ], + [ + 243, + 120 + ], + [ + 277, + 137 + ], + [ + 230, + 143 + ] + ] + ] + }, + { + "task_id": "0000336", + "id": "RH20T_cfg1_task_0081_user_0010_scene_0009_cfg_0001_819112070044#90", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000336/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 202 + ], + [ + 305, + 190 + ], + [ + 329, + 198 + ], + [ + 324, + 206 + ], + [ + 352, + 210 + ], + [ + 344, + 217 + ] + ], + [ + [ + 331, + 211 + ], + [ + 353, + 207 + ], + [ + 326, + 201 + ], + [ + 335, + 223 + ], + [ + 304, + 205 + ], + [ + 311, + 227 + ] + ], + [ + [ + 323, + 208 + ], + [ + 307, + 226 + ], + [ + 322, + 200 + ], + [ + 325, + 216 + ], + [ + 337, + 182 + ], + [ + 342, + 197 + ] + ], + [ + [ + 336, + 212 + ], + [ + 338, + 193 + ], + [ + 334, + 195 + ], + [ + 338, + 229 + ], + [ + 332, + 213 + ], + [ + 335, + 247 + ] + ] + ] + }, + { + "task_id": "0000341", + "id": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_039422060546#212", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000341/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 119 + ], + [ + 325, + 148 + ], + [ + 366, + 127 + ], + [ + 303, + 110 + ], + [ + 376, + 97 + ], + [ + 312, + 78 + ] + ], + [ + [ + 304, + 97 + ], + [ + 293, + 76 + ], + [ + 307, + 122 + ], + [ + 302, + 73 + ], + [ + 317, + 140 + ], + [ + 311, + 92 + ] + ], + [ + [ + 326, + 107 + ], + [ + 323, + 135 + ], + [ + 298, + 97 + ], + [ + 353, + 116 + ], + [ + 302, + 69 + ], + [ + 356, + 90 + ] + ], + [ + [ + 312, + 98 + ], + [ + 321, + 122 + ], + [ + 340, + 81 + ], + [ + 285, + 115 + ], + [ + 331, + 54 + ], + [ + 273, + 88 + ] + ] + ] + }, + { + "task_id": "0000342", + "id": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_039422060546#497", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000342/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 168 + ], + [ + 288, + 156 + ], + [ + 291, + 181 + ], + [ + 328, + 153 + ], + [ + 311, + 191 + ], + [ + 348, + 165 + ] + ], + [ + [ + 306, + 159 + ], + [ + 317, + 148 + ], + [ + 280, + 146 + ], + [ + 331, + 173 + ], + [ + 272, + 157 + ], + [ + 320, + 183 + ] + ], + [ + [ + 330, + 91 + ], + [ + 319, + 119 + ], + [ + 344, + 101 + ], + [ + 314, + 80 + ], + [ + 355, + 75 + ], + [ + 326, + 50 + ] + ], + [ + [ + 320, + 126 + ], + [ + 316, + 112 + ], + [ + 330, + 150 + ], + [ + 311, + 102 + ], + [ + 333, + 161 + ], + [ + 314, + 116 + ] + ] + ] + }, + { + "task_id": "0000345", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0008_cfg_0006_104122060811#233", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000345/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 360, + 182 + ], + [ + 352, + 179 + ], + [ + 346, + 186 + ], + [ + 373, + 178 + ], + [ + 355, + 189 + ], + [ + 382, + 181 + ] + ], + [ + [ + 337, + 177 + ], + [ + 344, + 187 + ], + [ + 350, + 169 + ], + [ + 325, + 185 + ], + [ + 343, + 159 + ], + [ + 317, + 176 + ] + ], + [ + [ + 354, + 178 + ], + [ + 362, + 170 + ], + [ + 365, + 177 + ], + [ + 342, + 179 + ], + [ + 357, + 185 + ], + [ + 334, + 187 + ] + ], + [ + [ + 365, + 190 + ], + [ + 374, + 180 + ], + [ + 364, + 179 + ], + [ + 366, + 201 + ], + [ + 356, + 188 + ], + [ + 357, + 210 + ] + ] + ] + }, + { + "task_id": "0000346", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0008_cfg_0006_104122060811#390", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000346/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 354, + 145 + ], + [ + 350, + 134 + ], + [ + 349, + 136 + ], + [ + 359, + 152 + ], + [ + 353, + 148 + ], + [ + 363, + 164 + ] + ], + [ + [ + 339, + 146 + ], + [ + 347, + 157 + ], + [ + 327, + 149 + ], + [ + 351, + 143 + ], + [ + 320, + 138 + ], + [ + 344, + 132 + ] + ], + [ + [ + 334, + 147 + ], + [ + 334, + 161 + ], + [ + 328, + 149 + ], + [ + 342, + 144 + ], + [ + 328, + 135 + ], + [ + 342, + 129 + ] + ], + [ + [ + 349, + 149 + ], + [ + 353, + 142 + ], + [ + 362, + 146 + ], + [ + 335, + 152 + ], + [ + 358, + 153 + ], + [ + 331, + 159 + ] + ] + ] + }, + { + "task_id": "0000347", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0008_cfg_0006_104122060811#662", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000347/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 396, + 134 + ], + [ + 406, + 126 + ], + [ + 407, + 134 + ], + [ + 386, + 135 + ], + [ + 398, + 142 + ], + [ + 377, + 143 + ] + ], + [ + [ + 395, + 141 + ], + [ + 393, + 145 + ], + [ + 384, + 133 + ], + [ + 407, + 150 + ], + [ + 386, + 129 + ], + [ + 409, + 146 + ] + ], + [ + [ + 397, + 141 + ], + [ + 407, + 132 + ], + [ + 389, + 142 + ], + [ + 405, + 139 + ], + [ + 379, + 151 + ], + [ + 396, + 147 + ] + ], + [ + [ + 407, + 125 + ], + [ + 409, + 138 + ], + [ + 419, + 120 + ], + [ + 395, + 129 + ], + [ + 417, + 106 + ], + [ + 393, + 116 + ] + ] + ] + }, + { + "task_id": "0000348", + "id": "RH20T_cfg4_task_0006_user_0010_scene_0008_cfg_0004_104122062823#157", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000348/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 524, + 154 + ], + [ + 510, + 165 + ], + [ + 497, + 173 + ], + [ + 554, + 132 + ], + [ + 510, + 162 + ], + [ + 565, + 122 + ] + ], + [ + [ + 466, + 129 + ], + [ + 471, + 107 + ], + [ + 471, + 152 + ], + [ + 460, + 104 + ], + [ + 466, + 176 + ], + [ + 454, + 128 + ] + ], + [ + [ + 480, + 147 + ], + [ + 513, + 135 + ], + [ + 468, + 115 + ], + [ + 492, + 179 + ], + [ + 436, + 127 + ], + [ + 459, + 190 + ] + ], + [ + [ + 496, + 122 + ], + [ + 505, + 93 + ], + [ + 532, + 132 + ], + [ + 461, + 114 + ], + [ + 522, + 159 + ], + [ + 454, + 140 + ] + ] + ] + }, + { + "task_id": "0000349", + "id": "RH20T_cfg4_task_0006_user_0010_scene_0008_cfg_0004_104122062823#391", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000349/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 83 + ], + [ + 405, + 73 + ], + [ + 391, + 49 + ], + [ + 394, + 116 + ], + [ + 376, + 59 + ], + [ + 381, + 129 + ] + ], + [ + [ + 383, + 84 + ], + [ + 378, + 106 + ], + [ + 354, + 74 + ], + [ + 411, + 95 + ], + [ + 359, + 54 + ], + [ + 414, + 74 + ] + ], + [ + [ + 369, + 88 + ], + [ + 376, + 61 + ], + [ + 398, + 100 + ], + [ + 341, + 77 + ], + [ + 391, + 125 + ], + [ + 337, + 102 + ] + ], + [ + [ + 389, + 86 + ], + [ + 380, + 62 + ], + [ + 422, + 76 + ], + [ + 356, + 96 + ], + [ + 433, + 102 + ], + [ + 365, + 122 + ] + ] + ] + }, + { + "task_id": "0000350", + "id": "RH20T_cfg1_task_0030_user_0015_scene_0001_cfg_0001_038522062547#248", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour out the test tube'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000350/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 440, + 31 + ], + [ + 464, + 36 + ], + [ + 444, + 9 + ], + [ + 435, + 53 + ], + [ + 420, + 4 + ], + [ + 411, + 47 + ] + ], + [ + [ + 405, + 44 + ], + [ + 389, + 49 + ], + [ + 407, + 18 + ], + [ + 403, + 70 + ], + [ + 422, + 14 + ], + [ + 417, + 64 + ] + ], + [ + [ + 438, + 50 + ], + [ + 425, + 74 + ], + [ + 427, + 42 + ], + [ + 450, + 60 + ], + [ + 439, + 18 + ], + [ + 464, + 34 + ] + ], + [ + [ + 420, + 34 + ], + [ + 431, + 31 + ], + [ + 438, + 50 + ], + [ + 401, + 17 + ], + [ + 427, + 52 + ], + [ + 392, + 20 + ] + ] + ] + }, + { + "task_id": "0000354", + "id": "RH20T_cfg5_task_0015_user_0007_scene_0006_cfg_0005_104122063678#263", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000354/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 292, + 224 + ], + [ + 270, + 205 + ], + [ + 310, + 204 + ], + [ + 273, + 245 + ], + [ + 331, + 223 + ], + [ + 296, + 265 + ] + ], + [ + [ + 286, + 227 + ], + [ + 274, + 210 + ], + [ + 278, + 252 + ], + [ + 295, + 201 + ], + [ + 291, + 271 + ], + [ + 309, + 218 + ] + ], + [ + [ + 264, + 224 + ], + [ + 285, + 243 + ], + [ + 256, + 221 + ], + [ + 271, + 227 + ], + [ + 235, + 202 + ], + [ + 252, + 210 + ] + ], + [ + [ + 272, + 214 + ], + [ + 280, + 196 + ], + [ + 251, + 229 + ], + [ + 291, + 201 + ], + [ + 245, + 246 + ], + [ + 283, + 218 + ] + ] + ] + }, + { + "task_id": "0000356", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0001_cfg_0004_104122062295#194", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000356/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 133, + 80 + ], + [ + 133, + 41 + ], + [ + 114, + 63 + ], + [ + 153, + 98 + ], + [ + 114, + 98 + ], + [ + 152, + 134 + ] + ], + [ + [ + 123, + 79 + ], + [ + 153, + 69 + ], + [ + 91, + 75 + ], + [ + 158, + 82 + ], + [ + 55, + 86 + ], + [ + 125, + 95 + ] + ], + [ + [ + 143, + 77 + ], + [ + 182, + 67 + ], + [ + 128, + 36 + ], + [ + 157, + 116 + ], + [ + 88, + 46 + ], + [ + 120, + 125 + ] + ], + [ + [ + 115, + 92 + ], + [ + 144, + 89 + ], + [ + 87, + 106 + ], + [ + 146, + 77 + ], + [ + 52, + 111 + ], + [ + 114, + 80 + ] + ] + ] + }, + { + "task_id": "0000357", + "id": "RH20T_cfg1_task_0066_user_0015_scene_0006_cfg_0001_038522062547#167", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000357/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 440, + 67 + ], + [ + 419, + 79 + ], + [ + 448, + 72 + ], + [ + 432, + 61 + ], + [ + 470, + 60 + ], + [ + 452, + 49 + ] + ], + [ + [ + 449, + 42 + ], + [ + 456, + 27 + ], + [ + 453, + 62 + ], + [ + 446, + 22 + ], + [ + 447, + 75 + ], + [ + 441, + 36 + ] + ], + [ + [ + 441, + 41 + ], + [ + 463, + 31 + ], + [ + 442, + 39 + ], + [ + 441, + 44 + ], + [ + 420, + 50 + ], + [ + 421, + 54 + ] + ], + [ + [ + 429, + 49 + ], + [ + 413, + 45 + ], + [ + 436, + 25 + ], + [ + 422, + 72 + ], + [ + 454, + 28 + ], + [ + 439, + 77 + ] + ] + ] + }, + { + "task_id": "0000358", + "id": "RH20T_cfg1_task_0066_user_0015_scene_0006_cfg_0001_038522062547#457", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000358/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 426, + 20 + ], + [ + 402, + 26 + ], + [ + 418, + 12 + ], + [ + 432, + 27 + ], + [ + 444, + 5 + ], + [ + 456, + 22 + ] + ], + [ + [ + 438, + 18 + ], + [ + 444, + 1 + ], + [ + 442, + 37 + ], + [ + 434, + 0 + ], + [ + 437, + 51 + ], + [ + 429, + 15 + ] + ], + [ + [ + 432, + 18 + ], + [ + 453, + 11 + ], + [ + 424, + 27 + ], + [ + 441, + 9 + ], + [ + 404, + 33 + ], + [ + 419, + 17 + ] + ], + [ + [ + 447, + 3 + ], + [ + 468, + 1 + ], + [ + 434, + 3 + ], + [ + 461, + 2 + ], + [ + 413, + 5 + ], + [ + 439, + 4 + ] + ] + ] + }, + { + "task_id": "0000359", + "id": "RH20T_cfg1_task_0066_user_0015_scene_0006_cfg_0001_038522062547#726", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000359/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 390, + 50 + ], + [ + 389, + 37 + ], + [ + 415, + 43 + ], + [ + 366, + 57 + ], + [ + 414, + 56 + ], + [ + 368, + 69 + ] + ], + [ + [ + 394, + 45 + ], + [ + 387, + 50 + ], + [ + 395, + 67 + ], + [ + 393, + 22 + ], + [ + 401, + 61 + ], + [ + 399, + 18 + ] + ], + [ + [ + 405, + 47 + ], + [ + 389, + 64 + ], + [ + 416, + 46 + ], + [ + 393, + 48 + ], + [ + 432, + 29 + ], + [ + 409, + 29 + ] + ], + [ + [ + 384, + 47 + ], + [ + 387, + 32 + ], + [ + 388, + 67 + ], + [ + 380, + 27 + ], + [ + 385, + 79 + ], + [ + 378, + 41 + ] + ] + ] + }, + { + "task_id": "0000361", + "id": "RH20T_cfg1_task_0066_user_0015_scene_0006_cfg_0001_038522062547#1314", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000361/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 297, + 29 + ], + [ + 283, + 20 + ], + [ + 307, + 43 + ], + [ + 285, + 15 + ], + [ + 320, + 52 + ], + [ + 299, + 24 + ] + ], + [ + [ + 336, + 30 + ], + [ + 316, + 40 + ], + [ + 347, + 52 + ], + [ + 324, + 9 + ], + [ + 365, + 42 + ], + [ + 343, + 0 + ] + ], + [ + [ + 300, + 54 + ], + [ + 283, + 44 + ], + [ + 289, + 75 + ], + [ + 311, + 34 + ], + [ + 305, + 84 + ], + [ + 327, + 45 + ] + ], + [ + [ + 318, + 49 + ], + [ + 318, + 34 + ], + [ + 321, + 69 + ], + [ + 315, + 29 + ], + [ + 320, + 81 + ], + [ + 314, + 43 + ] + ] + ] + }, + { + "task_id": "0000365", + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_038522062547#71", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000365/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 384, + 193 + ], + [ + 351, + 199 + ], + [ + 383, + 169 + ], + [ + 385, + 215 + ], + [ + 419, + 162 + ], + [ + 417, + 209 + ] + ], + [ + [ + 401, + 177 + ], + [ + 429, + 160 + ], + [ + 384, + 154 + ], + [ + 419, + 202 + ], + [ + 357, + 171 + ], + [ + 390, + 219 + ] + ], + [ + [ + 380, + 203 + ], + [ + 389, + 191 + ], + [ + 398, + 231 + ], + [ + 363, + 177 + ], + [ + 389, + 240 + ], + [ + 356, + 189 + ] + ], + [ + [ + 371, + 180 + ], + [ + 345, + 197 + ], + [ + 379, + 184 + ], + [ + 364, + 177 + ], + [ + 407, + 166 + ], + [ + 389, + 161 + ] + ] + ] + }, + { + "task_id": "0000369", + "id": "RH20T_cfg7_task_0047_user_0020_scene_0004_cfg_0007_037522061512#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000369/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 201, + 88 + ], + [ + 222, + 99 + ], + [ + 186, + 108 + ], + [ + 215, + 69 + ], + [ + 165, + 97 + ], + [ + 194, + 59 + ] + ], + [ + [ + 210, + 100 + ], + [ + 225, + 82 + ], + [ + 211, + 110 + ], + [ + 210, + 90 + ], + [ + 197, + 127 + ], + [ + 194, + 110 + ] + ], + [ + [ + 190, + 91 + ], + [ + 168, + 77 + ], + [ + 202, + 69 + ], + [ + 177, + 113 + ], + [ + 222, + 83 + ], + [ + 198, + 125 + ] + ], + [ + [ + 204, + 96 + ], + [ + 208, + 108 + ], + [ + 228, + 102 + ], + [ + 180, + 89 + ], + [ + 224, + 89 + ], + [ + 174, + 75 + ] + ] + ] + }, + { + "task_id": "0000370", + "id": "RH20T_cfg1_task_0077_user_0010_scene_0006_cfg_0001_038522063145#181", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000370/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 439, + 134 + ], + [ + 454, + 126 + ], + [ + 444, + 120 + ], + [ + 434, + 146 + ], + [ + 427, + 129 + ], + [ + 419, + 154 + ] + ], + [ + [ + 461, + 125 + ], + [ + 467, + 108 + ], + [ + 479, + 135 + ], + [ + 444, + 115 + ], + [ + 473, + 150 + ], + [ + 439, + 130 + ] + ], + [ + [ + 447, + 111 + ], + [ + 442, + 103 + ], + [ + 442, + 131 + ], + [ + 453, + 91 + ], + [ + 447, + 138 + ], + [ + 458, + 99 + ] + ], + [ + [ + 457, + 98 + ], + [ + 456, + 115 + ], + [ + 462, + 84 + ], + [ + 453, + 111 + ], + [ + 463, + 68 + ], + [ + 454, + 95 + ] + ] + ] + }, + { + "task_id": "0000371", + "id": "RH20T_cfg1_task_0016_user_0014_scene_0008_cfg_0001_035622060973#223", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000371/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 240, + 185 + ], + [ + 254, + 197 + ], + [ + 246, + 192 + ], + [ + 234, + 177 + ], + [ + 234, + 180 + ], + [ + 220, + 164 + ] + ], + [ + [ + 232, + 167 + ], + [ + 232, + 149 + ], + [ + 251, + 172 + ], + [ + 214, + 162 + ], + [ + 251, + 190 + ], + [ + 215, + 180 + ] + ], + [ + [ + 225, + 187 + ], + [ + 227, + 205 + ], + [ + 245, + 183 + ], + [ + 205, + 191 + ], + [ + 243, + 164 + ], + [ + 202, + 172 + ] + ], + [ + [ + 220, + 177 + ], + [ + 217, + 170 + ], + [ + 202, + 188 + ], + [ + 238, + 166 + ], + [ + 205, + 194 + ], + [ + 240, + 174 + ] + ] + ] + }, + { + "task_id": "0000372", + "id": "RH20T_cfg5_task_0004_user_0014_scene_0007_cfg_0005_037522062165#212", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000372/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 173, + 269 + ], + [ + 170, + 254 + ], + [ + 146, + 257 + ], + [ + 200, + 281 + ], + [ + 149, + 271 + ], + [ + 200, + 293 + ] + ], + [ + [ + 165, + 292 + ], + [ + 182, + 318 + ], + [ + 190, + 272 + ], + [ + 139, + 312 + ], + [ + 173, + 248 + ], + [ + 123, + 287 + ] + ], + [ + [ + 171, + 291 + ], + [ + 200, + 284 + ], + [ + 166, + 263 + ], + [ + 176, + 321 + ], + [ + 135, + 269 + ], + [ + 146, + 329 + ] + ], + [ + [ + 189, + 276 + ], + [ + 172, + 265 + ], + [ + 196, + 300 + ], + [ + 181, + 251 + ], + [ + 214, + 313 + ], + [ + 199, + 263 + ] + ] + ] + }, + { + "task_id": "0000374", + "id": "RH20T_cfg2_task_0128_user_0007_scene_0003_cfg_0002_036422060215#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the USB drive to the docking station'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000374/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 224, + 154 + ], + [ + 226, + 138 + ], + [ + 244, + 150 + ], + [ + 204, + 159 + ], + [ + 241, + 165 + ], + [ + 202, + 174 + ] + ], + [ + [ + 220, + 174 + ], + [ + 197, + 173 + ], + [ + 222, + 153 + ], + [ + 218, + 195 + ], + [ + 244, + 154 + ], + [ + 240, + 196 + ] + ], + [ + [ + 230, + 139 + ], + [ + 223, + 128 + ], + [ + 231, + 121 + ], + [ + 228, + 159 + ], + [ + 237, + 132 + ], + [ + 234, + 168 + ] + ], + [ + [ + 220, + 165 + ], + [ + 211, + 153 + ], + [ + 220, + 148 + ], + [ + 219, + 181 + ], + [ + 229, + 160 + ], + [ + 228, + 193 + ] + ] + ] + }, + { + "task_id": "0000375", + "id": "RH20T_cfg4_task_0032_user_0016_scene_0002_cfg_0004_104122063550#54", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000375/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 288, + 136 + ], + [ + 288, + 159 + ], + [ + 311, + 140 + ], + [ + 264, + 131 + ], + [ + 310, + 118 + ], + [ + 264, + 108 + ] + ], + [ + [ + 275, + 161 + ], + [ + 288, + 138 + ], + [ + 271, + 161 + ], + [ + 279, + 161 + ], + [ + 258, + 183 + ], + [ + 265, + 186 + ] + ], + [ + [ + 292, + 147 + ], + [ + 286, + 134 + ], + [ + 312, + 162 + ], + [ + 272, + 133 + ], + [ + 316, + 173 + ], + [ + 278, + 145 + ] + ], + [ + [ + 281, + 124 + ], + [ + 300, + 142 + ], + [ + 262, + 142 + ], + [ + 302, + 106 + ], + [ + 242, + 124 + ], + [ + 282, + 87 + ] + ] + ] + }, + { + "task_id": "0000379", + "id": "RH20T_cfg1_task_0070_user_0014_scene_0008_cfg_0001_750612070851#156", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000379/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 293, + 271 + ], + [ + 255, + 271 + ], + [ + 291, + 304 + ], + [ + 294, + 243 + ], + [ + 332, + 303 + ], + [ + 330, + 243 + ] + ], + [ + [ + 328, + 287 + ], + [ + 298, + 314 + ], + [ + 320, + 268 + ], + [ + 334, + 304 + ], + [ + 350, + 241 + ], + [ + 361, + 280 + ] + ], + [ + [ + 314, + 298 + ], + [ + 319, + 282 + ], + [ + 277, + 302 + ], + [ + 351, + 294 + ], + [ + 275, + 316 + ], + [ + 345, + 308 + ] + ], + [ + [ + 345, + 335 + ], + [ + 357, + 333 + ], + [ + 379, + 330 + ], + [ + 309, + 341 + ], + [ + 366, + 332 + ], + [ + 300, + 342 + ] + ] + ] + }, + { + "task_id": "0000381", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_036422060909#124", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000381/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 307, + 149 + ], + [ + 330, + 150 + ], + [ + 309, + 173 + ], + [ + 305, + 126 + ], + [ + 285, + 172 + ], + [ + 282, + 125 + ] + ], + [ + [ + 309, + 110 + ], + [ + 282, + 112 + ], + [ + 312, + 136 + ], + [ + 306, + 84 + ], + [ + 338, + 134 + ], + [ + 331, + 83 + ] + ], + [ + [ + 323, + 123 + ], + [ + 327, + 106 + ], + [ + 318, + 105 + ], + [ + 327, + 143 + ], + [ + 314, + 121 + ], + [ + 322, + 159 + ] + ], + [ + [ + 305, + 108 + ], + [ + 284, + 108 + ], + [ + 319, + 111 + ], + [ + 292, + 105 + ], + [ + 340, + 111 + ], + [ + 312, + 105 + ] + ] + ] + }, + { + "task_id": "0000383", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_036422060909#299", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000383/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 125 + ], + [ + 339, + 109 + ], + [ + 340, + 108 + ], + [ + 327, + 143 + ], + [ + 335, + 123 + ], + [ + 323, + 158 + ] + ], + [ + [ + 323, + 114 + ], + [ + 304, + 109 + ], + [ + 318, + 138 + ], + [ + 327, + 90 + ], + [ + 337, + 144 + ], + [ + 347, + 95 + ] + ], + [ + [ + 313, + 121 + ], + [ + 300, + 103 + ], + [ + 329, + 106 + ], + [ + 296, + 137 + ], + [ + 341, + 123 + ], + [ + 309, + 153 + ] + ], + [ + [ + 349, + 156 + ], + [ + 324, + 156 + ], + [ + 348, + 133 + ], + [ + 350, + 180 + ], + [ + 372, + 134 + ], + [ + 375, + 181 + ] + ] + ] + }, + { + "task_id": "0000385", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_036422060909#342", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000385/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 332, + 163 + ], + [ + 308, + 156 + ], + [ + 329, + 163 + ], + [ + 334, + 163 + ], + [ + 354, + 171 + ], + [ + 356, + 170 + ] + ], + [ + [ + 329, + 141 + ], + [ + 336, + 125 + ], + [ + 334, + 122 + ], + [ + 324, + 160 + ], + [ + 328, + 137 + ], + [ + 319, + 175 + ] + ], + [ + [ + 350, + 155 + ], + [ + 359, + 145 + ], + [ + 349, + 132 + ], + [ + 352, + 178 + ], + [ + 339, + 142 + ], + [ + 343, + 189 + ] + ], + [ + [ + 309, + 130 + ], + [ + 317, + 155 + ], + [ + 324, + 130 + ], + [ + 293, + 129 + ], + [ + 316, + 107 + ], + [ + 285, + 104 + ] + ] + ] + }, + { + "task_id": "0000386", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_036422060909#404", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000386/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 110 + ], + [ + 343, + 95 + ], + [ + 343, + 131 + ], + [ + 309, + 89 + ], + [ + 327, + 144 + ], + [ + 293, + 103 + ] + ], + [ + [ + 316, + 108 + ], + [ + 298, + 115 + ], + [ + 299, + 102 + ], + [ + 331, + 115 + ], + [ + 318, + 94 + ], + [ + 350, + 109 + ] + ], + [ + [ + 337, + 103 + ], + [ + 339, + 89 + ], + [ + 332, + 81 + ], + [ + 343, + 125 + ], + [ + 331, + 94 + ], + [ + 341, + 137 + ] + ], + [ + [ + 344, + 77 + ], + [ + 354, + 55 + ], + [ + 341, + 90 + ], + [ + 347, + 65 + ], + [ + 332, + 110 + ], + [ + 338, + 85 + ] + ] + ] + }, + { + "task_id": "0000388", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_036422060909#544", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000388/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 115 + ], + [ + 352, + 100 + ], + [ + 350, + 96 + ], + [ + 344, + 135 + ], + [ + 346, + 110 + ], + [ + 340, + 148 + ] + ], + [ + [ + 337, + 86 + ], + [ + 327, + 87 + ], + [ + 335, + 112 + ], + [ + 340, + 60 + ], + [ + 344, + 110 + ], + [ + 348, + 61 + ] + ], + [ + [ + 366, + 109 + ], + [ + 379, + 100 + ], + [ + 351, + 89 + ], + [ + 382, + 129 + ], + [ + 340, + 99 + ], + [ + 370, + 136 + ] + ], + [ + [ + 336, + 110 + ], + [ + 345, + 128 + ], + [ + 320, + 123 + ], + [ + 351, + 96 + ], + [ + 310, + 104 + ], + [ + 342, + 77 + ] + ] + ] + }, + { + "task_id": "0000389", + "id": "RH20T_cfg2_task_0054_user_0007_scene_0009_cfg_0002_104122061850#27", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the moving object'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000389/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 395, + 157 + ], + [ + 381, + 140 + ], + [ + 380, + 174 + ], + [ + 412, + 139 + ], + [ + 394, + 191 + ], + [ + 427, + 156 + ] + ], + [ + [ + 394, + 173 + ], + [ + 387, + 197 + ], + [ + 392, + 165 + ], + [ + 396, + 180 + ], + [ + 399, + 140 + ], + [ + 402, + 157 + ] + ], + [ + [ + 402, + 166 + ], + [ + 422, + 160 + ], + [ + 401, + 186 + ], + [ + 403, + 144 + ], + [ + 381, + 192 + ], + [ + 383, + 150 + ] + ], + [ + [ + 401, + 158 + ], + [ + 397, + 143 + ], + [ + 383, + 173 + ], + [ + 420, + 142 + ], + [ + 387, + 186 + ], + [ + 423, + 156 + ] + ] + ] + }, + { + "task_id": "0000390", + "id": "RH20T_cfg2_task_0054_user_0007_scene_0009_cfg_0002_104122061850#86", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the moving object'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000390/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 423, + 178 + ], + [ + 418, + 164 + ], + [ + 399, + 188 + ], + [ + 448, + 167 + ], + [ + 405, + 201 + ], + [ + 452, + 181 + ] + ], + [ + [ + 422, + 176 + ], + [ + 449, + 170 + ], + [ + 426, + 200 + ], + [ + 418, + 153 + ], + [ + 399, + 205 + ], + [ + 392, + 158 + ] + ], + [ + [ + 422, + 191 + ], + [ + 421, + 217 + ], + [ + 395, + 191 + ], + [ + 451, + 191 + ], + [ + 398, + 166 + ], + [ + 452, + 164 + ] + ], + [ + [ + 441, + 166 + ], + [ + 424, + 148 + ], + [ + 422, + 185 + ], + [ + 461, + 147 + ], + [ + 438, + 205 + ], + [ + 479, + 167 + ] + ] + ] + }, + { + "task_id": "0000391", + "id": "RH20T_cfg4_task_0047_user_0014_scene_0005_cfg_0004_104122062823#125", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000391/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 251, + 8 + ], + [ + 271, + 0 + ], + [ + 227, + 18 + ], + [ + 272, + 0 + ], + [ + 210, + 32 + ], + [ + 254, + 13 + ] + ], + [ + [ + 276, + 6 + ], + [ + 266, + 0 + ], + [ + 291, + 0 + ], + [ + 259, + 23 + ], + [ + 299, + 17 + ], + [ + 269, + 51 + ] + ], + [ + [ + 237, + 3 + ], + [ + 233, + 35 + ], + [ + 248, + 13 + ], + [ + 228, + 0 + ], + [ + 253, + 0 + ], + [ + 232, + 0 + ] + ], + [ + [ + 242, + 9 + ], + [ + 275, + 4 + ], + [ + 242, + 32 + ], + [ + 242, + 0 + ], + [ + 209, + 38 + ], + [ + 210, + 0 + ] + ] + ] + }, + { + "task_id": "0000392", + "id": "RH20T_cfg4_task_0047_user_0014_scene_0005_cfg_0004_104122062823#274", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000392/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 211, + 13 + ], + [ + 206, + 0 + ], + [ + 243, + 8 + ], + [ + 179, + 18 + ], + [ + 246, + 36 + ], + [ + 185, + 46 + ] + ], + [ + [ + 233, + 0 + ], + [ + 216, + 0 + ], + [ + 206, + 14 + ], + [ + 259, + 0 + ], + [ + 223, + 32 + ], + [ + 278, + 0 + ] + ], + [ + [ + 189, + 0 + ], + [ + 173, + 0 + ], + [ + 215, + 0 + ], + [ + 162, + 0 + ], + [ + 229, + 0 + ], + [ + 178, + 18 + ] + ], + [ + [ + 207, + 0 + ], + [ + 209, + 0 + ], + [ + 173, + 0 + ], + [ + 241, + 0 + ], + [ + 170, + 0 + ], + [ + 241, + 0 + ] + ] + ] + }, + { + "task_id": "0000396", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_104122061850#480", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000396/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 402, + 212 + ], + [ + 380, + 218 + ], + [ + 414, + 200 + ], + [ + 389, + 225 + ], + [ + 438, + 193 + ], + [ + 414, + 219 + ] + ], + [ + [ + 391, + 187 + ], + [ + 386, + 172 + ], + [ + 378, + 210 + ], + [ + 403, + 165 + ], + [ + 383, + 223 + ], + [ + 406, + 179 + ] + ], + [ + [ + 381, + 168 + ], + [ + 387, + 181 + ], + [ + 377, + 144 + ], + [ + 385, + 191 + ], + [ + 372, + 134 + ], + [ + 380, + 178 + ] + ], + [ + [ + 380, + 154 + ], + [ + 384, + 132 + ], + [ + 363, + 138 + ], + [ + 395, + 169 + ], + [ + 357, + 161 + ], + [ + 391, + 192 + ] + ] + ] + }, + { + "task_id": "0000400", + "id": "RH20T_cfg4_task_0066_user_0007_scene_0007_cfg_0004_104122062823#168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000400/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 509, + 198 + ], + [ + 515, + 172 + ], + [ + 506, + 169 + ], + [ + 512, + 224 + ], + [ + 499, + 197 + ], + [ + 506, + 252 + ] + ], + [ + [ + 487, + 154 + ], + [ + 498, + 125 + ], + [ + 527, + 166 + ], + [ + 449, + 143 + ], + [ + 514, + 192 + ], + [ + 441, + 169 + ] + ], + [ + [ + 490, + 150 + ], + [ + 460, + 172 + ], + [ + 489, + 122 + ], + [ + 490, + 175 + ], + [ + 519, + 100 + ], + [ + 517, + 155 + ] + ], + [ + [ + 467, + 114 + ], + [ + 499, + 98 + ], + [ + 473, + 86 + ], + [ + 462, + 138 + ], + [ + 438, + 104 + ], + [ + 431, + 154 + ] + ] + ] + }, + { + "task_id": "0000402", + "id": "RH20T_cfg4_task_0066_user_0007_scene_0007_cfg_0004_104122062823#803", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000402/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 494, + 82 + ], + [ + 514, + 103 + ], + [ + 508, + 87 + ], + [ + 482, + 77 + ], + [ + 486, + 64 + ], + [ + 463, + 57 + ] + ], + [ + [ + 460, + 77 + ], + [ + 465, + 66 + ], + [ + 464, + 47 + ], + [ + 456, + 104 + ], + [ + 458, + 57 + ], + [ + 451, + 117 + ] + ], + [ + [ + 478, + 75 + ], + [ + 482, + 46 + ], + [ + 513, + 79 + ], + [ + 445, + 72 + ], + [ + 508, + 107 + ], + [ + 443, + 99 + ] + ], + [ + [ + 491, + 94 + ], + [ + 479, + 65 + ], + [ + 473, + 96 + ], + [ + 512, + 92 + ], + [ + 484, + 122 + ], + [ + 524, + 121 + ] + ] + ] + }, + { + "task_id": "0000403", + "id": "RH20T_cfg4_task_0066_user_0007_scene_0007_cfg_0004_104122062823#980", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000403/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 458, + 61 + ], + [ + 452, + 74 + ], + [ + 429, + 60 + ], + [ + 490, + 62 + ], + [ + 436, + 49 + ], + [ + 494, + 50 + ] + ], + [ + [ + 465, + 66 + ], + [ + 471, + 38 + ], + [ + 497, + 72 + ], + [ + 435, + 60 + ], + [ + 490, + 97 + ], + [ + 431, + 85 + ] + ], + [ + [ + 478, + 59 + ], + [ + 480, + 53 + ], + [ + 465, + 31 + ], + [ + 490, + 87 + ], + [ + 463, + 37 + ], + [ + 487, + 90 + ] + ], + [ + [ + 443, + 95 + ], + [ + 456, + 90 + ], + [ + 424, + 72 + ], + [ + 463, + 117 + ], + [ + 414, + 77 + ], + [ + 450, + 119 + ] + ] + ] + }, + { + "task_id": "0000405", + "id": "RH20T_cfg4_task_0066_user_0007_scene_0007_cfg_0004_104122062823#1312", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000405/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 387, + 44 + ], + [ + 390, + 18 + ], + [ + 414, + 53 + ], + [ + 361, + 36 + ], + [ + 410, + 76 + ], + [ + 360, + 59 + ] + ], + [ + [ + 387, + 55 + ], + [ + 398, + 38 + ], + [ + 406, + 74 + ], + [ + 368, + 36 + ], + [ + 395, + 87 + ], + [ + 359, + 51 + ] + ], + [ + [ + 408, + 63 + ], + [ + 406, + 66 + ], + [ + 386, + 46 + ], + [ + 431, + 79 + ], + [ + 389, + 44 + ], + [ + 432, + 75 + ] + ], + [ + [ + 373, + 52 + ], + [ + 355, + 30 + ], + [ + 359, + 57 + ], + [ + 387, + 46 + ], + [ + 376, + 77 + ], + [ + 405, + 68 + ] + ] + ] + }, + { + "task_id": "0000407", + "id": "RH20T_cfg2_task_0071_user_0007_scene_0006_cfg_0002_037522062165#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000407/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 323, + 215 + ], + [ + 325, + 194 + ], + [ + 348, + 221 + ], + [ + 298, + 209 + ], + [ + 346, + 243 + ], + [ + 295, + 230 + ] + ], + [ + [ + 320, + 261 + ], + [ + 323, + 288 + ], + [ + 332, + 256 + ], + [ + 305, + 267 + ], + [ + 329, + 230 + ], + [ + 302, + 240 + ] + ], + [ + [ + 317, + 274 + ], + [ + 317, + 248 + ], + [ + 314, + 272 + ], + [ + 320, + 276 + ], + [ + 313, + 300 + ], + [ + 319, + 301 + ] + ], + [ + [ + 321, + 251 + ], + [ + 322, + 237 + ], + [ + 343, + 239 + ], + [ + 298, + 263 + ], + [ + 341, + 251 + ], + [ + 298, + 274 + ] + ] + ] + }, + { + "task_id": "0000409", + "id": "RH20T_cfg1_task_0033_user_0014_scene_0006_cfg_0001_039422060546#128", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000409/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 236, + 181 + ], + [ + 230, + 151 + ], + [ + 235, + 183 + ], + [ + 238, + 178 + ], + [ + 241, + 211 + ], + [ + 245, + 210 + ] + ], + [ + [ + 261, + 180 + ], + [ + 256, + 163 + ], + [ + 248, + 207 + ], + [ + 272, + 154 + ], + [ + 253, + 221 + ], + [ + 276, + 170 + ] + ], + [ + [ + 280, + 176 + ], + [ + 248, + 167 + ], + [ + 268, + 191 + ], + [ + 291, + 162 + ], + [ + 302, + 201 + ], + [ + 321, + 170 + ] + ], + [ + [ + 260, + 144 + ], + [ + 238, + 167 + ], + [ + 260, + 138 + ], + [ + 259, + 151 + ], + [ + 280, + 115 + ], + [ + 283, + 125 + ] + ] + ] + }, + { + "task_id": "0000411", + "id": "RH20T_cfg2_task_0057_user_0016_scene_0001_cfg_0002_104122063678#582", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000411/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 487, + 344 + ], + [ + 496, + 332 + ], + [ + 494, + 360 + ], + [ + 481, + 316 + ], + [ + 486, + 360 + ], + [ + 474, + 328 + ] + ], + [ + [ + 496, + 360 + ], + [ + 520, + 360 + ], + [ + 481, + 360 + ], + [ + 510, + 351 + ], + [ + 458, + 359 + ], + [ + 488, + 326 + ] + ], + [ + [ + 492, + 356 + ], + [ + 485, + 360 + ], + [ + 496, + 330 + ], + [ + 488, + 360 + ], + [ + 503, + 314 + ], + [ + 496, + 360 + ] + ], + [ + [ + 470, + 344 + ], + [ + 500, + 344 + ], + [ + 470, + 314 + ], + [ + 469, + 360 + ], + [ + 440, + 313 + ], + [ + 439, + 360 + ] + ] + ] + }, + { + "task_id": "0000413", + "id": "RH20T_cfg3_task_0018_user_0011_scene_0004_cfg_0003_104422070011#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000413/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 311, + 178 + ], + [ + 312, + 172 + ], + [ + 286, + 184 + ], + [ + 336, + 172 + ], + [ + 284, + 191 + ], + [ + 337, + 179 + ] + ], + [ + [ + 282, + 162 + ], + [ + 256, + 159 + ], + [ + 285, + 157 + ], + [ + 279, + 169 + ], + [ + 310, + 160 + ], + [ + 307, + 172 + ] + ], + [ + [ + 303, + 150 + ], + [ + 280, + 150 + ], + [ + 310, + 156 + ], + [ + 295, + 144 + ], + [ + 332, + 156 + ], + [ + 320, + 144 + ] + ], + [ + [ + 300, + 161 + ], + [ + 307, + 149 + ], + [ + 277, + 168 + ], + [ + 322, + 155 + ], + [ + 272, + 179 + ], + [ + 315, + 166 + ] + ] + ] + }, + { + "task_id": "0000414", + "id": "RH20T_cfg3_task_0032_user_0011_scene_0001_cfg_0003_038522062288#161", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000414/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 132, + 218 + ], + [ + 165, + 219 + ], + [ + 129, + 245 + ], + [ + 135, + 190 + ], + [ + 95, + 246 + ], + [ + 100, + 188 + ] + ], + [ + [ + 160, + 206 + ], + [ + 134, + 220 + ], + [ + 150, + 194 + ], + [ + 172, + 218 + ], + [ + 174, + 182 + ], + [ + 198, + 204 + ] + ], + [ + [ + 165, + 228 + ], + [ + 157, + 209 + ], + [ + 149, + 213 + ], + [ + 182, + 244 + ], + [ + 158, + 231 + ], + [ + 190, + 261 + ] + ], + [ + [ + 154, + 235 + ], + [ + 123, + 235 + ], + [ + 149, + 258 + ], + [ + 160, + 211 + ], + [ + 178, + 257 + ], + [ + 190, + 212 + ] + ] + ] + }, + { + "task_id": "0000417", + "id": "RH20T_cfg3_task_0032_user_0011_scene_0001_cfg_0003_038522062288#358", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000417/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 214, + 256 + ], + [ + 193, + 277 + ], + [ + 237, + 276 + ], + [ + 192, + 236 + ], + [ + 255, + 255 + ], + [ + 212, + 217 + ] + ], + [ + [ + 204, + 239 + ], + [ + 193, + 219 + ], + [ + 186, + 226 + ], + [ + 223, + 253 + ], + [ + 197, + 245 + ], + [ + 233, + 271 + ] + ], + [ + [ + 224, + 231 + ], + [ + 220, + 238 + ], + [ + 245, + 209 + ], + [ + 202, + 253 + ], + [ + 250, + 200 + ], + [ + 205, + 247 + ] + ], + [ + [ + 201, + 209 + ], + [ + 213, + 181 + ], + [ + 230, + 221 + ], + [ + 171, + 195 + ], + [ + 218, + 250 + ], + [ + 157, + 225 + ] + ] + ] + }, + { + "task_id": "0000418", + "id": "RH20T_cfg3_task_0032_user_0011_scene_0001_cfg_0003_038522062288#430", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000418/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 253, + 268 + ], + [ + 282, + 247 + ], + [ + 240, + 257 + ], + [ + 265, + 278 + ], + [ + 209, + 279 + ], + [ + 238, + 298 + ] + ], + [ + [ + 300, + 284 + ], + [ + 291, + 298 + ], + [ + 292, + 319 + ], + [ + 307, + 251 + ], + [ + 303, + 305 + ], + [ + 317, + 234 + ] + ], + [ + [ + 293, + 265 + ], + [ + 286, + 244 + ], + [ + 275, + 243 + ], + [ + 312, + 289 + ], + [ + 282, + 261 + ], + [ + 317, + 307 + ] + ], + [ + [ + 305, + 271 + ], + [ + 334, + 256 + ], + [ + 295, + 259 + ], + [ + 316, + 286 + ], + [ + 266, + 274 + ], + [ + 283, + 303 + ] + ] + ] + }, + { + "task_id": "0000421", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0009_cfg_0004_036422060909#232", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000421/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 173 + ], + [ + 319, + 158 + ], + [ + 324, + 154 + ], + [ + 303, + 194 + ], + [ + 319, + 168 + ], + [ + 299, + 207 + ] + ], + [ + [ + 340, + 170 + ], + [ + 319, + 182 + ], + [ + 327, + 166 + ], + [ + 354, + 174 + ], + [ + 347, + 155 + ], + [ + 375, + 162 + ] + ], + [ + [ + 340, + 170 + ], + [ + 341, + 145 + ], + [ + 313, + 166 + ], + [ + 367, + 173 + ], + [ + 312, + 192 + ], + [ + 366, + 199 + ] + ], + [ + [ + 302, + 164 + ], + [ + 317, + 185 + ], + [ + 280, + 175 + ], + [ + 322, + 154 + ], + [ + 265, + 154 + ], + [ + 307, + 135 + ] + ] + ] + }, + { + "task_id": "0000424", + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_039422060546#103", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000424/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 259, + 161 + ], + [ + 261, + 132 + ], + [ + 265, + 193 + ], + [ + 254, + 131 + ], + [ + 264, + 218 + ], + [ + 253, + 157 + ] + ], + [ + [ + 262, + 161 + ], + [ + 266, + 199 + ], + [ + 305, + 157 + ], + [ + 217, + 166 + ], + [ + 300, + 123 + ], + [ + 216, + 131 + ] + ], + [ + [ + 300, + 162 + ], + [ + 315, + 154 + ], + [ + 302, + 205 + ], + [ + 298, + 118 + ], + [ + 285, + 219 + ], + [ + 281, + 124 + ] + ], + [ + [ + 291, + 121 + ], + [ + 286, + 93 + ], + [ + 330, + 106 + ], + [ + 253, + 135 + ], + [ + 338, + 136 + ], + [ + 256, + 166 + ] + ] + ] + }, + { + "task_id": "0000425", + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_039422060546#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000425/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 183, + 166 + ], + [ + 178, + 144 + ], + [ + 183, + 201 + ], + [ + 183, + 134 + ], + [ + 187, + 218 + ], + [ + 186, + 155 + ] + ], + [ + [ + 176, + 180 + ], + [ + 145, + 150 + ], + [ + 143, + 207 + ], + [ + 208, + 154 + ], + [ + 173, + 234 + ], + [ + 235, + 182 + ] + ], + [ + [ + 150, + 194 + ], + [ + 187, + 178 + ], + [ + 172, + 212 + ], + [ + 124, + 173 + ], + [ + 138, + 226 + ], + [ + 86, + 190 + ] + ], + [ + [ + 159, + 146 + ], + [ + 171, + 118 + ], + [ + 124, + 160 + ], + [ + 190, + 133 + ], + [ + 115, + 185 + ], + [ + 179, + 157 + ] + ] + ] + }, + { + "task_id": "0000426", + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_039422060546#241", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000426/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 259, + 201 + ], + [ + 216, + 206 + ], + [ + 257, + 237 + ], + [ + 261, + 162 + ], + [ + 296, + 231 + ], + [ + 302, + 158 + ] + ], + [ + [ + 237, + 254 + ], + [ + 186, + 258 + ], + [ + 241, + 293 + ], + [ + 232, + 211 + ], + [ + 289, + 289 + ], + [ + 285, + 208 + ] + ], + [ + [ + 243, + 222 + ], + [ + 245, + 183 + ], + [ + 261, + 251 + ], + [ + 227, + 198 + ], + [ + 258, + 288 + ], + [ + 226, + 232 + ] + ], + [ + [ + 247, + 242 + ], + [ + 286, + 228 + ], + [ + 227, + 221 + ], + [ + 269, + 267 + ], + [ + 186, + 235 + ], + [ + 223, + 285 + ] + ] + ] + }, + { + "task_id": "0000427", + "id": "RH20T_cfg7_task_0036_user_0014_scene_0005_cfg_0007_104122061330#134", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Shake the test tube'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000427/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 170 + ], + [ + 334, + 171 + ], + [ + 310, + 198 + ], + [ + 309, + 142 + ], + [ + 286, + 197 + ], + [ + 285, + 142 + ] + ], + [ + [ + 321, + 149 + ], + [ + 309, + 173 + ], + [ + 326, + 160 + ], + [ + 315, + 136 + ], + [ + 337, + 137 + ], + [ + 327, + 111 + ] + ], + [ + [ + 335, + 160 + ], + [ + 345, + 154 + ], + [ + 357, + 175 + ], + [ + 314, + 144 + ], + [ + 348, + 180 + ], + [ + 307, + 150 + ] + ], + [ + [ + 331, + 176 + ], + [ + 325, + 205 + ], + [ + 348, + 180 + ], + [ + 313, + 172 + ], + [ + 354, + 152 + ], + [ + 320, + 142 + ] + ] + ] + }, + { + "task_id": "0000429", + "id": "RH20T_cfg1_task_0083_user_0014_scene_0003_cfg_0001_038522062547#149", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Swat with a flyswatter'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000429/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 466, + 79 + ], + [ + 448, + 62 + ], + [ + 484, + 61 + ], + [ + 449, + 95 + ], + [ + 502, + 79 + ], + [ + 465, + 112 + ] + ], + [ + [ + 446, + 76 + ], + [ + 426, + 87 + ], + [ + 452, + 98 + ], + [ + 439, + 54 + ], + [ + 470, + 88 + ], + [ + 458, + 44 + ] + ], + [ + [ + 451, + 106 + ], + [ + 468, + 87 + ], + [ + 470, + 122 + ], + [ + 433, + 91 + ], + [ + 453, + 140 + ], + [ + 416, + 110 + ] + ], + [ + [ + 453, + 93 + ], + [ + 460, + 79 + ], + [ + 468, + 111 + ], + [ + 439, + 76 + ], + [ + 461, + 123 + ], + [ + 433, + 90 + ] + ] + ] + }, + { + "task_id": "0000432", + "id": "RH20T_cfg3_task_0032_user_0011_scene_0002_cfg_0003_104122063550#130", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000432/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 295, + 121 + ], + [ + 273, + 113 + ], + [ + 286, + 141 + ], + [ + 303, + 102 + ], + [ + 308, + 149 + ], + [ + 325, + 109 + ] + ], + [ + [ + 295, + 138 + ], + [ + 296, + 124 + ], + [ + 284, + 157 + ], + [ + 305, + 120 + ], + [ + 284, + 170 + ], + [ + 303, + 134 + ] + ], + [ + [ + 303, + 172 + ], + [ + 321, + 164 + ], + [ + 292, + 186 + ], + [ + 312, + 160 + ], + [ + 274, + 193 + ], + [ + 295, + 168 + ] + ], + [ + [ + 281, + 131 + ], + [ + 299, + 126 + ], + [ + 263, + 118 + ], + [ + 298, + 143 + ], + [ + 246, + 123 + ], + [ + 281, + 147 + ] + ] + ] + }, + { + "task_id": "0000433", + "id": "RH20T_cfg3_task_0070_user_0010_scene_0008_cfg_0003_104122062823#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000433/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 234, + 36 + ], + [ + 258, + 30 + ], + [ + 224, + 46 + ], + [ + 245, + 24 + ], + [ + 198, + 51 + ], + [ + 217, + 30 + ] + ], + [ + [ + 272, + 81 + ], + [ + 299, + 85 + ], + [ + 256, + 93 + ], + [ + 287, + 71 + ], + [ + 229, + 89 + ], + [ + 262, + 68 + ] + ], + [ + [ + 252, + 73 + ], + [ + 258, + 50 + ], + [ + 280, + 80 + ], + [ + 223, + 65 + ], + [ + 272, + 100 + ], + [ + 218, + 86 + ] + ], + [ + [ + 266, + 87 + ], + [ + 264, + 64 + ], + [ + 284, + 96 + ], + [ + 247, + 78 + ], + [ + 286, + 120 + ], + [ + 249, + 103 + ] + ] + ] + }, + { + "task_id": "0000434", + "id": "RH20T_cfg3_task_0070_user_0010_scene_0008_cfg_0003_104122062823#271", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000434/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 268, + 74 + ], + [ + 274, + 59 + ], + [ + 274, + 95 + ], + [ + 261, + 51 + ], + [ + 268, + 112 + ], + [ + 255, + 68 + ] + ], + [ + [ + 235, + 59 + ], + [ + 209, + 48 + ], + [ + 245, + 32 + ], + [ + 224, + 85 + ], + [ + 269, + 43 + ], + [ + 248, + 94 + ] + ], + [ + [ + 247, + 63 + ], + [ + 254, + 40 + ], + [ + 274, + 73 + ], + [ + 220, + 53 + ], + [ + 267, + 94 + ], + [ + 215, + 76 + ] + ], + [ + [ + 265, + 78 + ], + [ + 272, + 59 + ], + [ + 250, + 90 + ], + [ + 283, + 64 + ], + [ + 243, + 110 + ], + [ + 276, + 85 + ] + ] + ] + }, + { + "task_id": "0000437", + "id": "RH20T_cfg1_task_0066_user_0015_scene_0006_cfg_0001_819112070044#667", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000437/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 254, + 252 + ], + [ + 274, + 255 + ], + [ + 242, + 241 + ], + [ + 265, + 263 + ], + [ + 222, + 239 + ], + [ + 247, + 260 + ] + ], + [ + [ + 279, + 257 + ], + [ + 281, + 241 + ], + [ + 299, + 248 + ], + [ + 258, + 266 + ], + [ + 297, + 264 + ], + [ + 257, + 282 + ] + ], + [ + [ + 283, + 263 + ], + [ + 289, + 273 + ], + [ + 262, + 256 + ], + [ + 303, + 270 + ], + [ + 257, + 247 + ], + [ + 297, + 261 + ] + ], + [ + [ + 305, + 261 + ], + [ + 299, + 251 + ], + [ + 285, + 252 + ], + [ + 324, + 270 + ], + [ + 290, + 262 + ], + [ + 330, + 281 + ] + ] + ] + }, + { + "task_id": "0000438", + "id": "RH20T_cfg1_task_0066_user_0015_scene_0006_cfg_0001_819112070044#910", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000438/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 257, + 280 + ], + [ + 247, + 257 + ], + [ + 237, + 291 + ], + [ + 276, + 270 + ], + [ + 247, + 316 + ], + [ + 285, + 293 + ] + ], + [ + [ + 243, + 299 + ], + [ + 248, + 323 + ], + [ + 227, + 292 + ], + [ + 259, + 305 + ], + [ + 223, + 268 + ], + [ + 255, + 282 + ] + ], + [ + [ + 248, + 278 + ], + [ + 248, + 262 + ], + [ + 270, + 269 + ], + [ + 225, + 288 + ], + [ + 270, + 284 + ], + [ + 225, + 303 + ] + ], + [ + [ + 227, + 291 + ], + [ + 216, + 288 + ], + [ + 205, + 289 + ], + [ + 249, + 293 + ], + [ + 215, + 292 + ], + [ + 258, + 296 + ] + ] + ] + }, + { + "task_id": "0000439", + "id": "RH20T_cfg1_task_0066_user_0015_scene_0006_cfg_0001_819112070044#1226", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000439/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 300, + 312 + ], + [ + 307, + 292 + ], + [ + 286, + 321 + ], + [ + 315, + 302 + ], + [ + 278, + 342 + ], + [ + 308, + 324 + ] + ], + [ + [ + 309, + 310 + ], + [ + 311, + 293 + ], + [ + 329, + 299 + ], + [ + 287, + 322 + ], + [ + 326, + 315 + ], + [ + 285, + 338 + ] + ], + [ + [ + 322, + 276 + ], + [ + 325, + 253 + ], + [ + 322, + 287 + ], + [ + 321, + 263 + ], + [ + 319, + 310 + ], + [ + 317, + 288 + ] + ], + [ + [ + 302, + 329 + ], + [ + 281, + 317 + ], + [ + 314, + 339 + ], + [ + 292, + 320 + ], + [ + 335, + 351 + ], + [ + 311, + 331 + ] + ] + ] + }, + { + "task_id": "0000443", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_038522062547#384", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000443/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 463, + 209 + ], + [ + 470, + 198 + ], + [ + 467, + 236 + ], + [ + 460, + 183 + ], + [ + 461, + 245 + ], + [ + 454, + 195 + ] + ], + [ + [ + 465, + 185 + ], + [ + 452, + 162 + ], + [ + 491, + 171 + ], + [ + 440, + 199 + ], + [ + 506, + 196 + ], + [ + 453, + 223 + ] + ], + [ + [ + 465, + 234 + ], + [ + 438, + 230 + ], + [ + 458, + 215 + ], + [ + 472, + 256 + ], + [ + 483, + 219 + ], + [ + 499, + 260 + ] + ], + [ + [ + 443, + 230 + ], + [ + 447, + 249 + ], + [ + 425, + 212 + ], + [ + 461, + 247 + ], + [ + 423, + 195 + ], + [ + 457, + 229 + ] + ] + ] + }, + { + "task_id": "0000444", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_038522062547#541", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000444/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 462, + 161 + ], + [ + 435, + 161 + ], + [ + 457, + 136 + ], + [ + 467, + 187 + ], + [ + 484, + 136 + ], + [ + 494, + 187 + ] + ], + [ + [ + 446, + 158 + ], + [ + 453, + 142 + ], + [ + 449, + 184 + ], + [ + 443, + 134 + ], + [ + 443, + 197 + ], + [ + 437, + 149 + ] + ], + [ + [ + 439, + 163 + ], + [ + 471, + 155 + ], + [ + 444, + 162 + ], + [ + 435, + 163 + ], + [ + 411, + 170 + ], + [ + 407, + 171 + ] + ], + [ + [ + 459, + 143 + ], + [ + 488, + 133 + ], + [ + 474, + 169 + ], + [ + 445, + 119 + ], + [ + 445, + 180 + ], + [ + 417, + 129 + ] + ] + ] + }, + { + "task_id": "0000446", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0007_cfg_0003_104122062823#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000446/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 181, + 179 + ], + [ + 179, + 202 + ], + [ + 187, + 152 + ], + [ + 175, + 209 + ], + [ + 190, + 125 + ], + [ + 176, + 182 + ] + ], + [ + [ + 181, + 104 + ], + [ + 188, + 77 + ], + [ + 217, + 113 + ], + [ + 144, + 94 + ], + [ + 209, + 136 + ], + [ + 139, + 120 + ] + ], + [ + [ + 181, + 149 + ], + [ + 175, + 118 + ], + [ + 215, + 155 + ], + [ + 147, + 142 + ], + [ + 218, + 182 + ], + [ + 153, + 170 + ] + ], + [ + [ + 187, + 114 + ], + [ + 169, + 137 + ], + [ + 217, + 136 + ], + [ + 155, + 90 + ], + [ + 237, + 112 + ], + [ + 173, + 63 + ] + ] + ] + }, + { + "task_id": "0000448", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0007_cfg_0003_104122062823#214", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000448/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 167, + 121 + ], + [ + 136, + 103 + ], + [ + 171, + 114 + ], + [ + 162, + 129 + ], + [ + 199, + 131 + ], + [ + 194, + 148 + ] + ], + [ + [ + 165, + 76 + ], + [ + 152, + 82 + ], + [ + 147, + 44 + ], + [ + 182, + 107 + ], + [ + 160, + 41 + ], + [ + 192, + 100 + ] + ], + [ + [ + 197, + 111 + ], + [ + 193, + 82 + ], + [ + 228, + 116 + ], + [ + 167, + 105 + ], + [ + 231, + 142 + ], + [ + 172, + 131 + ] + ], + [ + [ + 218, + 88 + ], + [ + 189, + 106 + ], + [ + 212, + 62 + ], + [ + 223, + 112 + ], + [ + 241, + 45 + ], + [ + 249, + 96 + ] + ] + ] + }, + { + "task_id": "0000450", + "id": "RH20T_cfg3_task_0059_user_0010_scene_0002_cfg_0003_104122063550#78", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000450/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 316, + 265 + ], + [ + 324, + 253 + ], + [ + 340, + 282 + ], + [ + 293, + 248 + ], + [ + 332, + 292 + ], + [ + 287, + 260 + ] + ], + [ + [ + 286, + 256 + ], + [ + 303, + 269 + ], + [ + 303, + 273 + ], + [ + 271, + 242 + ], + [ + 284, + 259 + ], + [ + 252, + 228 + ] + ], + [ + [ + 306, + 253 + ], + [ + 308, + 278 + ], + [ + 329, + 259 + ], + [ + 282, + 247 + ], + [ + 327, + 236 + ], + [ + 281, + 223 + ] + ], + [ + [ + 288, + 270 + ], + [ + 302, + 243 + ], + [ + 291, + 283 + ], + [ + 285, + 258 + ], + [ + 276, + 311 + ], + [ + 272, + 283 + ] + ] + ] + }, + { + "task_id": "0000452", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_750612070853#202", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000452/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 446, + 166 + ], + [ + 454, + 118 + ], + [ + 406, + 149 + ], + [ + 493, + 187 + ], + [ + 402, + 189 + ], + [ + 483, + 231 + ] + ], + [ + [ + 426, + 130 + ], + [ + 386, + 127 + ], + [ + 449, + 134 + ], + [ + 399, + 125 + ], + [ + 494, + 138 + ], + [ + 448, + 129 + ] + ], + [ + [ + 387, + 119 + ], + [ + 341, + 138 + ], + [ + 419, + 154 + ], + [ + 360, + 89 + ], + [ + 464, + 134 + ], + [ + 402, + 73 + ] + ], + [ + [ + 501, + 120 + ], + [ + 481, + 119 + ], + [ + 503, + 65 + ], + [ + 499, + 173 + ], + [ + 529, + 60 + ], + [ + 524, + 181 + ] + ] + ] + }, + { + "task_id": "0000453", + "id": "RH20T_cfg7_task_0050_user_0020_scene_0008_cfg_0007_104122061018#123", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the dish on the dish rack'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000453/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 137 + ], + [ + 293, + 121 + ], + [ + 263, + 116 + ], + [ + 304, + 159 + ], + [ + 257, + 131 + ], + [ + 295, + 172 + ] + ], + [ + [ + 268, + 154 + ], + [ + 294, + 172 + ], + [ + 260, + 164 + ], + [ + 278, + 143 + ], + [ + 235, + 147 + ], + [ + 251, + 124 + ] + ], + [ + [ + 262, + 156 + ], + [ + 286, + 141 + ], + [ + 245, + 133 + ], + [ + 279, + 179 + ], + [ + 221, + 148 + ], + [ + 255, + 195 + ] + ], + [ + [ + 276, + 126 + ], + [ + 272, + 97 + ], + [ + 295, + 119 + ], + [ + 258, + 132 + ], + [ + 298, + 150 + ], + [ + 262, + 159 + ] + ] + ] + }, + { + "task_id": "0000456", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0009_cfg_0007_104122064161#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000456/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 373, + 163 + ], + [ + 389, + 149 + ], + [ + 360, + 158 + ], + [ + 387, + 168 + ], + [ + 344, + 172 + ], + [ + 370, + 183 + ] + ], + [ + [ + 377, + 162 + ], + [ + 374, + 184 + ], + [ + 374, + 157 + ], + [ + 380, + 166 + ], + [ + 377, + 135 + ], + [ + 384, + 142 + ] + ], + [ + [ + 364, + 172 + ], + [ + 380, + 188 + ], + [ + 352, + 188 + ], + [ + 376, + 155 + ], + [ + 338, + 173 + ], + [ + 360, + 139 + ] + ], + [ + [ + 396, + 175 + ], + [ + 410, + 189 + ], + [ + 378, + 185 + ], + [ + 415, + 164 + ], + [ + 363, + 172 + ], + [ + 401, + 149 + ] + ] + ] + }, + { + "task_id": "0000460", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122061330#464", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000460/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 342, + 176 + ], + [ + 361, + 176 + ], + [ + 322, + 181 + ], + [ + 361, + 171 + ], + [ + 305, + 181 + ], + [ + 343, + 171 + ] + ], + [ + [ + 316, + 175 + ], + [ + 342, + 173 + ], + [ + 312, + 149 + ], + [ + 319, + 202 + ], + [ + 286, + 152 + ], + [ + 292, + 205 + ] + ], + [ + [ + 311, + 178 + ], + [ + 293, + 195 + ], + [ + 312, + 165 + ], + [ + 310, + 193 + ], + [ + 330, + 148 + ], + [ + 330, + 175 + ] + ], + [ + [ + 341, + 163 + ], + [ + 330, + 156 + ], + [ + 334, + 188 + ], + [ + 348, + 137 + ], + [ + 344, + 193 + ], + [ + 357, + 145 + ] + ] + ] + }, + { + "task_id": "0000462", + "id": "RH20T_cfg5_task_0123_user_0010_scene_0004_cfg_0005_104122061850#266", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the knife on the knife rack'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000462/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 362, + 137 + ], + [ + 380, + 153 + ], + [ + 370, + 130 + ], + [ + 355, + 143 + ], + [ + 351, + 113 + ], + [ + 338, + 128 + ] + ], + [ + [ + 367, + 131 + ], + [ + 391, + 126 + ], + [ + 362, + 107 + ], + [ + 372, + 155 + ], + [ + 338, + 113 + ], + [ + 348, + 161 + ] + ], + [ + [ + 406, + 134 + ], + [ + 384, + 125 + ], + [ + 411, + 138 + ], + [ + 401, + 128 + ], + [ + 433, + 147 + ], + [ + 425, + 137 + ] + ], + [ + [ + 386, + 118 + ], + [ + 384, + 102 + ], + [ + 374, + 136 + ], + [ + 397, + 101 + ], + [ + 375, + 151 + ], + [ + 398, + 116 + ] + ] + ] + }, + { + "task_id": "0000466", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122064161#755", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000466/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 364, + 270 + ], + [ + 375, + 294 + ], + [ + 343, + 286 + ], + [ + 385, + 255 + ], + [ + 331, + 262 + ], + [ + 374, + 231 + ] + ], + [ + [ + 340, + 245 + ], + [ + 340, + 243 + ], + [ + 337, + 272 + ], + [ + 343, + 219 + ], + [ + 337, + 273 + ], + [ + 343, + 222 + ] + ], + [ + [ + 313, + 267 + ], + [ + 307, + 247 + ], + [ + 328, + 282 + ], + [ + 300, + 254 + ], + [ + 334, + 302 + ], + [ + 306, + 272 + ] + ], + [ + [ + 366, + 225 + ], + [ + 339, + 221 + ], + [ + 372, + 202 + ], + [ + 360, + 247 + ], + [ + 399, + 207 + ], + [ + 385, + 251 + ] + ] + ] + }, + { + "task_id": "0000470", + "id": "RH20T_cfg2_task_0018_user_0003_scene_0010_cfg_0002_104122063678#206", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000470/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 329, + 303 + ], + [ + 318, + 286 + ], + [ + 352, + 303 + ], + [ + 304, + 303 + ], + [ + 365, + 322 + ], + [ + 314, + 322 + ] + ], + [ + [ + 346, + 327 + ], + [ + 330, + 306 + ], + [ + 353, + 310 + ], + [ + 338, + 343 + ], + [ + 371, + 333 + ], + [ + 354, + 360 + ] + ], + [ + [ + 361, + 304 + ], + [ + 361, + 291 + ], + [ + 345, + 326 + ], + [ + 377, + 283 + ], + [ + 346, + 337 + ], + [ + 376, + 295 + ] + ], + [ + [ + 358, + 321 + ], + [ + 339, + 339 + ], + [ + 369, + 345 + ], + [ + 347, + 297 + ], + [ + 387, + 328 + ], + [ + 366, + 281 + ] + ] + ] + }, + { + "task_id": "0000471", + "id": "RH20T_cfg1_task_0001_user_0009_scene_0003_cfg_0001_819112071037#111", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000471/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 337, + 111 + ], + [ + 330, + 81 + ], + [ + 391, + 72 + ], + [ + 289, + 146 + ], + [ + 407, + 106 + ], + [ + 292, + 185 + ] + ], + [ + [ + 346, + 89 + ], + [ + 359, + 63 + ], + [ + 344, + 30 + ], + [ + 348, + 140 + ], + [ + 326, + 58 + ], + [ + 334, + 174 + ] + ], + [ + [ + 335, + 11 + ], + [ + 274, + 11 + ], + [ + 350, + 18 + ], + [ + 315, + 2 + ], + [ + 409, + 17 + ], + [ + 391, + 1 + ] + ], + [ + [ + 309, + 71 + ], + [ + 310, + 0 + ], + [ + 242, + 70 + ], + [ + 375, + 73 + ], + [ + 245, + 133 + ], + [ + 370, + 135 + ] + ] + ] + }, + { + "task_id": "0000472", + "id": "RH20T_cfg6_task_0129_user_0020_scene_0009_cfg_0006_104122061018#213", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000472/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 221 + ], + [ + 304, + 186 + ], + [ + 340, + 219 + ], + [ + 272, + 223 + ], + [ + 342, + 255 + ], + [ + 273, + 258 + ] + ], + [ + [ + 325, + 217 + ], + [ + 312, + 184 + ], + [ + 319, + 213 + ], + [ + 331, + 221 + ], + [ + 331, + 243 + ], + [ + 345, + 256 + ] + ], + [ + [ + 312, + 199 + ], + [ + 337, + 176 + ], + [ + 306, + 198 + ], + [ + 318, + 200 + ], + [ + 282, + 220 + ], + [ + 289, + 225 + ] + ], + [ + [ + 313, + 254 + ], + [ + 298, + 277 + ], + [ + 288, + 260 + ], + [ + 336, + 249 + ], + [ + 305, + 235 + ], + [ + 352, + 226 + ] + ] + ] + }, + { + "task_id": "0000475", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_104122061850#463", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000475/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 116 + ], + [ + 295, + 114 + ], + [ + 301, + 91 + ], + [ + 312, + 142 + ], + [ + 312, + 94 + ], + [ + 322, + 143 + ] + ], + [ + [ + 283, + 108 + ], + [ + 273, + 131 + ], + [ + 277, + 124 + ], + [ + 290, + 90 + ], + [ + 285, + 104 + ], + [ + 299, + 69 + ] + ], + [ + [ + 301, + 116 + ], + [ + 320, + 129 + ], + [ + 321, + 115 + ], + [ + 280, + 118 + ], + [ + 304, + 104 + ], + [ + 263, + 106 + ] + ], + [ + [ + 300, + 116 + ], + [ + 298, + 95 + ], + [ + 281, + 133 + ], + [ + 319, + 100 + ], + [ + 284, + 151 + ], + [ + 320, + 119 + ] + ] + ] + }, + { + "task_id": "0000476", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_104122061850#533", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000476/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 386, + 145 + ], + [ + 373, + 115 + ], + [ + 403, + 136 + ], + [ + 368, + 155 + ], + [ + 415, + 164 + ], + [ + 382, + 185 + ] + ], + [ + [ + 367, + 150 + ], + [ + 345, + 143 + ], + [ + 367, + 176 + ], + [ + 366, + 123 + ], + [ + 390, + 185 + ], + [ + 390, + 130 + ] + ], + [ + [ + 388, + 176 + ], + [ + 399, + 195 + ], + [ + 392, + 149 + ], + [ + 385, + 201 + ], + [ + 382, + 132 + ], + [ + 376, + 183 + ] + ], + [ + [ + 381, + 141 + ], + [ + 382, + 122 + ], + [ + 360, + 161 + ], + [ + 401, + 123 + ], + [ + 360, + 177 + ], + [ + 398, + 140 + ] + ] + ] + }, + { + "task_id": "0000477", + "id": "RH20T_cfg1_task_0043_user_0015_scene_0004_cfg_0001_750612070853#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000477/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 451, + 200 + ], + [ + 470, + 224 + ], + [ + 477, + 160 + ], + [ + 424, + 241 + ], + [ + 457, + 129 + ], + [ + 400, + 215 + ] + ], + [ + [ + 393, + 175 + ], + [ + 369, + 179 + ], + [ + 375, + 133 + ], + [ + 411, + 215 + ], + [ + 402, + 124 + ], + [ + 440, + 214 + ] + ], + [ + [ + 429, + 153 + ], + [ + 428, + 109 + ], + [ + 378, + 154 + ], + [ + 481, + 151 + ], + [ + 382, + 192 + ], + [ + 479, + 190 + ] + ], + [ + [ + 415, + 134 + ], + [ + 377, + 135 + ], + [ + 450, + 129 + ], + [ + 376, + 140 + ], + [ + 497, + 128 + ], + [ + 421, + 139 + ] + ] + ] + }, + { + "task_id": "0000479", + "id": "RH20T_cfg5_task_0096_user_0007_scene_0008_cfg_0005_105422061350#481", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000479/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 260, + 0 + ], + [ + 240, + 5 + ], + [ + 278, + 9 + ], + [ + 240, + 0 + ], + [ + 296, + 0 + ], + [ + 260, + 0 + ] + ], + [ + [ + 306, + 18 + ], + [ + 296, + 7 + ], + [ + 329, + 6 + ], + [ + 282, + 31 + ], + [ + 341, + 17 + ], + [ + 291, + 43 + ] + ], + [ + [ + 276, + 17 + ], + [ + 303, + 10 + ], + [ + 270, + 0 + ], + [ + 281, + 44 + ], + [ + 242, + 0 + ], + [ + 255, + 51 + ] + ], + [ + [ + 275, + 7 + ], + [ + 279, + 0 + ], + [ + 302, + 18 + ], + [ + 249, + 0 + ], + [ + 297, + 40 + ], + [ + 247, + 19 + ] + ] + ] + }, + { + "task_id": "0000481", + "id": "RH20T_cfg1_task_0016_user_0014_scene_0001_cfg_0001_038522062547#413", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000481/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 382, + 52 + ], + [ + 381, + 35 + ], + [ + 401, + 41 + ], + [ + 364, + 61 + ], + [ + 402, + 59 + ], + [ + 365, + 79 + ] + ], + [ + [ + 387, + 38 + ], + [ + 388, + 43 + ], + [ + 391, + 61 + ], + [ + 383, + 14 + ], + [ + 390, + 56 + ], + [ + 382, + 11 + ] + ], + [ + [ + 389, + 59 + ], + [ + 390, + 46 + ], + [ + 412, + 57 + ], + [ + 366, + 62 + ], + [ + 409, + 70 + ], + [ + 365, + 75 + ] + ], + [ + [ + 374, + 54 + ], + [ + 374, + 35 + ], + [ + 361, + 65 + ], + [ + 389, + 43 + ], + [ + 361, + 84 + ], + [ + 390, + 63 + ] + ] + ] + }, + { + "task_id": "0000482", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122061602#135", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000482/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 426, + 147 + ], + [ + 419, + 169 + ], + [ + 407, + 143 + ], + [ + 445, + 151 + ], + [ + 415, + 121 + ], + [ + 453, + 130 + ] + ], + [ + [ + 446, + 117 + ], + [ + 435, + 131 + ], + [ + 466, + 133 + ], + [ + 428, + 101 + ], + [ + 479, + 118 + ], + [ + 439, + 85 + ] + ], + [ + [ + 452, + 140 + ], + [ + 458, + 120 + ], + [ + 469, + 155 + ], + [ + 437, + 126 + ], + [ + 463, + 173 + ], + [ + 433, + 144 + ] + ], + [ + [ + 455, + 156 + ], + [ + 436, + 156 + ], + [ + 456, + 132 + ], + [ + 454, + 180 + ], + [ + 475, + 132 + ], + [ + 473, + 179 + ] + ] + ] + }, + { + "task_id": "0000483", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122061602#322", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000483/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 474, + 132 + ], + [ + 450, + 139 + ], + [ + 480, + 155 + ], + [ + 467, + 110 + ], + [ + 505, + 148 + ], + [ + 491, + 103 + ] + ], + [ + [ + 460, + 160 + ], + [ + 443, + 144 + ], + [ + 476, + 142 + ], + [ + 444, + 178 + ], + [ + 494, + 158 + ], + [ + 461, + 194 + ] + ], + [ + [ + 475, + 137 + ], + [ + 495, + 146 + ], + [ + 490, + 117 + ], + [ + 461, + 157 + ], + [ + 471, + 109 + ], + [ + 443, + 148 + ] + ], + [ + [ + 458, + 152 + ], + [ + 464, + 135 + ], + [ + 474, + 167 + ], + [ + 443, + 138 + ], + [ + 467, + 183 + ], + [ + 438, + 154 + ] + ] + ] + }, + { + "task_id": "0000484", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122061602#471", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000484/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 460, + 184 + ], + [ + 477, + 183 + ], + [ + 473, + 204 + ], + [ + 448, + 166 + ], + [ + 454, + 206 + ], + [ + 431, + 168 + ] + ], + [ + [ + 470, + 173 + ], + [ + 491, + 188 + ], + [ + 468, + 171 + ], + [ + 473, + 174 + ], + [ + 447, + 156 + ], + [ + 453, + 160 + ] + ], + [ + [ + 445, + 164 + ], + [ + 425, + 173 + ], + [ + 457, + 157 + ], + [ + 434, + 170 + ], + [ + 477, + 148 + ], + [ + 452, + 161 + ] + ], + [ + [ + 451, + 167 + ], + [ + 460, + 156 + ], + [ + 466, + 187 + ], + [ + 437, + 148 + ], + [ + 457, + 197 + ], + [ + 429, + 159 + ] + ] + ] + }, + { + "task_id": "0000487", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0004_cfg_0006_104122061018#522", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000487/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 287, + 135 + ], + [ + 306, + 146 + ], + [ + 257, + 127 + ], + [ + 314, + 142 + ], + [ + 241, + 117 + ], + [ + 296, + 132 + ] + ], + [ + [ + 299, + 113 + ], + [ + 286, + 90 + ], + [ + 298, + 133 + ], + [ + 300, + 91 + ], + [ + 310, + 157 + ], + [ + 315, + 118 + ] + ], + [ + [ + 320, + 112 + ], + [ + 291, + 96 + ], + [ + 324, + 93 + ], + [ + 317, + 129 + ], + [ + 356, + 111 + ], + [ + 346, + 145 + ] + ], + [ + [ + 307, + 125 + ], + [ + 320, + 154 + ], + [ + 278, + 138 + ], + [ + 336, + 112 + ], + [ + 266, + 109 + ], + [ + 323, + 84 + ] + ] + ] + }, + { + "task_id": "0000488", + "id": "RH20T_cfg6_task_0038_user_0020_scene_0010_cfg_0006_104122064161#144", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000488/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 411, + 192 + ], + [ + 385, + 192 + ], + [ + 411, + 199 + ], + [ + 411, + 185 + ], + [ + 437, + 199 + ], + [ + 440, + 186 + ] + ], + [ + [ + 405, + 205 + ], + [ + 408, + 201 + ], + [ + 404, + 232 + ], + [ + 405, + 179 + ], + [ + 401, + 234 + ], + [ + 402, + 184 + ] + ], + [ + [ + 380, + 203 + ], + [ + 400, + 189 + ], + [ + 365, + 190 + ], + [ + 397, + 216 + ], + [ + 345, + 203 + ], + [ + 376, + 231 + ] + ], + [ + [ + 434, + 201 + ], + [ + 463, + 200 + ], + [ + 432, + 182 + ], + [ + 436, + 222 + ], + [ + 405, + 184 + ], + [ + 407, + 223 + ] + ] + ] + }, + { + "task_id": "0000489", + "id": "RH20T_cfg7_task_0036_user_0014_scene_0001_cfg_0007_037522061512#74", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Shake the test tube'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000489/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 273, + 116 + ], + [ + 272, + 103 + ], + [ + 283, + 98 + ], + [ + 263, + 133 + ], + [ + 285, + 110 + ], + [ + 263, + 146 + ] + ], + [ + [ + 272, + 119 + ], + [ + 274, + 124 + ], + [ + 286, + 136 + ], + [ + 256, + 102 + ], + [ + 283, + 131 + ], + [ + 255, + 98 + ] + ], + [ + [ + 266, + 84 + ], + [ + 243, + 81 + ], + [ + 266, + 67 + ], + [ + 265, + 99 + ], + [ + 290, + 70 + ], + [ + 286, + 102 + ] + ], + [ + [ + 274, + 100 + ], + [ + 265, + 88 + ], + [ + 282, + 81 + ], + [ + 265, + 120 + ], + [ + 290, + 93 + ], + [ + 273, + 130 + ] + ] + ] + }, + { + "task_id": "0000491", + "id": "RH20T_cfg6_task_0029_user_0014_scene_0005_cfg_0006_104122060811#149", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000491/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 415, + 151 + ], + [ + 413, + 136 + ], + [ + 423, + 145 + ], + [ + 408, + 157 + ], + [ + 424, + 160 + ], + [ + 410, + 173 + ] + ], + [ + [ + 397, + 146 + ], + [ + 409, + 149 + ], + [ + 387, + 142 + ], + [ + 407, + 151 + ], + [ + 375, + 139 + ], + [ + 396, + 148 + ] + ], + [ + [ + 409, + 143 + ], + [ + 418, + 133 + ], + [ + 421, + 141 + ], + [ + 397, + 144 + ], + [ + 413, + 150 + ], + [ + 389, + 153 + ] + ], + [ + [ + 414, + 145 + ], + [ + 423, + 157 + ], + [ + 423, + 135 + ], + [ + 405, + 153 + ], + [ + 414, + 123 + ], + [ + 396, + 141 + ] + ] + ] + }, + { + "task_id": "0000492", + "id": "RH20T_cfg5_task_0010_user_0010_scene_0007_cfg_0005_037522062165#269", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000492/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 235, + 168 + ], + [ + 237, + 150 + ], + [ + 219, + 154 + ], + [ + 252, + 183 + ], + [ + 218, + 170 + ], + [ + 250, + 199 + ] + ], + [ + [ + 216, + 179 + ], + [ + 197, + 190 + ], + [ + 230, + 198 + ], + [ + 200, + 160 + ], + [ + 249, + 187 + ], + [ + 219, + 148 + ] + ], + [ + [ + 231, + 175 + ], + [ + 248, + 165 + ], + [ + 246, + 194 + ], + [ + 217, + 156 + ], + [ + 229, + 206 + ], + [ + 198, + 166 + ] + ], + [ + [ + 227, + 162 + ], + [ + 249, + 172 + ], + [ + 218, + 177 + ], + [ + 235, + 147 + ], + [ + 193, + 166 + ], + [ + 212, + 137 + ] + ] + ] + }, + { + "task_id": "0000493", + "id": "RH20T_cfg5_task_0085_user_0007_scene_0007_cfg_0005_104422070044#141", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the bubble ring from the assembled bubble ring and ball'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000493/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 421, + 182 + ], + [ + 424, + 163 + ], + [ + 411, + 179 + ], + [ + 431, + 184 + ], + [ + 407, + 198 + ], + [ + 427, + 202 + ] + ], + [ + [ + 414, + 184 + ], + [ + 414, + 172 + ], + [ + 406, + 197 + ], + [ + 422, + 172 + ], + [ + 407, + 208 + ], + [ + 422, + 184 + ] + ], + [ + [ + 404, + 196 + ], + [ + 401, + 207 + ], + [ + 396, + 207 + ], + [ + 412, + 184 + ], + [ + 399, + 195 + ], + [ + 416, + 172 + ] + ], + [ + [ + 401, + 191 + ], + [ + 408, + 205 + ], + [ + 410, + 197 + ], + [ + 393, + 185 + ], + [ + 403, + 182 + ], + [ + 387, + 171 + ] + ] + ] + }, + { + "task_id": "0000494", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122063633#339", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000494/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 237, + 135 + ], + [ + 259, + 136 + ], + [ + 236, + 155 + ], + [ + 238, + 116 + ], + [ + 213, + 154 + ], + [ + 217, + 115 + ] + ], + [ + [ + 218, + 140 + ], + [ + 221, + 120 + ], + [ + 220, + 142 + ], + [ + 216, + 138 + ], + [ + 217, + 162 + ], + [ + 213, + 160 + ] + ], + [ + [ + 224, + 124 + ], + [ + 203, + 118 + ], + [ + 225, + 115 + ], + [ + 224, + 133 + ], + [ + 247, + 122 + ], + [ + 244, + 139 + ] + ], + [ + [ + 226, + 119 + ], + [ + 246, + 121 + ], + [ + 218, + 110 + ], + [ + 232, + 127 + ], + [ + 196, + 108 + ], + [ + 213, + 125 + ] + ] + ] + }, + { + "task_id": "0000495", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122063633#743", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000495/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 173, + 142 + ], + [ + 194, + 145 + ], + [ + 170, + 129 + ], + [ + 176, + 154 + ], + [ + 148, + 125 + ], + [ + 155, + 151 + ] + ], + [ + [ + 195, + 132 + ], + [ + 200, + 114 + ], + [ + 214, + 138 + ], + [ + 175, + 126 + ], + [ + 209, + 155 + ], + [ + 170, + 144 + ] + ], + [ + [ + 177, + 117 + ], + [ + 177, + 97 + ], + [ + 182, + 111 + ], + [ + 172, + 124 + ], + [ + 182, + 130 + ], + [ + 171, + 144 + ] + ], + [ + [ + 191, + 110 + ], + [ + 200, + 91 + ], + [ + 171, + 103 + ], + [ + 210, + 116 + ], + [ + 162, + 123 + ], + [ + 201, + 134 + ] + ] + ] + }, + { + "task_id": "0000496", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122063633#1159", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000496/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 228, + 76 + ], + [ + 226, + 67 + ], + [ + 218, + 62 + ], + [ + 239, + 91 + ], + [ + 220, + 71 + ], + [ + 240, + 99 + ] + ], + [ + [ + 207, + 90 + ], + [ + 198, + 82 + ], + [ + 222, + 77 + ], + [ + 192, + 103 + ], + [ + 232, + 85 + ], + [ + 202, + 112 + ] + ], + [ + [ + 227, + 94 + ], + [ + 229, + 75 + ], + [ + 229, + 95 + ], + [ + 225, + 92 + ], + [ + 227, + 113 + ], + [ + 224, + 112 + ] + ], + [ + [ + 211, + 80 + ], + [ + 208, + 97 + ], + [ + 214, + 73 + ], + [ + 206, + 88 + ], + [ + 218, + 56 + ], + [ + 210, + 70 + ] + ] + ] + }, + { + "task_id": "0000497", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122063633#1565", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000497/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 132, + 313 + ], + [ + 128, + 293 + ], + [ + 139, + 288 + ], + [ + 124, + 340 + ], + [ + 142, + 306 + ], + [ + 128, + 358 + ] + ], + [ + [ + 165, + 308 + ], + [ + 185, + 288 + ], + [ + 165, + 326 + ], + [ + 165, + 289 + ], + [ + 145, + 348 + ], + [ + 143, + 311 + ] + ], + [ + [ + 152, + 284 + ], + [ + 176, + 303 + ], + [ + 162, + 283 + ], + [ + 140, + 286 + ], + [ + 140, + 266 + ], + [ + 115, + 267 + ] + ], + [ + [ + 129, + 315 + ], + [ + 112, + 291 + ], + [ + 101, + 336 + ], + [ + 155, + 296 + ], + [ + 119, + 360 + ], + [ + 171, + 319 + ] + ] + ] + }, + { + "task_id": "0000500", + "id": "RH20T_cfg3_task_0059_user_0010_scene_0004_cfg_0003_104422070011#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000500/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 364, + 251 + ], + [ + 347, + 235 + ], + [ + 381, + 228 + ], + [ + 346, + 274 + ], + [ + 396, + 244 + ], + [ + 362, + 288 + ] + ], + [ + [ + 348, + 227 + ], + [ + 329, + 239 + ], + [ + 365, + 225 + ], + [ + 332, + 229 + ], + [ + 382, + 214 + ], + [ + 349, + 219 + ] + ], + [ + [ + 359, + 244 + ], + [ + 373, + 242 + ], + [ + 338, + 260 + ], + [ + 378, + 230 + ], + [ + 327, + 261 + ], + [ + 365, + 233 + ] + ], + [ + [ + 374, + 239 + ], + [ + 399, + 248 + ], + [ + 365, + 236 + ], + [ + 385, + 242 + ], + [ + 342, + 228 + ], + [ + 358, + 233 + ] + ] + ] + }, + { + "task_id": "0000502", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_039422060546#66", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000502/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 327, + 153 + ], + [ + 326, + 133 + ], + [ + 332, + 178 + ], + [ + 322, + 131 + ], + [ + 332, + 195 + ], + [ + 323, + 149 + ] + ], + [ + [ + 350, + 162 + ], + [ + 380, + 159 + ], + [ + 350, + 168 + ], + [ + 350, + 156 + ], + [ + 322, + 170 + ], + [ + 319, + 159 + ] + ], + [ + [ + 354, + 155 + ], + [ + 365, + 167 + ], + [ + 328, + 146 + ], + [ + 381, + 165 + ], + [ + 315, + 132 + ], + [ + 370, + 151 + ] + ], + [ + [ + 311, + 156 + ], + [ + 306, + 145 + ], + [ + 319, + 127 + ], + [ + 303, + 184 + ], + [ + 326, + 138 + ], + [ + 308, + 199 + ] + ] + ] + }, + { + "task_id": "0000503", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_039422060546#143", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000503/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 355, + 129 + ], + [ + 341, + 155 + ], + [ + 371, + 138 + ], + [ + 340, + 121 + ], + [ + 386, + 110 + ], + [ + 354, + 96 + ] + ], + [ + [ + 355, + 119 + ], + [ + 328, + 135 + ], + [ + 354, + 129 + ], + [ + 356, + 108 + ], + [ + 379, + 114 + ], + [ + 384, + 91 + ] + ], + [ + [ + 377, + 161 + ], + [ + 360, + 178 + ], + [ + 364, + 170 + ], + [ + 390, + 153 + ], + [ + 383, + 152 + ], + [ + 408, + 136 + ] + ], + [ + [ + 355, + 138 + ], + [ + 356, + 119 + ], + [ + 347, + 162 + ], + [ + 362, + 115 + ], + [ + 346, + 178 + ], + [ + 360, + 133 + ] + ] + ] + }, + { + "task_id": "0000504", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_039422060546#235", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000504/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 385, + 114 + ], + [ + 386, + 93 + ], + [ + 382, + 136 + ], + [ + 388, + 94 + ], + [ + 381, + 154 + ], + [ + 386, + 113 + ] + ], + [ + [ + 386, + 132 + ], + [ + 369, + 156 + ], + [ + 410, + 149 + ], + [ + 362, + 115 + ], + [ + 428, + 124 + ], + [ + 380, + 91 + ] + ], + [ + [ + 413, + 113 + ], + [ + 385, + 109 + ], + [ + 421, + 98 + ], + [ + 405, + 127 + ], + [ + 450, + 102 + ], + [ + 431, + 131 + ] + ], + [ + [ + 374, + 123 + ], + [ + 349, + 118 + ], + [ + 383, + 99 + ], + [ + 366, + 147 + ], + [ + 409, + 103 + ], + [ + 393, + 153 + ] + ] + ] + }, + { + "task_id": "0000505", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_039422060546#373", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000505/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 355, + 158 + ], + [ + 330, + 171 + ], + [ + 348, + 141 + ], + [ + 362, + 174 + ], + [ + 374, + 127 + ], + [ + 386, + 161 + ] + ], + [ + [ + 379, + 139 + ], + [ + 380, + 116 + ], + [ + 409, + 141 + ], + [ + 350, + 137 + ], + [ + 407, + 162 + ], + [ + 350, + 157 + ] + ], + [ + [ + 385, + 146 + ], + [ + 382, + 127 + ], + [ + 401, + 123 + ], + [ + 371, + 167 + ], + [ + 405, + 143 + ], + [ + 373, + 187 + ] + ], + [ + [ + 383, + 119 + ], + [ + 393, + 91 + ], + [ + 356, + 114 + ], + [ + 413, + 124 + ], + [ + 346, + 141 + ], + [ + 403, + 153 + ] + ] + ] + }, + { + "task_id": "0000507", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061330#326", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000507/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 337, + 183 + ], + [ + 314, + 166 + ], + [ + 332, + 192 + ], + [ + 342, + 172 + ], + [ + 354, + 208 + ], + [ + 366, + 189 + ] + ], + [ + [ + 302, + 203 + ], + [ + 309, + 218 + ], + [ + 322, + 185 + ], + [ + 282, + 221 + ], + [ + 314, + 172 + ], + [ + 277, + 207 + ] + ], + [ + [ + 356, + 200 + ], + [ + 341, + 203 + ], + [ + 344, + 176 + ], + [ + 367, + 224 + ], + [ + 360, + 172 + ], + [ + 383, + 222 + ] + ], + [ + [ + 334, + 193 + ], + [ + 327, + 198 + ], + [ + 324, + 217 + ], + [ + 344, + 168 + ], + [ + 331, + 211 + ], + [ + 349, + 165 + ] + ] + ] + }, + { + "task_id": "0000510", + "id": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_038522062547#166", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000510/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 450, + 131 + ], + [ + 429, + 114 + ], + [ + 468, + 110 + ], + [ + 431, + 151 + ], + [ + 490, + 128 + ], + [ + 452, + 169 + ] + ], + [ + [ + 424, + 162 + ], + [ + 403, + 163 + ], + [ + 410, + 152 + ], + [ + 437, + 172 + ], + [ + 432, + 151 + ], + [ + 459, + 171 + ] + ], + [ + [ + 449, + 146 + ], + [ + 455, + 140 + ], + [ + 454, + 171 + ], + [ + 443, + 122 + ], + [ + 448, + 176 + ], + [ + 438, + 128 + ] + ], + [ + [ + 428, + 164 + ], + [ + 434, + 158 + ], + [ + 417, + 184 + ], + [ + 440, + 143 + ], + [ + 409, + 191 + ], + [ + 433, + 149 + ] + ] + ] + }, + { + "task_id": "0000511", + "id": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_038522062547#229", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000511/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 411, + 145 + ], + [ + 415, + 136 + ], + [ + 417, + 169 + ], + [ + 406, + 121 + ], + [ + 414, + 176 + ], + [ + 403, + 131 + ] + ], + [ + [ + 444, + 159 + ], + [ + 459, + 169 + ], + [ + 446, + 183 + ], + [ + 442, + 135 + ], + [ + 433, + 173 + ], + [ + 429, + 127 + ] + ], + [ + [ + 418, + 161 + ], + [ + 434, + 151 + ], + [ + 419, + 140 + ], + [ + 418, + 184 + ], + [ + 405, + 150 + ], + [ + 404, + 192 + ] + ], + [ + [ + 391, + 139 + ], + [ + 375, + 156 + ], + [ + 407, + 149 + ], + [ + 375, + 129 + ], + [ + 424, + 133 + ], + [ + 392, + 111 + ] + ] + ] + }, + { + "task_id": "0000513", + "id": "RH20T_cfg3_task_0007_user_0016_scene_0006_cfg_0003_036422060909#250", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000513/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 291, + 108 + ], + [ + 295, + 121 + ], + [ + 281, + 126 + ], + [ + 300, + 91 + ], + [ + 276, + 113 + ], + [ + 296, + 76 + ] + ], + [ + [ + 307, + 120 + ], + [ + 295, + 101 + ], + [ + 319, + 112 + ], + [ + 295, + 129 + ], + [ + 331, + 131 + ], + [ + 306, + 147 + ] + ], + [ + [ + 324, + 100 + ], + [ + 330, + 77 + ], + [ + 303, + 92 + ], + [ + 343, + 108 + ], + [ + 297, + 115 + ], + [ + 337, + 130 + ] + ], + [ + [ + 301, + 111 + ], + [ + 304, + 95 + ], + [ + 316, + 99 + ], + [ + 285, + 123 + ], + [ + 313, + 113 + ], + [ + 283, + 137 + ] + ] + ] + }, + { + "task_id": "0000518", + "id": "RH20T_cfg7_task_0034_user_0014_scene_0001_cfg_0007_104122063633#310", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000518/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 170, + 135 + ], + [ + 169, + 156 + ], + [ + 152, + 136 + ], + [ + 188, + 134 + ], + [ + 153, + 115 + ], + [ + 188, + 112 + ] + ], + [ + [ + 193, + 174 + ], + [ + 214, + 175 + ], + [ + 196, + 181 + ], + [ + 189, + 166 + ], + [ + 176, + 179 + ], + [ + 167, + 164 + ] + ], + [ + [ + 191, + 140 + ], + [ + 173, + 134 + ], + [ + 203, + 136 + ], + [ + 179, + 145 + ], + [ + 221, + 142 + ], + [ + 199, + 152 + ] + ], + [ + [ + 175, + 150 + ], + [ + 179, + 131 + ], + [ + 178, + 159 + ], + [ + 171, + 141 + ], + [ + 173, + 178 + ], + [ + 166, + 161 + ] + ] + ] + }, + { + "task_id": "0000520", + "id": "RH20T_cfg7_task_0034_user_0014_scene_0001_cfg_0007_104122063633#517", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000520/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 196, + 130 + ], + [ + 200, + 113 + ], + [ + 200, + 141 + ], + [ + 192, + 119 + ], + [ + 196, + 158 + ], + [ + 188, + 137 + ] + ], + [ + [ + 182, + 116 + ], + [ + 166, + 118 + ], + [ + 171, + 102 + ], + [ + 192, + 130 + ], + [ + 187, + 100 + ], + [ + 208, + 129 + ] + ], + [ + [ + 203, + 130 + ], + [ + 222, + 124 + ], + [ + 197, + 113 + ], + [ + 207, + 146 + ], + [ + 177, + 119 + ], + [ + 188, + 153 + ] + ], + [ + [ + 198, + 122 + ], + [ + 209, + 128 + ], + [ + 209, + 106 + ], + [ + 187, + 139 + ], + [ + 199, + 101 + ], + [ + 177, + 133 + ] + ] + ] + }, + { + "task_id": "0000522", + "id": "RH20T_cfg4_task_0066_user_0015_scene_0007_cfg_0004_104122062295#134", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000522/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 351, + 307 + ], + [ + 365, + 271 + ], + [ + 315, + 274 + ], + [ + 390, + 344 + ], + [ + 305, + 306 + ], + [ + 374, + 360 + ] + ], + [ + [ + 399, + 315 + ], + [ + 370, + 332 + ], + [ + 388, + 359 + ], + [ + 410, + 268 + ], + [ + 413, + 341 + ], + [ + 434, + 256 + ] + ], + [ + [ + 289, + 327 + ], + [ + 250, + 357 + ], + [ + 313, + 319 + ], + [ + 269, + 335 + ], + [ + 349, + 292 + ], + [ + 302, + 310 + ] + ], + [ + [ + 339, + 296 + ], + [ + 350, + 329 + ], + [ + 379, + 278 + ], + [ + 297, + 314 + ], + [ + 368, + 242 + ], + [ + 283, + 278 + ] + ] + ] + }, + { + "task_id": "0000523", + "id": "RH20T_cfg4_task_0066_user_0015_scene_0007_cfg_0004_104122062295#823", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000523/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 445, + 265 + ], + [ + 436, + 316 + ], + [ + 455, + 263 + ], + [ + 437, + 266 + ], + [ + 464, + 208 + ], + [ + 446, + 221 + ] + ], + [ + [ + 416, + 279 + ], + [ + 434, + 256 + ], + [ + 456, + 275 + ], + [ + 373, + 283 + ], + [ + 438, + 295 + ], + [ + 359, + 303 + ] + ], + [ + [ + 476, + 307 + ], + [ + 444, + 268 + ], + [ + 523, + 285 + ], + [ + 434, + 326 + ], + [ + 556, + 325 + ], + [ + 465, + 360 + ] + ], + [ + [ + 441, + 294 + ], + [ + 462, + 257 + ], + [ + 401, + 263 + ], + [ + 486, + 328 + ], + [ + 385, + 295 + ], + [ + 463, + 360 + ] + ] + ] + }, + { + "task_id": "0000524", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0006_cfg_0004_104122062823#144", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000524/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 281, + 82 + ], + [ + 285, + 77 + ], + [ + 259, + 97 + ], + [ + 305, + 66 + ], + [ + 254, + 104 + ], + [ + 302, + 72 + ] + ], + [ + [ + 281, + 74 + ], + [ + 273, + 84 + ], + [ + 256, + 56 + ], + [ + 306, + 92 + ], + [ + 264, + 44 + ], + [ + 317, + 83 + ] + ], + [ + [ + 255, + 74 + ], + [ + 259, + 103 + ], + [ + 278, + 67 + ], + [ + 231, + 81 + ], + [ + 275, + 37 + ], + [ + 225, + 49 + ] + ], + [ + [ + 282, + 87 + ], + [ + 283, + 63 + ], + [ + 312, + 89 + ], + [ + 252, + 85 + ], + [ + 310, + 111 + ], + [ + 253, + 107 + ] + ] + ] + }, + { + "task_id": "0000525", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0006_cfg_0004_104122062823#217", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000525/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 77 + ], + [ + 266, + 50 + ], + [ + 293, + 85 + ], + [ + 241, + 70 + ], + [ + 293, + 109 + ], + [ + 242, + 94 + ] + ], + [ + [ + 264, + 78 + ], + [ + 277, + 69 + ], + [ + 260, + 50 + ], + [ + 268, + 105 + ], + [ + 244, + 58 + ], + [ + 254, + 115 + ] + ], + [ + [ + 266, + 75 + ], + [ + 250, + 98 + ], + [ + 258, + 59 + ], + [ + 274, + 93 + ], + [ + 274, + 35 + ], + [ + 293, + 68 + ] + ], + [ + [ + 294, + 87 + ], + [ + 279, + 80 + ], + [ + 316, + 86 + ], + [ + 269, + 87 + ], + [ + 333, + 93 + ], + [ + 285, + 95 + ] + ] + ] + }, + { + "task_id": "0000526", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0006_cfg_0004_104122062823#287", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000526/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 129 + ], + [ + 276, + 114 + ], + [ + 238, + 133 + ], + [ + 292, + 126 + ], + [ + 230, + 147 + ], + [ + 282, + 139 + ] + ], + [ + [ + 230, + 105 + ], + [ + 244, + 120 + ], + [ + 251, + 86 + ], + [ + 208, + 124 + ], + [ + 236, + 69 + ], + [ + 191, + 109 + ] + ], + [ + [ + 235, + 95 + ], + [ + 234, + 118 + ], + [ + 264, + 82 + ], + [ + 208, + 107 + ], + [ + 264, + 61 + ], + [ + 211, + 86 + ] + ], + [ + [ + 257, + 98 + ], + [ + 253, + 73 + ], + [ + 288, + 92 + ], + [ + 227, + 104 + ], + [ + 290, + 116 + ], + [ + 233, + 127 + ] + ] + ] + }, + { + "task_id": "0000529", + "id": "RH20T_cfg5_task_0001_user_0010_scene_0003_cfg_0005_104122061850#281", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000529/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 144 + ], + [ + 394, + 159 + ], + [ + 363, + 152 + ], + [ + 380, + 134 + ], + [ + 341, + 137 + ], + [ + 356, + 118 + ] + ], + [ + [ + 375, + 144 + ], + [ + 375, + 129 + ], + [ + 366, + 167 + ], + [ + 384, + 122 + ], + [ + 366, + 180 + ], + [ + 384, + 136 + ] + ], + [ + [ + 381, + 160 + ], + [ + 383, + 139 + ], + [ + 375, + 179 + ], + [ + 387, + 139 + ], + [ + 372, + 201 + ], + [ + 384, + 161 + ] + ], + [ + [ + 361, + 174 + ], + [ + 386, + 183 + ], + [ + 357, + 161 + ], + [ + 365, + 185 + ], + [ + 332, + 151 + ], + [ + 342, + 176 + ] + ] + ] + }, + { + "task_id": "0000532", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0005_cfg_0005_105422061350#301", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000532/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 252, + 98 + ], + [ + 263, + 102 + ], + [ + 263, + 72 + ], + [ + 240, + 124 + ], + [ + 252, + 70 + ], + [ + 230, + 119 + ] + ], + [ + [ + 263, + 114 + ], + [ + 234, + 123 + ], + [ + 274, + 141 + ], + [ + 253, + 87 + ], + [ + 303, + 131 + ], + [ + 281, + 78 + ] + ], + [ + [ + 237, + 87 + ], + [ + 209, + 100 + ], + [ + 227, + 60 + ], + [ + 248, + 116 + ], + [ + 255, + 47 + ], + [ + 277, + 103 + ] + ], + [ + [ + 250, + 92 + ], + [ + 249, + 70 + ], + [ + 262, + 72 + ], + [ + 238, + 114 + ], + [ + 262, + 93 + ], + [ + 239, + 133 + ] + ] + ] + }, + { + "task_id": "0000534", + "id": "RH20T_cfg1_task_0051_user_0015_scene_0002_cfg_0001_750612070853#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the dish off the dish rack'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000534/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 264, + 79 + ], + [ + 229, + 90 + ], + [ + 241, + 105 + ], + [ + 291, + 50 + ], + [ + 270, + 94 + ], + [ + 321, + 42 + ] + ], + [ + [ + 309, + 101 + ], + [ + 284, + 62 + ], + [ + 269, + 122 + ], + [ + 350, + 80 + ], + [ + 293, + 156 + ], + [ + 371, + 117 + ] + ], + [ + [ + 294, + 59 + ], + [ + 292, + 101 + ], + [ + 247, + 50 + ], + [ + 338, + 67 + ], + [ + 251, + 9 + ], + [ + 339, + 27 + ] + ], + [ + [ + 322, + 59 + ], + [ + 310, + 56 + ], + [ + 318, + 106 + ], + [ + 326, + 10 + ], + [ + 331, + 114 + ], + [ + 340, + 8 + ] + ] + ] + }, + { + "task_id": "0000535", + "id": "RH20T_cfg1_task_0059_user_0001_scene_0001_cfg_0001_038522062547#185", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000535/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 369, + 218 + ], + [ + 369, + 205 + ], + [ + 398, + 222 + ], + [ + 340, + 215 + ], + [ + 396, + 233 + ], + [ + 341, + 227 + ] + ], + [ + [ + 396, + 204 + ], + [ + 392, + 232 + ], + [ + 388, + 214 + ], + [ + 404, + 192 + ], + [ + 391, + 189 + ], + [ + 407, + 164 + ] + ], + [ + [ + 381, + 221 + ], + [ + 406, + 230 + ], + [ + 395, + 201 + ], + [ + 367, + 243 + ], + [ + 372, + 193 + ], + [ + 344, + 234 + ] + ], + [ + [ + 379, + 212 + ], + [ + 396, + 223 + ], + [ + 365, + 237 + ], + [ + 392, + 187 + ], + [ + 351, + 225 + ], + [ + 376, + 178 + ] + ] + ] + }, + { + "task_id": "0000539", + "id": "RH20T_cfg6_task_0046_user_0020_scene_0010_cfg_0006_104122060811#316", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000539/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 369, + 114 + ], + [ + 381, + 115 + ], + [ + 363, + 125 + ], + [ + 375, + 104 + ], + [ + 350, + 124 + ], + [ + 362, + 103 + ] + ], + [ + [ + 377, + 114 + ], + [ + 377, + 121 + ], + [ + 383, + 102 + ], + [ + 372, + 127 + ], + [ + 384, + 94 + ], + [ + 372, + 120 + ] + ], + [ + [ + 381, + 108 + ], + [ + 389, + 96 + ], + [ + 392, + 117 + ], + [ + 369, + 100 + ], + [ + 385, + 129 + ], + [ + 361, + 112 + ] + ], + [ + [ + 375, + 106 + ], + [ + 382, + 118 + ], + [ + 387, + 98 + ], + [ + 363, + 114 + ], + [ + 379, + 85 + ], + [ + 355, + 102 + ] + ] + ] + }, + { + "task_id": "0000540", + "id": "RH20T_cfg4_task_0081_user_0007_scene_0004_cfg_0004_038522062288#164", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000540/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 207, + 184 + ], + [ + 218, + 163 + ], + [ + 182, + 168 + ], + [ + 231, + 199 + ], + [ + 170, + 191 + ], + [ + 221, + 222 + ] + ], + [ + [ + 168, + 174 + ], + [ + 184, + 155 + ], + [ + 144, + 169 + ], + [ + 191, + 179 + ], + [ + 128, + 189 + ], + [ + 175, + 197 + ] + ], + [ + [ + 176, + 190 + ], + [ + 173, + 173 + ], + [ + 148, + 180 + ], + [ + 205, + 200 + ], + [ + 152, + 196 + ], + [ + 206, + 215 + ] + ], + [ + [ + 162, + 182 + ], + [ + 146, + 205 + ], + [ + 166, + 199 + ], + [ + 157, + 164 + ], + [ + 180, + 179 + ], + [ + 172, + 142 + ] + ] + ] + }, + { + "task_id": "0000541", + "id": "RH20T_cfg4_task_0081_user_0007_scene_0004_cfg_0004_038522062288#323", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000541/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 241, + 167 + ], + [ + 251, + 194 + ], + [ + 232, + 173 + ], + [ + 252, + 161 + ], + [ + 223, + 148 + ], + [ + 241, + 133 + ] + ], + [ + [ + 243, + 187 + ], + [ + 244, + 169 + ], + [ + 216, + 198 + ], + [ + 269, + 177 + ], + [ + 217, + 214 + ], + [ + 267, + 193 + ] + ], + [ + [ + 243, + 203 + ], + [ + 252, + 210 + ], + [ + 215, + 217 + ], + [ + 271, + 190 + ], + [ + 203, + 210 + ], + [ + 263, + 182 + ] + ], + [ + [ + 255, + 217 + ], + [ + 281, + 227 + ], + [ + 266, + 191 + ], + [ + 244, + 242 + ], + [ + 239, + 181 + ], + [ + 217, + 232 + ] + ] + ] + }, + { + "task_id": "0000542", + "id": "RH20T_cfg4_task_0029_user_0016_scene_0009_cfg_0004_036422060909#54", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000542/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 430, + 63 + ], + [ + 421, + 43 + ], + [ + 460, + 50 + ], + [ + 401, + 76 + ], + [ + 470, + 71 + ], + [ + 410, + 97 + ] + ], + [ + [ + 424, + 102 + ], + [ + 434, + 84 + ], + [ + 405, + 82 + ], + [ + 444, + 123 + ], + [ + 397, + 99 + ], + [ + 434, + 139 + ] + ], + [ + [ + 401, + 110 + ], + [ + 379, + 103 + ], + [ + 406, + 80 + ], + [ + 396, + 140 + ], + [ + 426, + 88 + ], + [ + 416, + 145 + ] + ], + [ + [ + 406, + 114 + ], + [ + 429, + 102 + ], + [ + 390, + 96 + ], + [ + 424, + 134 + ], + [ + 367, + 108 + ], + [ + 400, + 148 + ] + ] + ] + }, + { + "task_id": "0000547", + "id": "RH20T_cfg4_task_0066_user_0015_scene_0001_cfg_0004_104122062823#144", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000547/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 243, + 58 + ], + [ + 239, + 31 + ], + [ + 267, + 68 + ], + [ + 221, + 49 + ], + [ + 270, + 93 + ], + [ + 226, + 73 + ] + ], + [ + [ + 256, + 71 + ], + [ + 273, + 55 + ], + [ + 276, + 92 + ], + [ + 235, + 49 + ], + [ + 260, + 106 + ], + [ + 220, + 65 + ] + ], + [ + [ + 255, + 49 + ], + [ + 269, + 33 + ], + [ + 230, + 33 + ], + [ + 280, + 66 + ], + [ + 215, + 49 + ], + [ + 266, + 84 + ] + ], + [ + [ + 235, + 62 + ], + [ + 260, + 45 + ], + [ + 221, + 36 + ], + [ + 249, + 85 + ], + [ + 194, + 54 + ], + [ + 225, + 102 + ] + ] + ] + }, + { + "task_id": "0000548", + "id": "RH20T_cfg3_task_0059_user_0010_scene_0004_cfg_0003_104122062295#181", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000548/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 248, + 209 + ], + [ + 282, + 214 + ], + [ + 268, + 246 + ], + [ + 226, + 168 + ], + [ + 238, + 239 + ], + [ + 197, + 167 + ] + ], + [ + [ + 215, + 270 + ], + [ + 191, + 219 + ], + [ + 208, + 276 + ], + [ + 220, + 265 + ], + [ + 235, + 334 + ], + [ + 241, + 310 + ] + ], + [ + [ + 231, + 273 + ], + [ + 219, + 243 + ], + [ + 207, + 244 + ], + [ + 259, + 307 + ], + [ + 219, + 271 + ], + [ + 269, + 331 + ] + ], + [ + [ + 176, + 258 + ], + [ + 138, + 254 + ], + [ + 209, + 240 + ], + [ + 135, + 279 + ], + [ + 249, + 243 + ], + [ + 180, + 286 + ] + ] + ] + }, + { + "task_id": "0000550", + "id": "RH20T_cfg6_task_0046_user_0020_scene_0010_cfg_0006_037522061512#308", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000550/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 234, + 138 + ], + [ + 227, + 157 + ], + [ + 215, + 139 + ], + [ + 252, + 137 + ], + [ + 222, + 118 + ], + [ + 260, + 117 + ] + ], + [ + [ + 223, + 115 + ], + [ + 238, + 108 + ], + [ + 226, + 136 + ], + [ + 220, + 92 + ], + [ + 210, + 144 + ], + [ + 203, + 98 + ] + ], + [ + [ + 202, + 99 + ], + [ + 215, + 118 + ], + [ + 205, + 86 + ], + [ + 199, + 111 + ], + [ + 192, + 66 + ], + [ + 187, + 93 + ] + ], + [ + [ + 234, + 103 + ], + [ + 228, + 124 + ], + [ + 213, + 101 + ], + [ + 255, + 104 + ], + [ + 219, + 81 + ], + [ + 260, + 83 + ] + ] + ] + }, + { + "task_id": "0000554", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_105422061350#236", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000554/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 127 + ], + [ + 258, + 107 + ], + [ + 256, + 149 + ], + [ + 292, + 103 + ], + [ + 272, + 170 + ], + [ + 309, + 123 + ] + ], + [ + [ + 281, + 119 + ], + [ + 278, + 142 + ], + [ + 308, + 114 + ], + [ + 256, + 124 + ], + [ + 311, + 93 + ], + [ + 260, + 104 + ] + ], + [ + [ + 235, + 129 + ], + [ + 258, + 137 + ], + [ + 232, + 156 + ], + [ + 238, + 103 + ], + [ + 206, + 148 + ], + [ + 213, + 93 + ] + ], + [ + [ + 258, + 121 + ], + [ + 262, + 103 + ], + [ + 286, + 122 + ], + [ + 229, + 120 + ], + [ + 281, + 139 + ], + [ + 226, + 137 + ] + ] + ] + }, + { + "task_id": "0000555", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_105422061350#457", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000555/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 237, + 110 + ], + [ + 223, + 106 + ], + [ + 261, + 105 + ], + [ + 214, + 114 + ], + [ + 273, + 108 + ], + [ + 227, + 117 + ] + ], + [ + [ + 242, + 84 + ], + [ + 226, + 107 + ], + [ + 236, + 83 + ], + [ + 247, + 85 + ], + [ + 254, + 58 + ], + [ + 263, + 63 + ] + ], + [ + [ + 219, + 98 + ], + [ + 215, + 110 + ], + [ + 238, + 81 + ], + [ + 200, + 114 + ], + [ + 240, + 71 + ], + [ + 204, + 103 + ] + ], + [ + [ + 253, + 102 + ], + [ + 258, + 84 + ], + [ + 279, + 98 + ], + [ + 226, + 106 + ], + [ + 273, + 114 + ], + [ + 223, + 123 + ] + ] + ] + }, + { + "task_id": "0000559", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0005_cfg_0003_104122062823#163", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000559/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 215, + 76 + ], + [ + 203, + 73 + ], + [ + 214, + 48 + ], + [ + 216, + 103 + ], + [ + 228, + 51 + ], + [ + 229, + 108 + ] + ], + [ + [ + 232, + 68 + ], + [ + 231, + 41 + ], + [ + 262, + 67 + ], + [ + 203, + 68 + ], + [ + 262, + 91 + ], + [ + 205, + 93 + ] + ], + [ + [ + 261, + 45 + ], + [ + 280, + 28 + ], + [ + 276, + 69 + ], + [ + 245, + 20 + ], + [ + 256, + 88 + ], + [ + 223, + 39 + ] + ], + [ + [ + 218, + 59 + ], + [ + 237, + 40 + ], + [ + 193, + 41 + ], + [ + 240, + 76 + ], + [ + 175, + 61 + ], + [ + 222, + 93 + ] + ] + ] + }, + { + "task_id": "0000560", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0005_cfg_0003_104122062823#202", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000560/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 252, + 74 + ], + [ + 248, + 55 + ], + [ + 261, + 92 + ], + [ + 241, + 54 + ], + [ + 265, + 112 + ], + [ + 245, + 75 + ] + ], + [ + [ + 236, + 26 + ], + [ + 222, + 41 + ], + [ + 257, + 44 + ], + [ + 214, + 6 + ], + [ + 274, + 29 + ], + [ + 230, + 0 + ] + ], + [ + [ + 231, + 56 + ], + [ + 230, + 29 + ], + [ + 259, + 56 + ], + [ + 203, + 56 + ], + [ + 259, + 81 + ], + [ + 205, + 81 + ] + ], + [ + [ + 226, + 42 + ], + [ + 247, + 45 + ], + [ + 203, + 31 + ], + [ + 246, + 52 + ], + [ + 183, + 29 + ], + [ + 227, + 49 + ] + ] + ] + }, + { + "task_id": "0000561", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0005_cfg_0003_104122062823#225", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000561/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 221, + 34 + ], + [ + 221, + 41 + ], + [ + 193, + 21 + ], + [ + 247, + 46 + ], + [ + 191, + 12 + ], + [ + 249, + 39 + ] + ], + [ + [ + 228, + 45 + ], + [ + 227, + 19 + ], + [ + 255, + 46 + ], + [ + 201, + 44 + ], + [ + 256, + 70 + ], + [ + 203, + 68 + ] + ], + [ + [ + 215, + 32 + ], + [ + 214, + 19 + ], + [ + 234, + 53 + ], + [ + 196, + 10 + ], + [ + 234, + 63 + ], + [ + 197, + 22 + ] + ], + [ + [ + 239, + 58 + ], + [ + 237, + 58 + ], + [ + 215, + 48 + ], + [ + 263, + 68 + ], + [ + 215, + 47 + ], + [ + 267, + 69 + ] + ] + ] + }, + { + "task_id": "0000563", + "id": "RH20T_cfg5_task_0048_user_0010_scene_0010_cfg_0005_104122061850#164", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000563/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 135 + ], + [ + 322, + 112 + ], + [ + 308, + 150 + ], + [ + 367, + 120 + ], + [ + 324, + 175 + ], + [ + 385, + 144 + ] + ], + [ + [ + 338, + 120 + ], + [ + 323, + 96 + ], + [ + 324, + 144 + ], + [ + 350, + 97 + ], + [ + 338, + 165 + ], + [ + 363, + 119 + ] + ], + [ + [ + 337, + 82 + ], + [ + 345, + 83 + ], + [ + 353, + 110 + ], + [ + 319, + 54 + ], + [ + 345, + 107 + ], + [ + 313, + 55 + ] + ], + [ + [ + 354, + 79 + ], + [ + 379, + 83 + ], + [ + 350, + 47 + ], + [ + 358, + 109 + ], + [ + 326, + 45 + ], + [ + 336, + 105 + ] + ] + ] + }, + { + "task_id": "0000564", + "id": "RH20T_cfg5_task_0048_user_0010_scene_0010_cfg_0005_104122061850#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000564/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 344, + 126 + ], + [ + 349, + 99 + ], + [ + 363, + 120 + ], + [ + 328, + 132 + ], + [ + 358, + 150 + ], + [ + 323, + 159 + ] + ], + [ + [ + 331, + 127 + ], + [ + 314, + 110 + ], + [ + 312, + 151 + ], + [ + 350, + 102 + ], + [ + 329, + 165 + ], + [ + 365, + 119 + ] + ], + [ + [ + 352, + 149 + ], + [ + 335, + 130 + ], + [ + 343, + 173 + ], + [ + 362, + 124 + ], + [ + 359, + 194 + ], + [ + 381, + 145 + ] + ], + [ + [ + 349, + 128 + ], + [ + 368, + 141 + ], + [ + 367, + 102 + ], + [ + 331, + 155 + ], + [ + 349, + 91 + ], + [ + 315, + 142 + ] + ] + ] + }, + { + "task_id": "0000565", + "id": "RH20T_cfg5_task_0048_user_0010_scene_0010_cfg_0005_104122061850#278", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000565/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 177 + ], + [ + 291, + 192 + ], + [ + 265, + 198 + ], + [ + 303, + 157 + ], + [ + 256, + 183 + ], + [ + 296, + 139 + ] + ], + [ + [ + 265, + 165 + ], + [ + 282, + 168 + ], + [ + 288, + 166 + ], + [ + 243, + 164 + ], + [ + 269, + 163 + ], + [ + 224, + 161 + ] + ], + [ + [ + 294, + 148 + ], + [ + 284, + 123 + ], + [ + 301, + 133 + ], + [ + 287, + 162 + ], + [ + 312, + 160 + ], + [ + 297, + 187 + ] + ], + [ + [ + 293, + 150 + ], + [ + 275, + 135 + ], + [ + 275, + 174 + ], + [ + 311, + 126 + ], + [ + 292, + 186 + ], + [ + 327, + 140 + ] + ] + ] + }, + { + "task_id": "0000566", + "id": "RH20T_cfg5_task_0048_user_0010_scene_0010_cfg_0005_104122061850#414", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000566/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 318, + 85 + ], + [ + 328, + 66 + ], + [ + 319, + 104 + ], + [ + 316, + 65 + ], + [ + 309, + 124 + ], + [ + 305, + 85 + ] + ], + [ + [ + 326, + 98 + ], + [ + 310, + 88 + ], + [ + 308, + 117 + ], + [ + 345, + 79 + ], + [ + 323, + 125 + ], + [ + 359, + 89 + ] + ], + [ + [ + 328, + 68 + ], + [ + 323, + 62 + ], + [ + 302, + 67 + ], + [ + 353, + 69 + ], + [ + 306, + 74 + ], + [ + 360, + 76 + ] + ], + [ + [ + 348, + 109 + ], + [ + 359, + 134 + ], + [ + 325, + 120 + ], + [ + 371, + 97 + ], + [ + 314, + 96 + ], + [ + 360, + 71 + ] + ] + ] + }, + { + "task_id": "0000567", + "id": "RH20T_cfg3_task_0073_user_0010_scene_0010_cfg_0003_104422070011#136", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000567/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 186, + 107 + ], + [ + 192, + 115 + ], + [ + 211, + 122 + ], + [ + 161, + 93 + ], + [ + 206, + 114 + ], + [ + 153, + 83 + ] + ], + [ + [ + 183, + 141 + ], + [ + 184, + 127 + ], + [ + 155, + 150 + ], + [ + 209, + 132 + ], + [ + 155, + 163 + ], + [ + 207, + 145 + ] + ], + [ + [ + 178, + 162 + ], + [ + 178, + 176 + ], + [ + 164, + 186 + ], + [ + 190, + 140 + ], + [ + 164, + 171 + ], + [ + 191, + 124 + ] + ], + [ + [ + 176, + 135 + ], + [ + 178, + 151 + ], + [ + 146, + 136 + ], + [ + 204, + 135 + ], + [ + 142, + 119 + ], + [ + 203, + 118 + ] + ] + ] + }, + { + "task_id": "0000568", + "id": "RH20T_cfg3_task_0073_user_0010_scene_0010_cfg_0003_104422070011#207", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000568/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 303, + 155 + ], + [ + 314, + 139 + ], + [ + 278, + 147 + ], + [ + 326, + 162 + ], + [ + 269, + 163 + ], + [ + 315, + 176 + ] + ], + [ + [ + 308, + 181 + ], + [ + 309, + 173 + ], + [ + 293, + 204 + ], + [ + 322, + 159 + ], + [ + 292, + 209 + ], + [ + 320, + 167 + ] + ], + [ + [ + 302, + 166 + ], + [ + 277, + 151 + ], + [ + 314, + 149 + ], + [ + 289, + 185 + ], + [ + 337, + 164 + ], + [ + 314, + 201 + ] + ], + [ + [ + 293, + 195 + ], + [ + 317, + 202 + ], + [ + 288, + 220 + ], + [ + 298, + 169 + ], + [ + 264, + 213 + ], + [ + 274, + 162 + ] + ] + ] + }, + { + "task_id": "0000569", + "id": "RH20T_cfg5_task_0075_user_0007_scene_0004_cfg_0005_104122063678#286", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Finish setting up the starting position of a chessboard that is almost arranged'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000569/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 262, + 201 + ], + [ + 265, + 185 + ], + [ + 258, + 221 + ], + [ + 266, + 183 + ], + [ + 256, + 235 + ], + [ + 264, + 198 + ] + ], + [ + [ + 266, + 173 + ], + [ + 264, + 198 + ], + [ + 243, + 170 + ], + [ + 288, + 176 + ], + [ + 245, + 145 + ], + [ + 289, + 152 + ] + ], + [ + [ + 256, + 208 + ], + [ + 245, + 218 + ], + [ + 256, + 231 + ], + [ + 257, + 186 + ], + [ + 269, + 220 + ], + [ + 269, + 174 + ] + ], + [ + [ + 248, + 197 + ], + [ + 270, + 199 + ], + [ + 260, + 190 + ], + [ + 235, + 205 + ], + [ + 240, + 188 + ], + [ + 214, + 203 + ] + ] + ] + }, + { + "task_id": "0000571", + "id": "RH20T_cfg1_task_0077_user_0010_scene_0006_cfg_0001_819112070044#180", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000571/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 356, + 169 + ], + [ + 366, + 180 + ], + [ + 345, + 183 + ], + [ + 367, + 154 + ], + [ + 335, + 172 + ], + [ + 357, + 143 + ] + ], + [ + [ + 351, + 188 + ], + [ + 340, + 176 + ], + [ + 358, + 193 + ], + [ + 343, + 184 + ], + [ + 369, + 206 + ], + [ + 355, + 197 + ] + ], + [ + [ + 341, + 175 + ], + [ + 329, + 172 + ], + [ + 337, + 194 + ], + [ + 345, + 156 + ], + [ + 350, + 198 + ], + [ + 359, + 159 + ] + ], + [ + [ + 350, + 177 + ], + [ + 352, + 162 + ], + [ + 366, + 171 + ], + [ + 334, + 183 + ], + [ + 364, + 184 + ], + [ + 332, + 197 + ] + ] + ] + }, + { + "task_id": "0000572", + "id": "RH20T_cfg4_task_0055_user_0010_scene_0005_cfg_0004_104122062823#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000572/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 235, + 41 + ], + [ + 253, + 34 + ], + [ + 211, + 19 + ], + [ + 260, + 64 + ], + [ + 189, + 25 + ], + [ + 241, + 73 + ] + ], + [ + [ + 183, + 59 + ], + [ + 200, + 32 + ], + [ + 210, + 78 + ], + [ + 154, + 39 + ], + [ + 194, + 105 + ], + [ + 136, + 68 + ] + ], + [ + [ + 207, + 35 + ], + [ + 194, + 11 + ], + [ + 236, + 43 + ], + [ + 180, + 27 + ], + [ + 246, + 64 + ], + [ + 193, + 48 + ] + ], + [ + [ + 197, + 53 + ], + [ + 208, + 21 + ], + [ + 232, + 66 + ], + [ + 161, + 40 + ], + [ + 222, + 99 + ], + [ + 149, + 74 + ] + ] + ] + }, + { + "task_id": "0000576", + "id": "RH20T_cfg5_task_0033_user_0010_scene_0010_cfg_0005_104122063678#443", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000576/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 399, + 228 + ], + [ + 376, + 224 + ], + [ + 402, + 205 + ], + [ + 396, + 252 + ], + [ + 426, + 209 + ], + [ + 419, + 256 + ] + ], + [ + [ + 400, + 216 + ], + [ + 383, + 231 + ], + [ + 404, + 208 + ], + [ + 395, + 224 + ], + [ + 421, + 193 + ], + [ + 413, + 207 + ] + ], + [ + [ + 399, + 224 + ], + [ + 407, + 211 + ], + [ + 380, + 235 + ], + [ + 416, + 214 + ], + [ + 373, + 247 + ], + [ + 408, + 226 + ] + ], + [ + [ + 381, + 252 + ], + [ + 377, + 227 + ], + [ + 357, + 255 + ], + [ + 406, + 249 + ], + [ + 360, + 281 + ], + [ + 411, + 274 + ] + ] + ] + }, + { + "task_id": "0000578", + "id": "RH20T_cfg5_task_0033_user_0010_scene_0010_cfg_0005_104122063678#721", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000578/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 352, + 279 + ], + [ + 341, + 257 + ], + [ + 336, + 273 + ], + [ + 367, + 284 + ], + [ + 348, + 298 + ], + [ + 379, + 308 + ] + ], + [ + [ + 350, + 263 + ], + [ + 357, + 247 + ], + [ + 336, + 281 + ], + [ + 364, + 246 + ], + [ + 329, + 296 + ], + [ + 357, + 261 + ] + ], + [ + [ + 329, + 282 + ], + [ + 344, + 289 + ], + [ + 326, + 256 + ], + [ + 333, + 308 + ], + [ + 313, + 251 + ], + [ + 320, + 301 + ] + ], + [ + [ + 339, + 246 + ], + [ + 314, + 243 + ], + [ + 328, + 262 + ], + [ + 348, + 231 + ], + [ + 356, + 265 + ], + [ + 373, + 233 + ] + ] + ] + }, + { + "task_id": "0000580", + "id": "RH20T_cfg4_task_0047_user_0014_scene_0014_cfg_0004_038522062288#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000580/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 260, + 285 + ], + [ + 232, + 260 + ], + [ + 271, + 290 + ], + [ + 249, + 280 + ], + [ + 298, + 315 + ], + [ + 281, + 309 + ] + ], + [ + [ + 321, + 255 + ], + [ + 284, + 244 + ], + [ + 333, + 217 + ], + [ + 308, + 295 + ], + [ + 372, + 228 + ], + [ + 347, + 308 + ] + ], + [ + [ + 288, + 261 + ], + [ + 275, + 228 + ], + [ + 251, + 281 + ], + [ + 322, + 241 + ], + [ + 264, + 311 + ], + [ + 332, + 271 + ] + ], + [ + [ + 297, + 256 + ], + [ + 298, + 260 + ], + [ + 337, + 250 + ], + [ + 256, + 261 + ], + [ + 340, + 245 + ], + [ + 251, + 256 + ] + ] + ] + }, + { + "task_id": "0000581", + "id": "RH20T_cfg1_task_0065_user_0014_scene_0004_cfg_0001_819112070044#99", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000581/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 407, + 175 + ], + [ + 418, + 168 + ], + [ + 391, + 166 + ], + [ + 422, + 184 + ], + [ + 381, + 174 + ], + [ + 411, + 190 + ] + ], + [ + [ + 418, + 184 + ], + [ + 421, + 169 + ], + [ + 432, + 177 + ], + [ + 403, + 191 + ], + [ + 428, + 191 + ], + [ + 400, + 205 + ] + ], + [ + [ + 420, + 204 + ], + [ + 429, + 187 + ], + [ + 413, + 201 + ], + [ + 426, + 207 + ], + [ + 403, + 219 + ], + [ + 417, + 224 + ] + ], + [ + [ + 430, + 188 + ], + [ + 441, + 197 + ], + [ + 416, + 202 + ], + [ + 444, + 174 + ], + [ + 404, + 193 + ], + [ + 432, + 164 + ] + ] + ] + }, + { + "task_id": "0000582", + "id": "RH20T_cfg1_task_0033_user_0014_scene_0006_cfg_0001_035622060973#89", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000582/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 236, + 103 + ], + [ + 237, + 89 + ], + [ + 251, + 114 + ], + [ + 222, + 93 + ], + [ + 249, + 127 + ], + [ + 221, + 106 + ] + ], + [ + [ + 232, + 125 + ], + [ + 220, + 141 + ], + [ + 238, + 125 + ], + [ + 224, + 126 + ], + [ + 249, + 110 + ], + [ + 236, + 109 + ] + ], + [ + [ + 241, + 117 + ], + [ + 237, + 105 + ], + [ + 230, + 105 + ], + [ + 252, + 129 + ], + [ + 234, + 117 + ], + [ + 255, + 140 + ] + ], + [ + [ + 258, + 122 + ], + [ + 248, + 106 + ], + [ + 242, + 132 + ], + [ + 273, + 113 + ], + [ + 251, + 148 + ], + [ + 283, + 129 + ] + ] + ] + }, + { + "task_id": "0000585", + "id": "RH20T_cfg6_task_0039_user_0014_scene_0002_cfg_0006_104122061018#120", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000585/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 344, + 61 + ], + [ + 321, + 72 + ], + [ + 342, + 88 + ], + [ + 346, + 31 + ], + [ + 363, + 77 + ], + [ + 368, + 22 + ] + ], + [ + [ + 336, + 71 + ], + [ + 344, + 47 + ], + [ + 319, + 51 + ], + [ + 353, + 93 + ], + [ + 313, + 73 + ], + [ + 345, + 114 + ] + ], + [ + [ + 322, + 55 + ], + [ + 307, + 32 + ], + [ + 315, + 74 + ], + [ + 330, + 33 + ], + [ + 330, + 97 + ], + [ + 346, + 59 + ] + ], + [ + [ + 325, + 59 + ], + [ + 350, + 49 + ], + [ + 316, + 28 + ], + [ + 334, + 88 + ], + [ + 288, + 38 + ], + [ + 308, + 99 + ] + ] + ] + }, + { + "task_id": "0000586", + "id": "RH20T_cfg5_task_0075_user_0007_scene_0001_cfg_0005_105422061350#456", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Finish setting up the starting position of a chessboard that is almost arranged'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000586/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 108, + 181 + ], + [ + 124, + 159 + ], + [ + 139, + 190 + ], + [ + 77, + 172 + ], + [ + 123, + 214 + ], + [ + 57, + 195 + ] + ], + [ + [ + 89, + 160 + ], + [ + 92, + 130 + ], + [ + 114, + 166 + ], + [ + 61, + 153 + ], + [ + 112, + 196 + ], + [ + 57, + 186 + ] + ], + [ + [ + 124, + 207 + ], + [ + 146, + 200 + ], + [ + 150, + 212 + ], + [ + 96, + 201 + ], + [ + 130, + 218 + ], + [ + 75, + 208 + ] + ], + [ + [ + 121, + 179 + ], + [ + 115, + 163 + ], + [ + 154, + 176 + ], + [ + 86, + 181 + ], + [ + 157, + 191 + ], + [ + 93, + 196 + ] + ] + ] + }, + { + "task_id": "0000588", + "id": "RH20T_cfg5_task_0007_user_0010_scene_0007_cfg_0005_104122063678#298", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000588/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 309 + ], + [ + 327, + 314 + ], + [ + 296, + 333 + ], + [ + 302, + 287 + ], + [ + 266, + 328 + ], + [ + 275, + 283 + ] + ], + [ + [ + 321, + 305 + ], + [ + 329, + 291 + ], + [ + 327, + 333 + ], + [ + 314, + 280 + ], + [ + 319, + 345 + ], + [ + 307, + 294 + ] + ], + [ + [ + 317, + 312 + ], + [ + 315, + 292 + ], + [ + 305, + 333 + ], + [ + 327, + 294 + ], + [ + 306, + 352 + ], + [ + 328, + 313 + ] + ], + [ + [ + 335, + 273 + ], + [ + 341, + 301 + ], + [ + 308, + 279 + ], + [ + 362, + 268 + ], + [ + 303, + 252 + ], + [ + 356, + 240 + ] + ] + ] + }, + { + "task_id": "0000591", + "id": "RH20T_cfg7_task_0063_user_0020_scene_0008_cfg_0007_104122063633#134", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a shovel to scoop up an object'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000591/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 211, + 283 + ], + [ + 238, + 283 + ], + [ + 208, + 311 + ], + [ + 213, + 255 + ], + [ + 180, + 311 + ], + [ + 186, + 255 + ] + ], + [ + [ + 181, + 248 + ], + [ + 201, + 249 + ], + [ + 168, + 235 + ], + [ + 196, + 262 + ], + [ + 146, + 233 + ], + [ + 174, + 261 + ] + ], + [ + [ + 193, + 260 + ], + [ + 201, + 235 + ], + [ + 191, + 265 + ], + [ + 196, + 255 + ], + [ + 183, + 290 + ], + [ + 188, + 282 + ] + ], + [ + [ + 167, + 266 + ], + [ + 141, + 280 + ], + [ + 167, + 246 + ], + [ + 167, + 283 + ], + [ + 191, + 234 + ], + [ + 190, + 270 + ] + ] + ] + }, + { + "task_id": "0000592", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0006_cfg_0005_036422060215#220", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000592/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 286, + 146 + ], + [ + 288, + 166 + ], + [ + 294, + 151 + ], + [ + 277, + 140 + ], + [ + 292, + 132 + ], + [ + 275, + 120 + ] + ], + [ + [ + 284, + 158 + ], + [ + 284, + 141 + ], + [ + 289, + 168 + ], + [ + 278, + 148 + ], + [ + 288, + 185 + ], + [ + 278, + 166 + ] + ], + [ + [ + 276, + 144 + ], + [ + 279, + 127 + ], + [ + 295, + 140 + ], + [ + 257, + 148 + ], + [ + 292, + 155 + ], + [ + 256, + 164 + ] + ], + [ + [ + 274, + 143 + ], + [ + 283, + 148 + ], + [ + 258, + 155 + ], + [ + 291, + 130 + ], + [ + 247, + 149 + ], + [ + 282, + 123 + ] + ] + ] + }, + { + "task_id": "0000595", + "id": "RH20T_cfg6_task_0022_user_0014_scene_0007_cfg_0006_104122061602#149", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Wave the flag'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000595/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 410, + 150 + ], + [ + 404, + 135 + ], + [ + 390, + 158 + ], + [ + 431, + 141 + ], + [ + 396, + 174 + ], + [ + 439, + 157 + ] + ], + [ + [ + 394, + 149 + ], + [ + 373, + 145 + ], + [ + 396, + 159 + ], + [ + 391, + 138 + ], + [ + 418, + 162 + ], + [ + 415, + 141 + ] + ], + [ + [ + 416, + 152 + ], + [ + 421, + 133 + ], + [ + 428, + 166 + ], + [ + 405, + 139 + ], + [ + 422, + 184 + ], + [ + 401, + 156 + ] + ], + [ + [ + 422, + 167 + ], + [ + 422, + 177 + ], + [ + 398, + 169 + ], + [ + 447, + 165 + ], + [ + 397, + 158 + ], + [ + 448, + 154 + ] + ] + ] + }, + { + "task_id": "0000596", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_037522061512#183", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000596/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 214, + 146 + ], + [ + 228, + 135 + ], + [ + 220, + 171 + ], + [ + 208, + 122 + ], + [ + 206, + 182 + ], + [ + 195, + 134 + ] + ], + [ + [ + 243, + 138 + ], + [ + 223, + 129 + ], + [ + 230, + 158 + ], + [ + 256, + 118 + ], + [ + 250, + 168 + ], + [ + 276, + 126 + ] + ], + [ + [ + 191, + 142 + ], + [ + 169, + 132 + ], + [ + 204, + 150 + ], + [ + 178, + 134 + ], + [ + 224, + 159 + ], + [ + 198, + 143 + ] + ], + [ + [ + 211, + 146 + ], + [ + 206, + 128 + ], + [ + 226, + 131 + ], + [ + 194, + 162 + ], + [ + 230, + 148 + ], + [ + 200, + 179 + ] + ] + ] + }, + { + "task_id": "0000599", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_037522061512#714", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000599/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 198, + 172 + ], + [ + 192, + 154 + ], + [ + 219, + 179 + ], + [ + 176, + 164 + ], + [ + 225, + 198 + ], + [ + 182, + 184 + ] + ], + [ + [ + 217, + 162 + ], + [ + 209, + 151 + ], + [ + 230, + 142 + ], + [ + 203, + 183 + ], + [ + 237, + 152 + ], + [ + 211, + 192 + ] + ], + [ + [ + 210, + 158 + ], + [ + 186, + 166 + ], + [ + 215, + 178 + ], + [ + 205, + 136 + ], + [ + 238, + 170 + ], + [ + 229, + 129 + ] + ], + [ + [ + 196, + 142 + ], + [ + 201, + 158 + ], + [ + 198, + 122 + ], + [ + 193, + 163 + ], + [ + 192, + 104 + ], + [ + 187, + 145 + ] + ] + ] + }, + { + "task_id": "0000601", + "id": "RH20T_cfg4_task_0103_user_0014_scene_0012_cfg_0004_104122062295#212", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000601/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 248, + 211 + ], + [ + 254, + 178 + ], + [ + 210, + 198 + ], + [ + 287, + 223 + ], + [ + 207, + 228 + ], + [ + 280, + 252 + ] + ], + [ + [ + 209, + 209 + ], + [ + 229, + 192 + ], + [ + 246, + 218 + ], + [ + 169, + 199 + ], + [ + 227, + 232 + ], + [ + 153, + 215 + ] + ], + [ + [ + 269, + 214 + ], + [ + 295, + 192 + ], + [ + 294, + 242 + ], + [ + 244, + 186 + ], + [ + 269, + 261 + ], + [ + 221, + 207 + ] + ], + [ + [ + 206, + 232 + ], + [ + 175, + 236 + ], + [ + 187, + 264 + ], + [ + 222, + 203 + ], + [ + 223, + 261 + ], + [ + 254, + 198 + ] + ] + ] + }, + { + "task_id": "0000602", + "id": "RH20T_cfg4_task_0020_user_0016_scene_0002_cfg_0004_036422060909#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000602/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 233 + ], + [ + 247, + 212 + ], + [ + 284, + 212 + ], + [ + 247, + 255 + ], + [ + 301, + 232 + ], + [ + 266, + 275 + ] + ], + [ + [ + 299, + 217 + ], + [ + 302, + 229 + ], + [ + 277, + 203 + ], + [ + 321, + 232 + ], + [ + 271, + 189 + ], + [ + 318, + 219 + ] + ], + [ + [ + 292, + 197 + ], + [ + 298, + 176 + ], + [ + 288, + 177 + ], + [ + 297, + 219 + ], + [ + 284, + 196 + ], + [ + 292, + 238 + ] + ], + [ + [ + 288, + 203 + ], + [ + 314, + 187 + ], + [ + 290, + 195 + ], + [ + 285, + 210 + ], + [ + 261, + 213 + ], + [ + 260, + 226 + ] + ] + ] + }, + { + "task_id": "0000604", + "id": "RH20T_cfg6_task_0039_user_0014_scene_0002_cfg_0006_104122064161#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000604/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 163 + ], + [ + 339, + 142 + ], + [ + 340, + 180 + ], + [ + 336, + 148 + ], + [ + 339, + 200 + ], + [ + 335, + 167 + ] + ], + [ + [ + 318, + 189 + ], + [ + 292, + 192 + ], + [ + 316, + 204 + ], + [ + 319, + 174 + ], + [ + 341, + 201 + ], + [ + 346, + 171 + ] + ], + [ + [ + 314, + 162 + ], + [ + 291, + 160 + ], + [ + 300, + 168 + ], + [ + 329, + 155 + ], + [ + 321, + 169 + ], + [ + 351, + 157 + ] + ], + [ + [ + 332, + 165 + ], + [ + 308, + 159 + ], + [ + 330, + 159 + ], + [ + 334, + 170 + ], + [ + 356, + 166 + ], + [ + 358, + 176 + ] + ] + ] + }, + { + "task_id": "0000606", + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_104122061850#313", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000606/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 350, + 202 + ], + [ + 328, + 191 + ], + [ + 339, + 205 + ], + [ + 361, + 199 + ], + [ + 360, + 215 + ], + [ + 384, + 210 + ] + ], + [ + [ + 325, + 220 + ], + [ + 317, + 199 + ], + [ + 305, + 233 + ], + [ + 346, + 207 + ], + [ + 314, + 255 + ], + [ + 356, + 230 + ] + ], + [ + [ + 345, + 219 + ], + [ + 340, + 207 + ], + [ + 321, + 231 + ], + [ + 369, + 208 + ], + [ + 327, + 241 + ], + [ + 372, + 219 + ] + ], + [ + [ + 365, + 208 + ], + [ + 374, + 230 + ], + [ + 346, + 219 + ], + [ + 384, + 195 + ], + [ + 338, + 198 + ], + [ + 375, + 174 + ] + ] + ] + }, + { + "task_id": "0000607", + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_104122061850#491", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000607/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 162 + ], + [ + 311, + 136 + ], + [ + 341, + 159 + ], + [ + 288, + 165 + ], + [ + 345, + 184 + ], + [ + 292, + 191 + ] + ], + [ + [ + 309, + 118 + ], + [ + 293, + 97 + ], + [ + 288, + 133 + ], + [ + 329, + 102 + ], + [ + 304, + 153 + ], + [ + 345, + 122 + ] + ], + [ + [ + 350, + 142 + ], + [ + 341, + 134 + ], + [ + 364, + 156 + ], + [ + 335, + 127 + ], + [ + 375, + 166 + ], + [ + 344, + 136 + ] + ], + [ + [ + 326, + 146 + ], + [ + 321, + 131 + ], + [ + 302, + 152 + ], + [ + 351, + 139 + ], + [ + 308, + 165 + ], + [ + 354, + 153 + ] + ] + ] + }, + { + "task_id": "0000609", + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_104122061850#894", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000609/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 382, + 281 + ], + [ + 389, + 267 + ], + [ + 382, + 304 + ], + [ + 382, + 257 + ], + [ + 374, + 321 + ], + [ + 373, + 272 + ] + ], + [ + [ + 384, + 279 + ], + [ + 407, + 266 + ], + [ + 401, + 303 + ], + [ + 368, + 256 + ], + [ + 379, + 315 + ], + [ + 347, + 269 + ] + ], + [ + [ + 399, + 309 + ], + [ + 398, + 336 + ], + [ + 419, + 319 + ], + [ + 381, + 299 + ], + [ + 421, + 291 + ], + [ + 382, + 273 + ] + ], + [ + [ + 385, + 296 + ], + [ + 383, + 285 + ], + [ + 364, + 277 + ], + [ + 407, + 316 + ], + [ + 367, + 287 + ], + [ + 408, + 325 + ] + ] + ] + }, + { + "task_id": "0000610", + "id": "RH20T_cfg1_task_0023_user_0014_scene_0002_cfg_0001_819112071037#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000610/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 174, + 133 + ], + [ + 161, + 69 + ], + [ + 231, + 120 + ], + [ + 107, + 148 + ], + [ + 239, + 175 + ], + [ + 123, + 210 + ] + ], + [ + [ + 161, + 120 + ], + [ + 144, + 117 + ], + [ + 180, + 49 + ], + [ + 143, + 191 + ], + [ + 204, + 41 + ], + [ + 162, + 204 + ] + ], + [ + [ + 186, + 113 + ], + [ + 191, + 53 + ], + [ + 122, + 117 + ], + [ + 240, + 110 + ], + [ + 121, + 175 + ], + [ + 233, + 160 + ] + ], + [ + [ + 223, + 103 + ], + [ + 262, + 140 + ], + [ + 205, + 103 + ], + [ + 237, + 103 + ], + [ + 161, + 60 + ], + [ + 202, + 69 + ] + ] + ] + }, + { + "task_id": "0000613", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0005_cfg_0004_104122062295#274", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000613/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 178, + 153 + ], + [ + 211, + 184 + ], + [ + 149, + 186 + ], + [ + 208, + 120 + ], + [ + 116, + 156 + ], + [ + 174, + 90 + ] + ], + [ + [ + 212, + 165 + ], + [ + 218, + 130 + ], + [ + 181, + 187 + ], + [ + 239, + 146 + ], + [ + 178, + 219 + ], + [ + 232, + 176 + ] + ], + [ + [ + 201, + 221 + ], + [ + 201, + 229 + ], + [ + 231, + 255 + ], + [ + 169, + 186 + ], + [ + 228, + 246 + ], + [ + 172, + 183 + ] + ], + [ + [ + 210, + 207 + ], + [ + 227, + 246 + ], + [ + 177, + 226 + ], + [ + 244, + 188 + ], + [ + 161, + 189 + ], + [ + 226, + 149 + ] + ] + ] + }, + { + "task_id": "0000614", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0005_cfg_0004_104122062295#469", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000614/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 192, + 294 + ], + [ + 221, + 332 + ], + [ + 182, + 273 + ], + [ + 201, + 312 + ], + [ + 154, + 236 + ], + [ + 176, + 279 + ] + ], + [ + [ + 231, + 253 + ], + [ + 204, + 218 + ], + [ + 224, + 255 + ], + [ + 238, + 252 + ], + [ + 256, + 295 + ], + [ + 263, + 285 + ] + ], + [ + [ + 201, + 273 + ], + [ + 214, + 264 + ], + [ + 198, + 223 + ], + [ + 205, + 328 + ], + [ + 188, + 234 + ], + [ + 193, + 330 + ] + ], + [ + [ + 212, + 251 + ], + [ + 229, + 219 + ], + [ + 186, + 214 + ], + [ + 239, + 290 + ], + [ + 173, + 243 + ], + [ + 222, + 317 + ] + ] + ] + }, + { + "task_id": "0000615", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0005_cfg_0004_104122062295#657", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000615/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 163, + 150 + ], + [ + 148, + 188 + ], + [ + 181, + 132 + ], + [ + 142, + 173 + ], + [ + 196, + 93 + ], + [ + 159, + 128 + ] + ], + [ + [ + 145, + 68 + ], + [ + 162, + 122 + ], + [ + 159, + 76 + ], + [ + 127, + 58 + ], + [ + 143, + 27 + ], + [ + 106, + 0 + ] + ], + [ + [ + 168, + 98 + ], + [ + 181, + 57 + ], + [ + 117, + 103 + ], + [ + 211, + 93 + ], + [ + 108, + 141 + ], + [ + 198, + 127 + ] + ], + [ + [ + 127, + 108 + ], + [ + 157, + 126 + ], + [ + 130, + 157 + ], + [ + 124, + 53 + ], + [ + 105, + 138 + ], + [ + 97, + 40 + ] + ] + ] + }, + { + "task_id": "0000625", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0007_cfg_0004_104122063550#48", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000625/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 123 + ], + [ + 280, + 101 + ], + [ + 289, + 135 + ], + [ + 244, + 111 + ], + [ + 275, + 157 + ], + [ + 230, + 132 + ] + ], + [ + [ + 288, + 170 + ], + [ + 297, + 188 + ], + [ + 305, + 153 + ], + [ + 270, + 187 + ], + [ + 297, + 134 + ], + [ + 260, + 167 + ] + ], + [ + [ + 300, + 154 + ], + [ + 278, + 162 + ], + [ + 296, + 171 + ], + [ + 304, + 135 + ], + [ + 316, + 163 + ], + [ + 325, + 127 + ] + ], + [ + [ + 272, + 156 + ], + [ + 271, + 144 + ], + [ + 295, + 168 + ], + [ + 249, + 143 + ], + [ + 295, + 178 + ], + [ + 251, + 155 + ] + ] + ] + }, + { + "task_id": "0000626", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0007_cfg_0004_104122063550#150", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000626/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 241 + ], + [ + 329, + 217 + ], + [ + 291, + 230 + ], + [ + 339, + 253 + ], + [ + 276, + 254 + ], + [ + 324, + 279 + ] + ], + [ + [ + 298, + 239 + ], + [ + 277, + 251 + ], + [ + 319, + 255 + ], + [ + 279, + 225 + ], + [ + 338, + 243 + ], + [ + 299, + 215 + ] + ], + [ + [ + 311, + 191 + ], + [ + 300, + 220 + ], + [ + 306, + 197 + ], + [ + 315, + 186 + ], + [ + 316, + 171 + ], + [ + 327, + 157 + ] + ], + [ + [ + 303, + 208 + ], + [ + 305, + 198 + ], + [ + 329, + 221 + ], + [ + 278, + 195 + ], + [ + 326, + 229 + ], + [ + 277, + 205 + ] + ] + ] + }, + { + "task_id": "0000627", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0007_cfg_0004_104122063550#219", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000627/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 212, + 130 + ], + [ + 212, + 130 + ], + [ + 226, + 152 + ], + [ + 197, + 108 + ], + [ + 227, + 153 + ], + [ + 197, + 107 + ] + ], + [ + [ + 219, + 96 + ], + [ + 195, + 109 + ], + [ + 220, + 106 + ], + [ + 217, + 84 + ], + [ + 241, + 93 + ], + [ + 241, + 70 + ] + ], + [ + [ + 217, + 90 + ], + [ + 212, + 75 + ], + [ + 231, + 109 + ], + [ + 203, + 72 + ], + [ + 235, + 121 + ], + [ + 208, + 86 + ] + ], + [ + [ + 205, + 121 + ], + [ + 208, + 111 + ], + [ + 231, + 125 + ], + [ + 179, + 117 + ], + [ + 229, + 136 + ], + [ + 174, + 128 + ] + ] + ] + }, + { + "task_id": "0000631", + "id": "RH20T_cfg4_task_0050_user_0014_scene_0010_cfg_0004_036422060909#126", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the dish on the dish rack'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000631/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 304, + 138 + ], + [ + 298, + 125 + ], + [ + 310, + 117 + ], + [ + 298, + 159 + ], + [ + 315, + 129 + ], + [ + 304, + 170 + ] + ], + [ + [ + 308, + 106 + ], + [ + 322, + 98 + ], + [ + 288, + 106 + ], + [ + 329, + 107 + ], + [ + 273, + 114 + ], + [ + 315, + 116 + ] + ], + [ + [ + 307, + 129 + ], + [ + 311, + 111 + ], + [ + 320, + 115 + ], + [ + 293, + 144 + ], + [ + 316, + 132 + ], + [ + 290, + 161 + ] + ], + [ + [ + 313, + 148 + ], + [ + 305, + 147 + ], + [ + 297, + 168 + ], + [ + 329, + 128 + ], + [ + 305, + 168 + ], + [ + 336, + 130 + ] + ] + ] + }, + { + "task_id": "0000633", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0001_cfg_0006_104122061018#477", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000633/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 275, + 124 + ], + [ + 269, + 126 + ], + [ + 269, + 91 + ], + [ + 280, + 156 + ], + [ + 275, + 91 + ], + [ + 284, + 153 + ] + ], + [ + [ + 257, + 133 + ], + [ + 259, + 106 + ], + [ + 240, + 110 + ], + [ + 276, + 157 + ], + [ + 239, + 134 + ], + [ + 273, + 180 + ] + ], + [ + [ + 242, + 155 + ], + [ + 252, + 123 + ], + [ + 275, + 172 + ], + [ + 209, + 138 + ], + [ + 264, + 201 + ], + [ + 201, + 167 + ] + ], + [ + [ + 228, + 103 + ], + [ + 262, + 89 + ], + [ + 240, + 133 + ], + [ + 217, + 74 + ], + [ + 206, + 148 + ], + [ + 184, + 87 + ] + ] + ] + }, + { + "task_id": "0000634", + "id": "RH20T_cfg2_task_0080_user_0014_scene_0006_cfg_0002_036422060215#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Flip over and spread out the paper that is laid flat on the table'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000634/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 244, + 158 + ], + [ + 247, + 143 + ], + [ + 257, + 169 + ], + [ + 232, + 146 + ], + [ + 254, + 183 + ], + [ + 229, + 160 + ] + ], + [ + [ + 226, + 156 + ], + [ + 212, + 169 + ], + [ + 235, + 171 + ], + [ + 217, + 142 + ], + [ + 248, + 158 + ], + [ + 231, + 129 + ] + ], + [ + [ + 256, + 156 + ], + [ + 270, + 167 + ], + [ + 243, + 170 + ], + [ + 270, + 142 + ], + [ + 228, + 160 + ], + [ + 255, + 131 + ] + ], + [ + [ + 251, + 158 + ], + [ + 267, + 168 + ], + [ + 245, + 170 + ], + [ + 257, + 147 + ], + [ + 227, + 160 + ], + [ + 241, + 137 + ] + ] + ] + }, + { + "task_id": "0000635", + "id": "RH20T_cfg6_task_0029_user_0014_scene_0003_cfg_0006_104122061330#215", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000635/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 365, + 95 + ], + [ + 345, + 102 + ], + [ + 369, + 123 + ], + [ + 360, + 67 + ], + [ + 387, + 116 + ], + [ + 379, + 62 + ] + ], + [ + [ + 369, + 70 + ], + [ + 393, + 81 + ], + [ + 371, + 48 + ], + [ + 366, + 91 + ], + [ + 346, + 38 + ], + [ + 344, + 81 + ] + ], + [ + [ + 353, + 68 + ], + [ + 326, + 65 + ], + [ + 341, + 85 + ], + [ + 363, + 52 + ], + [ + 370, + 88 + ], + [ + 390, + 54 + ] + ], + [ + [ + 354, + 64 + ], + [ + 346, + 52 + ], + [ + 343, + 90 + ], + [ + 364, + 38 + ], + [ + 350, + 99 + ], + [ + 370, + 50 + ] + ] + ] + }, + { + "task_id": "0000638", + "id": "RH20T_cfg3_task_0077_user_0010_scene_0009_cfg_0003_036422060909#221", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000638/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 163 + ], + [ + 294, + 140 + ], + [ + 302, + 171 + ], + [ + 265, + 155 + ], + [ + 292, + 194 + ], + [ + 254, + 180 + ] + ], + [ + [ + 269, + 134 + ], + [ + 293, + 129 + ], + [ + 272, + 119 + ], + [ + 265, + 149 + ], + [ + 249, + 124 + ], + [ + 241, + 154 + ] + ], + [ + [ + 252, + 131 + ], + [ + 226, + 125 + ], + [ + 259, + 107 + ], + [ + 246, + 154 + ], + [ + 285, + 113 + ], + [ + 270, + 160 + ] + ], + [ + [ + 260, + 156 + ], + [ + 258, + 138 + ], + [ + 253, + 139 + ], + [ + 268, + 174 + ], + [ + 255, + 156 + ], + [ + 269, + 191 + ] + ] + ] + }, + { + "task_id": "0000639", + "id": "RH20T_cfg3_task_0077_user_0010_scene_0009_cfg_0003_036422060909#341", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000639/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 173 + ], + [ + 346, + 153 + ], + [ + 331, + 161 + ], + [ + 365, + 186 + ], + [ + 333, + 180 + ], + [ + 365, + 205 + ] + ], + [ + [ + 336, + 173 + ], + [ + 357, + 187 + ], + [ + 326, + 185 + ], + [ + 347, + 160 + ], + [ + 306, + 172 + ], + [ + 324, + 145 + ] + ], + [ + [ + 344, + 163 + ], + [ + 353, + 155 + ], + [ + 349, + 186 + ], + [ + 340, + 138 + ], + [ + 339, + 196 + ], + [ + 330, + 145 + ] + ], + [ + [ + 328, + 185 + ], + [ + 326, + 196 + ], + [ + 336, + 164 + ], + [ + 318, + 206 + ], + [ + 338, + 151 + ], + [ + 319, + 194 + ] + ] + ] + }, + { + "task_id": "0000642", + "id": "RH20T_cfg1_task_0073_user_0001_scene_0003_cfg_0001_819112070044#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000642/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 491, + 142 + ], + [ + 472, + 131 + ], + [ + 506, + 123 + ], + [ + 477, + 160 + ], + [ + 525, + 134 + ], + [ + 495, + 170 + ] + ], + [ + [ + 452, + 124 + ], + [ + 453, + 132 + ], + [ + 477, + 124 + ], + [ + 428, + 124 + ], + [ + 477, + 115 + ], + [ + 426, + 115 + ] + ], + [ + [ + 478, + 134 + ], + [ + 486, + 115 + ], + [ + 494, + 128 + ], + [ + 461, + 141 + ], + [ + 487, + 146 + ], + [ + 454, + 160 + ] + ], + [ + [ + 473, + 124 + ], + [ + 496, + 126 + ], + [ + 463, + 145 + ], + [ + 483, + 102 + ], + [ + 443, + 142 + ], + [ + 463, + 100 + ] + ] + ] + }, + { + "task_id": "0000644", + "id": "RH20T_cfg7_task_0044_user_0020_scene_0001_cfg_0007_104122063633#112", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000644/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 228, + 154 + ], + [ + 205, + 155 + ], + [ + 226, + 147 + ], + [ + 229, + 159 + ], + [ + 249, + 146 + ], + [ + 251, + 158 + ] + ], + [ + [ + 212, + 161 + ], + [ + 213, + 143 + ], + [ + 214, + 149 + ], + [ + 210, + 174 + ], + [ + 212, + 166 + ], + [ + 208, + 192 + ] + ], + [ + [ + 191, + 144 + ], + [ + 211, + 155 + ], + [ + 183, + 163 + ], + [ + 200, + 125 + ], + [ + 163, + 152 + ], + [ + 179, + 114 + ] + ], + [ + [ + 204, + 152 + ], + [ + 205, + 170 + ], + [ + 206, + 138 + ], + [ + 203, + 164 + ], + [ + 205, + 121 + ], + [ + 202, + 147 + ] + ] + ] + }, + { + "task_id": "0000647", + "id": "RH20T_cfg1_task_0059_user_0001_scene_0001_cfg_0001_035622060973#389", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000647/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 273, + 110 + ], + [ + 285, + 122 + ], + [ + 281, + 117 + ], + [ + 264, + 102 + ], + [ + 271, + 106 + ], + [ + 253, + 90 + ] + ], + [ + [ + 243, + 102 + ], + [ + 234, + 85 + ], + [ + 260, + 96 + ], + [ + 226, + 108 + ], + [ + 270, + 112 + ], + [ + 236, + 125 + ] + ], + [ + [ + 254, + 116 + ], + [ + 254, + 104 + ], + [ + 241, + 128 + ], + [ + 268, + 104 + ], + [ + 242, + 139 + ], + [ + 267, + 115 + ] + ], + [ + [ + 237, + 105 + ], + [ + 254, + 113 + ], + [ + 233, + 92 + ], + [ + 240, + 118 + ], + [ + 216, + 84 + ], + [ + 225, + 111 + ] + ] + ] + }, + { + "task_id": "0000650", + "id": "RH20T_cfg1_task_0085_user_0015_scene_0002_cfg_0001_035622060973#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the bubble ring from the assembled bubble ring and ball'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000650/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 235, + 204 + ], + [ + 217, + 215 + ], + [ + 245, + 221 + ], + [ + 224, + 186 + ], + [ + 263, + 210 + ], + [ + 241, + 176 + ] + ], + [ + [ + 229, + 190 + ], + [ + 218, + 187 + ], + [ + 245, + 179 + ], + [ + 212, + 202 + ], + [ + 256, + 182 + ], + [ + 223, + 207 + ] + ], + [ + [ + 227, + 209 + ], + [ + 223, + 199 + ], + [ + 216, + 195 + ], + [ + 240, + 224 + ], + [ + 220, + 204 + ], + [ + 243, + 233 + ] + ], + [ + [ + 225, + 207 + ], + [ + 213, + 193 + ], + [ + 239, + 205 + ], + [ + 209, + 209 + ], + [ + 251, + 218 + ], + [ + 221, + 224 + ] + ] + ] + }, + { + "task_id": "0000653", + "id": "RH20T_cfg6_task_0045_user_0020_scene_0010_cfg_0006_104122064161#294", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the brush on the pen rack'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000653/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 384, + 201 + ], + [ + 360, + 202 + ], + [ + 383, + 176 + ], + [ + 386, + 226 + ], + [ + 408, + 174 + ], + [ + 411, + 225 + ] + ], + [ + [ + 385, + 240 + ], + [ + 409, + 244 + ], + [ + 387, + 250 + ], + [ + 382, + 229 + ], + [ + 364, + 246 + ], + [ + 357, + 225 + ] + ], + [ + [ + 400, + 231 + ], + [ + 374, + 229 + ], + [ + 397, + 252 + ], + [ + 403, + 212 + ], + [ + 425, + 255 + ], + [ + 429, + 214 + ] + ], + [ + [ + 377, + 224 + ], + [ + 372, + 237 + ], + [ + 383, + 245 + ], + [ + 371, + 201 + ], + [ + 387, + 233 + ], + [ + 375, + 190 + ] + ] + ] + }, + { + "task_id": "0000654", + "id": "RH20T_cfg3_task_0062_user_0010_scene_0008_cfg_0003_104122062295#170", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the toilet paper on its holder'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000654/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 341, + 188 + ], + [ + 313, + 162 + ], + [ + 328, + 216 + ], + [ + 352, + 162 + ], + [ + 356, + 241 + ], + [ + 377, + 186 + ] + ], + [ + [ + 312, + 141 + ], + [ + 306, + 109 + ], + [ + 281, + 128 + ], + [ + 338, + 151 + ], + [ + 286, + 164 + ], + [ + 345, + 184 + ] + ], + [ + [ + 357, + 149 + ], + [ + 348, + 179 + ], + [ + 357, + 179 + ], + [ + 357, + 115 + ], + [ + 365, + 152 + ], + [ + 366, + 89 + ] + ], + [ + [ + 330, + 164 + ], + [ + 328, + 131 + ], + [ + 298, + 152 + ], + [ + 365, + 176 + ], + [ + 302, + 181 + ], + [ + 366, + 205 + ] + ] + ] + }, + { + "task_id": "0000656", + "id": "RH20T_cfg4_task_0033_user_0016_scene_0008_cfg_0004_104122062823#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000656/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 493, + 138 + ], + [ + 505, + 106 + ], + [ + 536, + 142 + ], + [ + 451, + 135 + ], + [ + 522, + 170 + ], + [ + 443, + 163 + ] + ], + [ + [ + 467, + 93 + ], + [ + 478, + 79 + ], + [ + 447, + 123 + ], + [ + 488, + 59 + ], + [ + 434, + 139 + ], + [ + 476, + 73 + ] + ], + [ + [ + 525, + 182 + ], + [ + 534, + 184 + ], + [ + 516, + 141 + ], + [ + 533, + 224 + ], + [ + 505, + 135 + ], + [ + 523, + 225 + ] + ], + [ + [ + 531, + 154 + ], + [ + 491, + 150 + ], + [ + 555, + 154 + ], + [ + 511, + 154 + ], + [ + 598, + 158 + ], + [ + 547, + 157 + ] + ] + ] + }, + { + "task_id": "0000657", + "id": "RH20T_cfg4_task_0033_user_0016_scene_0008_cfg_0004_104122062823#145", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000657/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 464, + 159 + ], + [ + 479, + 127 + ], + [ + 502, + 159 + ], + [ + 425, + 158 + ], + [ + 487, + 187 + ], + [ + 414, + 187 + ] + ], + [ + [ + 451, + 190 + ], + [ + 433, + 157 + ], + [ + 489, + 178 + ], + [ + 413, + 203 + ], + [ + 510, + 213 + ], + [ + 431, + 240 + ] + ], + [ + [ + 438, + 192 + ], + [ + 442, + 231 + ], + [ + 480, + 195 + ], + [ + 396, + 188 + ], + [ + 475, + 159 + ], + [ + 395, + 152 + ] + ], + [ + [ + 500, + 132 + ], + [ + 534, + 102 + ], + [ + 533, + 154 + ], + [ + 465, + 109 + ], + [ + 501, + 179 + ], + [ + 436, + 137 + ] + ] + ] + }, + { + "task_id": "0000662", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0003_cfg_0005_104122061850#248", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000662/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 333, + 85 + ], + [ + 327, + 68 + ], + [ + 316, + 102 + ], + [ + 349, + 68 + ], + [ + 323, + 116 + ], + [ + 354, + 84 + ] + ], + [ + [ + 340, + 118 + ], + [ + 324, + 107 + ], + [ + 348, + 129 + ], + [ + 331, + 107 + ], + [ + 365, + 141 + ], + [ + 349, + 119 + ] + ], + [ + [ + 358, + 76 + ], + [ + 372, + 89 + ], + [ + 377, + 57 + ], + [ + 340, + 95 + ], + [ + 363, + 45 + ], + [ + 328, + 83 + ] + ], + [ + [ + 342, + 61 + ], + [ + 339, + 43 + ], + [ + 321, + 72 + ], + [ + 364, + 49 + ], + [ + 324, + 90 + ], + [ + 369, + 68 + ] + ] + ] + }, + { + "task_id": "0000663", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0003_cfg_0005_104122061850#431", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000663/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 336, + 69 + ], + [ + 331, + 51 + ], + [ + 360, + 61 + ], + [ + 312, + 77 + ], + [ + 365, + 80 + ], + [ + 317, + 96 + ] + ], + [ + [ + 364, + 115 + ], + [ + 349, + 102 + ], + [ + 383, + 100 + ], + [ + 345, + 129 + ], + [ + 396, + 113 + ], + [ + 359, + 141 + ] + ], + [ + [ + 348, + 78 + ], + [ + 360, + 93 + ], + [ + 359, + 58 + ], + [ + 338, + 97 + ], + [ + 346, + 45 + ], + [ + 327, + 83 + ] + ], + [ + [ + 342, + 93 + ], + [ + 337, + 76 + ], + [ + 320, + 104 + ], + [ + 364, + 83 + ], + [ + 326, + 119 + ], + [ + 368, + 99 + ] + ] + ] + }, + { + "task_id": "0000664", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0005_cfg_0006_104122061602#169", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000664/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 438, + 83 + ], + [ + 430, + 59 + ], + [ + 457, + 85 + ], + [ + 420, + 81 + ], + [ + 465, + 109 + ], + [ + 428, + 103 + ] + ], + [ + [ + 423, + 81 + ], + [ + 434, + 66 + ], + [ + 432, + 102 + ], + [ + 414, + 60 + ], + [ + 421, + 114 + ], + [ + 405, + 74 + ] + ], + [ + [ + 407, + 80 + ], + [ + 426, + 81 + ], + [ + 401, + 103 + ], + [ + 413, + 57 + ], + [ + 382, + 103 + ], + [ + 393, + 56 + ] + ], + [ + [ + 406, + 59 + ], + [ + 385, + 74 + ], + [ + 422, + 75 + ], + [ + 391, + 43 + ], + [ + 443, + 59 + ], + [ + 412, + 27 + ] + ] + ] + }, + { + "task_id": "0000666", + "id": "RH20T_cfg3_task_0007_user_0016_scene_0002_cfg_0003_104122062295#175", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000666/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 122, + 258 + ], + [ + 95, + 225 + ], + [ + 165, + 234 + ], + [ + 78, + 284 + ], + [ + 186, + 264 + ], + [ + 105, + 314 + ] + ], + [ + [ + 44, + 208 + ], + [ + 70, + 163 + ], + [ + 7, + 191 + ], + [ + 82, + 227 + ], + [ + 0, + 237 + ], + [ + 55, + 276 + ] + ], + [ + [ + 19, + 250 + ], + [ + 45, + 247 + ], + [ + 0, + 250 + ], + [ + 66, + 249 + ], + [ + 0, + 253 + ], + [ + 42, + 252 + ] + ], + [ + [ + 77, + 234 + ], + [ + 79, + 198 + ], + [ + 55, + 208 + ], + [ + 100, + 264 + ], + [ + 54, + 241 + ], + [ + 97, + 297 + ] + ] + ] + }, + { + "task_id": "0000667", + "id": "RH20T_cfg7_task_0065_user_0014_scene_0003_cfg_0007_104122061602#177", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000667/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 526, + 198 + ], + [ + 535, + 180 + ], + [ + 547, + 214 + ], + [ + 507, + 182 + ], + [ + 538, + 231 + ], + [ + 500, + 199 + ] + ], + [ + [ + 553, + 205 + ], + [ + 536, + 222 + ], + [ + 574, + 209 + ], + [ + 534, + 201 + ], + [ + 592, + 191 + ], + [ + 550, + 185 + ] + ], + [ + [ + 502, + 199 + ], + [ + 481, + 212 + ], + [ + 500, + 193 + ], + [ + 505, + 205 + ], + [ + 521, + 181 + ], + [ + 528, + 191 + ] + ], + [ + [ + 537, + 198 + ], + [ + 535, + 224 + ], + [ + 521, + 198 + ], + [ + 555, + 198 + ], + [ + 524, + 173 + ], + [ + 558, + 170 + ] + ] + ] + }, + { + "task_id": "0000669", + "id": "RH20T_cfg7_task_0065_user_0014_scene_0003_cfg_0007_104122061602#569", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000669/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 495, + 199 + ], + [ + 470, + 204 + ], + [ + 487, + 189 + ], + [ + 502, + 209 + ], + [ + 514, + 183 + ], + [ + 526, + 204 + ] + ], + [ + [ + 510, + 188 + ], + [ + 515, + 168 + ], + [ + 531, + 200 + ], + [ + 492, + 177 + ], + [ + 525, + 219 + ], + [ + 488, + 195 + ] + ], + [ + [ + 542, + 183 + ], + [ + 531, + 166 + ], + [ + 555, + 165 + ], + [ + 529, + 199 + ], + [ + 567, + 182 + ], + [ + 539, + 217 + ] + ], + [ + [ + 515, + 190 + ], + [ + 500, + 209 + ], + [ + 533, + 204 + ], + [ + 497, + 178 + ], + [ + 549, + 184 + ], + [ + 511, + 160 + ] + ] + ] + }, + { + "task_id": "0000670", + "id": "RH20T_cfg5_task_0059_user_0007_scene_0003_cfg_0005_104422070044#384", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000670/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 365, + 207 + ], + [ + 361, + 187 + ], + [ + 384, + 201 + ], + [ + 347, + 212 + ], + [ + 389, + 220 + ], + [ + 351, + 232 + ] + ], + [ + [ + 367, + 199 + ], + [ + 365, + 186 + ], + [ + 380, + 209 + ], + [ + 354, + 189 + ], + [ + 381, + 222 + ], + [ + 355, + 201 + ] + ], + [ + [ + 389, + 191 + ], + [ + 372, + 197 + ], + [ + 383, + 172 + ], + [ + 395, + 211 + ], + [ + 399, + 168 + ], + [ + 411, + 205 + ] + ], + [ + [ + 364, + 206 + ], + [ + 379, + 196 + ], + [ + 362, + 197 + ], + [ + 367, + 214 + ], + [ + 346, + 209 + ], + [ + 352, + 225 + ] + ] + ] + }, + { + "task_id": "0000673", + "id": "RH20T_cfg7_task_0055_user_0020_scene_0007_cfg_0007_104122061330#315", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000673/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 320, + 189 + ], + [ + 315, + 207 + ], + [ + 294, + 193 + ], + [ + 347, + 184 + ], + [ + 300, + 177 + ], + [ + 351, + 168 + ] + ], + [ + [ + 308, + 188 + ], + [ + 295, + 186 + ], + [ + 298, + 215 + ], + [ + 317, + 160 + ], + [ + 310, + 215 + ], + [ + 328, + 163 + ] + ], + [ + [ + 281, + 189 + ], + [ + 297, + 211 + ], + [ + 260, + 209 + ], + [ + 302, + 169 + ], + [ + 246, + 187 + ], + [ + 285, + 148 + ] + ], + [ + [ + 317, + 216 + ], + [ + 341, + 229 + ], + [ + 325, + 218 + ], + [ + 310, + 215 + ], + [ + 299, + 203 + ], + [ + 287, + 202 + ] + ] + ] + }, + { + "task_id": "0000674", + "id": "RH20T_cfg7_task_0055_user_0020_scene_0007_cfg_0007_104122061330#405", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000674/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 308, + 202 + ], + [ + 335, + 201 + ], + [ + 319, + 183 + ], + [ + 298, + 219 + ], + [ + 289, + 184 + ], + [ + 271, + 221 + ] + ], + [ + [ + 312, + 205 + ], + [ + 286, + 196 + ], + [ + 301, + 197 + ], + [ + 323, + 212 + ], + [ + 331, + 207 + ], + [ + 349, + 220 + ] + ], + [ + [ + 307, + 192 + ], + [ + 296, + 185 + ], + [ + 296, + 220 + ], + [ + 318, + 165 + ], + [ + 307, + 225 + ], + [ + 328, + 173 + ] + ], + [ + [ + 325, + 195 + ], + [ + 306, + 188 + ], + [ + 313, + 176 + ], + [ + 335, + 213 + ], + [ + 334, + 182 + ], + [ + 355, + 221 + ] + ] + ] + }, + { + "task_id": "0000675", + "id": "RH20T_cfg7_task_0055_user_0020_scene_0007_cfg_0007_104122061330#586", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000675/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 127 + ], + [ + 322, + 119 + ], + [ + 325, + 153 + ], + [ + 343, + 100 + ], + [ + 335, + 159 + ], + [ + 352, + 108 + ] + ], + [ + [ + 348, + 114 + ], + [ + 323, + 111 + ], + [ + 344, + 141 + ], + [ + 351, + 86 + ], + [ + 370, + 145 + ], + [ + 377, + 88 + ] + ], + [ + [ + 357, + 133 + ], + [ + 373, + 147 + ], + [ + 352, + 154 + ], + [ + 361, + 112 + ], + [ + 335, + 139 + ], + [ + 345, + 96 + ] + ], + [ + [ + 365, + 146 + ], + [ + 365, + 119 + ], + [ + 338, + 147 + ], + [ + 392, + 146 + ], + [ + 338, + 174 + ], + [ + 392, + 173 + ] + ] + ] + }, + { + "task_id": "0000678", + "id": "RH20T_cfg1_task_0064_user_0001_scene_0008_cfg_0001_038522062547#187", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000678/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 434, + 226 + ], + [ + 421, + 251 + ], + [ + 407, + 212 + ], + [ + 462, + 240 + ], + [ + 420, + 186 + ], + [ + 476, + 214 + ] + ], + [ + [ + 418, + 229 + ], + [ + 424, + 219 + ], + [ + 447, + 240 + ], + [ + 389, + 218 + ], + [ + 440, + 247 + ], + [ + 386, + 226 + ] + ], + [ + [ + 424, + 208 + ], + [ + 418, + 234 + ], + [ + 441, + 200 + ], + [ + 410, + 215 + ], + [ + 446, + 174 + ], + [ + 415, + 192 + ] + ], + [ + [ + 385, + 236 + ], + [ + 375, + 215 + ], + [ + 366, + 255 + ], + [ + 405, + 217 + ], + [ + 376, + 278 + ], + [ + 417, + 239 + ] + ] + ] + }, + { + "task_id": "0000680", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_750612070851#214", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000680/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 199 + ], + [ + 366, + 219 + ], + [ + 365, + 168 + ], + [ + 361, + 229 + ], + [ + 362, + 152 + ], + [ + 358, + 210 + ] + ], + [ + [ + 407, + 225 + ], + [ + 393, + 196 + ], + [ + 377, + 228 + ], + [ + 434, + 222 + ], + [ + 391, + 260 + ], + [ + 450, + 252 + ] + ], + [ + [ + 373, + 180 + ], + [ + 376, + 150 + ], + [ + 338, + 182 + ], + [ + 407, + 178 + ], + [ + 336, + 209 + ], + [ + 402, + 205 + ] + ], + [ + [ + 359, + 191 + ], + [ + 352, + 227 + ], + [ + 364, + 197 + ], + [ + 354, + 184 + ], + [ + 370, + 165 + ], + [ + 361, + 146 + ] + ] + ] + }, + { + "task_id": "0000681", + "id": "RH20T_cfg1_task_0046_user_0001_scene_0003_cfg_0001_039422060546#155", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000681/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 322, + 328 + ], + [ + 310, + 360 + ], + [ + 316, + 330 + ], + [ + 329, + 325 + ], + [ + 327, + 293 + ], + [ + 342, + 282 + ] + ], + [ + [ + 403, + 330 + ], + [ + 363, + 308 + ], + [ + 423, + 294 + ], + [ + 381, + 360 + ], + [ + 464, + 316 + ], + [ + 425, + 360 + ] + ], + [ + [ + 380, + 360 + ], + [ + 405, + 360 + ], + [ + 351, + 360 + ], + [ + 407, + 345 + ], + [ + 323, + 360 + ], + [ + 382, + 315 + ] + ], + [ + [ + 360, + 318 + ], + [ + 360, + 293 + ], + [ + 403, + 325 + ], + [ + 318, + 312 + ], + [ + 399, + 347 + ], + [ + 320, + 334 + ] + ] + ] + }, + { + "task_id": "0000684", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0005_cfg_0005_036422060215#220", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000684/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 254, + 218 + ], + [ + 236, + 234 + ], + [ + 241, + 204 + ], + [ + 269, + 233 + ], + [ + 258, + 189 + ], + [ + 287, + 217 + ] + ], + [ + [ + 254, + 209 + ], + [ + 257, + 192 + ], + [ + 278, + 208 + ], + [ + 229, + 209 + ], + [ + 275, + 224 + ], + [ + 227, + 225 + ] + ], + [ + [ + 278, + 230 + ], + [ + 297, + 245 + ], + [ + 293, + 211 + ], + [ + 262, + 249 + ], + [ + 274, + 196 + ], + [ + 242, + 233 + ] + ], + [ + [ + 252, + 226 + ], + [ + 255, + 212 + ], + [ + 269, + 243 + ], + [ + 235, + 209 + ], + [ + 266, + 258 + ], + [ + 230, + 223 + ] + ] + ] + }, + { + "task_id": "0000685", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0005_cfg_0005_036422060215#415", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000685/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 218, + 158 + ], + [ + 200, + 170 + ], + [ + 228, + 171 + ], + [ + 208, + 145 + ], + [ + 246, + 159 + ], + [ + 226, + 134 + ] + ], + [ + [ + 230, + 127 + ], + [ + 217, + 112 + ], + [ + 222, + 145 + ], + [ + 237, + 110 + ], + [ + 235, + 158 + ], + [ + 249, + 124 + ] + ], + [ + [ + 219, + 141 + ], + [ + 219, + 123 + ], + [ + 241, + 142 + ], + [ + 197, + 140 + ], + [ + 241, + 158 + ], + [ + 198, + 156 + ] + ], + [ + [ + 214, + 169 + ], + [ + 193, + 170 + ], + [ + 207, + 176 + ], + [ + 221, + 161 + ], + [ + 227, + 174 + ], + [ + 242, + 159 + ] + ] + ] + }, + { + "task_id": "0000689", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_038522063145#234", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000689/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 497, + 136 + ], + [ + 516, + 142 + ], + [ + 480, + 135 + ], + [ + 514, + 137 + ], + [ + 464, + 129 + ], + [ + 497, + 131 + ] + ], + [ + [ + 517, + 174 + ], + [ + 514, + 181 + ], + [ + 544, + 180 + ], + [ + 492, + 168 + ], + [ + 546, + 172 + ], + [ + 496, + 161 + ] + ], + [ + [ + 508, + 157 + ], + [ + 515, + 139 + ], + [ + 530, + 169 + ], + [ + 487, + 146 + ], + [ + 522, + 186 + ], + [ + 481, + 162 + ] + ], + [ + [ + 502, + 172 + ], + [ + 525, + 162 + ], + [ + 498, + 180 + ], + [ + 506, + 163 + ], + [ + 476, + 190 + ], + [ + 485, + 172 + ] + ] + ] + }, + { + "task_id": "0000690", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_038522063145#375", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000690/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 491, + 145 + ], + [ + 498, + 127 + ], + [ + 512, + 156 + ], + [ + 471, + 134 + ], + [ + 505, + 173 + ], + [ + 466, + 151 + ] + ], + [ + [ + 495, + 149 + ], + [ + 499, + 150 + ], + [ + 512, + 131 + ], + [ + 479, + 167 + ], + [ + 509, + 129 + ], + [ + 475, + 166 + ] + ], + [ + [ + 505, + 156 + ], + [ + 521, + 141 + ], + [ + 487, + 138 + ], + [ + 523, + 174 + ], + [ + 470, + 152 + ], + [ + 507, + 189 + ] + ], + [ + [ + 518, + 146 + ], + [ + 494, + 140 + ], + [ + 523, + 129 + ], + [ + 513, + 165 + ], + [ + 547, + 135 + ], + [ + 537, + 172 + ] + ] + ] + }, + { + "task_id": "0000694", + "id": "RH20T_cfg1_task_0066_user_0010_scene_0004_cfg_0001_819112070044#550", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000694/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 250, + 301 + ], + [ + 241, + 280 + ], + [ + 251, + 311 + ], + [ + 249, + 289 + ], + [ + 261, + 333 + ], + [ + 260, + 312 + ] + ], + [ + [ + 239, + 301 + ], + [ + 237, + 286 + ], + [ + 263, + 290 + ], + [ + 214, + 312 + ], + [ + 264, + 304 + ], + [ + 217, + 325 + ] + ], + [ + [ + 235, + 291 + ], + [ + 239, + 293 + ], + [ + 215, + 278 + ], + [ + 255, + 304 + ], + [ + 211, + 275 + ], + [ + 252, + 302 + ] + ], + [ + [ + 261, + 283 + ], + [ + 250, + 264 + ], + [ + 251, + 273 + ], + [ + 270, + 292 + ], + [ + 263, + 293 + ], + [ + 281, + 311 + ] + ] + ] + }, + { + "task_id": "0000695", + "id": "RH20T_cfg1_task_0066_user_0010_scene_0004_cfg_0001_819112070044#729", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000695/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 268, + 264 + ], + [ + 249, + 247 + ], + [ + 251, + 282 + ], + [ + 284, + 245 + ], + [ + 269, + 299 + ], + [ + 303, + 261 + ] + ], + [ + [ + 268, + 267 + ], + [ + 256, + 289 + ], + [ + 286, + 276 + ], + [ + 249, + 257 + ], + [ + 297, + 255 + ], + [ + 261, + 235 + ] + ], + [ + [ + 267, + 280 + ], + [ + 266, + 265 + ], + [ + 279, + 262 + ], + [ + 254, + 299 + ], + [ + 280, + 276 + ], + [ + 255, + 312 + ] + ], + [ + [ + 276, + 258 + ], + [ + 299, + 266 + ], + [ + 283, + 235 + ], + [ + 270, + 280 + ], + [ + 260, + 228 + ], + [ + 248, + 273 + ] + ] + ] + }, + { + "task_id": "0000696", + "id": "RH20T_cfg1_task_0066_user_0010_scene_0004_cfg_0001_819112070044#985", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000696/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 162, + 250 + ], + [ + 147, + 237 + ], + [ + 178, + 233 + ], + [ + 145, + 268 + ], + [ + 193, + 247 + ], + [ + 161, + 282 + ] + ], + [ + [ + 173, + 234 + ], + [ + 196, + 234 + ], + [ + 167, + 240 + ], + [ + 179, + 228 + ], + [ + 143, + 241 + ], + [ + 157, + 228 + ] + ], + [ + [ + 139, + 253 + ], + [ + 147, + 268 + ], + [ + 153, + 263 + ], + [ + 124, + 241 + ], + [ + 146, + 249 + ], + [ + 118, + 227 + ] + ], + [ + [ + 165, + 255 + ], + [ + 162, + 238 + ], + [ + 189, + 247 + ], + [ + 141, + 262 + ], + [ + 191, + 263 + ], + [ + 145, + 278 + ] + ] + ] + }, + { + "task_id": "0000698", + "id": "RH20T_cfg3_task_0059_user_0010_scene_0002_cfg_0003_104122062295#171", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000698/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 291, + 161 + ], + [ + 287, + 135 + ], + [ + 256, + 155 + ], + [ + 328, + 168 + ], + [ + 262, + 178 + ], + [ + 330, + 191 + ] + ], + [ + [ + 278, + 151 + ], + [ + 304, + 127 + ], + [ + 296, + 173 + ], + [ + 258, + 126 + ], + [ + 271, + 197 + ], + [ + 230, + 152 + ] + ], + [ + [ + 315, + 139 + ], + [ + 324, + 116 + ], + [ + 324, + 169 + ], + [ + 304, + 106 + ], + [ + 315, + 196 + ], + [ + 293, + 133 + ] + ], + [ + [ + 263, + 174 + ], + [ + 237, + 206 + ], + [ + 231, + 149 + ], + [ + 295, + 198 + ], + [ + 256, + 118 + ], + [ + 320, + 165 + ] + ] + ] + }, + { + "task_id": "0000699", + "id": "RH20T_cfg6_task_0060_user_0014_scene_0006_cfg_0006_104122061602#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to decrease the volume of the speaker'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000699/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 438, + 153 + ], + [ + 417, + 160 + ], + [ + 422, + 140 + ], + [ + 456, + 168 + ], + [ + 442, + 134 + ], + [ + 477, + 161 + ] + ], + [ + [ + 445, + 185 + ], + [ + 456, + 165 + ], + [ + 450, + 197 + ], + [ + 440, + 175 + ], + [ + 439, + 217 + ], + [ + 429, + 194 + ] + ], + [ + [ + 447, + 161 + ], + [ + 447, + 172 + ], + [ + 432, + 179 + ], + [ + 462, + 143 + ], + [ + 432, + 167 + ], + [ + 463, + 130 + ] + ], + [ + [ + 446, + 176 + ], + [ + 463, + 186 + ], + [ + 445, + 200 + ], + [ + 447, + 152 + ], + [ + 430, + 190 + ], + [ + 431, + 144 + ] + ] + ] + }, + { + "task_id": "0000700", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_819112071037#302", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000700/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 352, + 89 + ], + [ + 355, + 8 + ], + [ + 276, + 88 + ], + [ + 427, + 90 + ], + [ + 277, + 159 + ], + [ + 420, + 161 + ] + ], + [ + [ + 321, + 64 + ], + [ + 249, + 119 + ], + [ + 301, + 76 + ], + [ + 351, + 46 + ], + [ + 359, + 33 + ], + [ + 433, + 0 + ] + ], + [ + [ + 442, + 30 + ], + [ + 350, + 33 + ], + [ + 425, + 0 + ], + [ + 463, + 104 + ], + [ + 507, + 0 + ], + [ + 562, + 100 + ] + ], + [ + [ + 366, + 42 + ], + [ + 351, + 0 + ], + [ + 432, + 20 + ], + [ + 302, + 63 + ], + [ + 452, + 79 + ], + [ + 315, + 121 + ] + ] + ] + }, + { + "task_id": "0000702", + "id": "RH20T_cfg1_task_0002_user_0001_scene_0006_cfg_0001_039422060546#192", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000702/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 144 + ], + [ + 286, + 146 + ], + [ + 289, + 174 + ], + [ + 280, + 115 + ], + [ + 287, + 174 + ], + [ + 278, + 111 + ] + ], + [ + [ + 265, + 145 + ], + [ + 241, + 164 + ], + [ + 254, + 128 + ], + [ + 277, + 163 + ], + [ + 277, + 109 + ], + [ + 303, + 142 + ] + ], + [ + [ + 298, + 128 + ], + [ + 291, + 105 + ], + [ + 285, + 152 + ], + [ + 309, + 106 + ], + [ + 293, + 172 + ], + [ + 314, + 127 + ] + ], + [ + [ + 266, + 106 + ], + [ + 271, + 118 + ], + [ + 298, + 103 + ], + [ + 234, + 109 + ], + [ + 295, + 89 + ], + [ + 227, + 95 + ] + ] + ] + }, + { + "task_id": "0000704", + "id": "RH20T_cfg7_task_0016_user_0020_scene_0004_cfg_0007_104122061330#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000704/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 342, + 138 + ], + [ + 327, + 157 + ], + [ + 349, + 157 + ], + [ + 334, + 118 + ], + [ + 362, + 140 + ], + [ + 349, + 101 + ] + ], + [ + [ + 346, + 137 + ], + [ + 372, + 138 + ], + [ + 357, + 143 + ], + [ + 334, + 131 + ], + [ + 333, + 143 + ], + [ + 309, + 131 + ] + ], + [ + [ + 334, + 116 + ], + [ + 350, + 122 + ], + [ + 353, + 107 + ], + [ + 313, + 126 + ], + [ + 338, + 103 + ], + [ + 298, + 120 + ] + ], + [ + [ + 354, + 135 + ], + [ + 336, + 154 + ], + [ + 367, + 148 + ], + [ + 343, + 123 + ], + [ + 387, + 128 + ], + [ + 361, + 105 + ] + ] + ] + }, + { + "task_id": "0000708", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122061018#142", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000708/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 292, + 85 + ], + [ + 278, + 84 + ], + [ + 314, + 79 + ], + [ + 268, + 91 + ], + [ + 331, + 80 + ], + [ + 283, + 93 + ] + ], + [ + [ + 303, + 90 + ], + [ + 322, + 90 + ], + [ + 290, + 71 + ], + [ + 315, + 109 + ], + [ + 272, + 72 + ], + [ + 297, + 109 + ] + ], + [ + [ + 297, + 91 + ], + [ + 276, + 103 + ], + [ + 285, + 67 + ], + [ + 310, + 114 + ], + [ + 305, + 56 + ], + [ + 329, + 102 + ] + ], + [ + [ + 286, + 97 + ], + [ + 287, + 84 + ], + [ + 274, + 75 + ], + [ + 299, + 120 + ], + [ + 273, + 88 + ], + [ + 296, + 131 + ] + ] + ] + }, + { + "task_id": "0000714", + "id": "RH20T_cfg2_task_0065_user_0007_scene_0012_cfg_0002_037522062165#166", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000714/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 188, + 346 + ], + [ + 161, + 338 + ], + [ + 198, + 317 + ], + [ + 178, + 360 + ], + [ + 224, + 325 + ], + [ + 206, + 360 + ] + ], + [ + [ + 174, + 341 + ], + [ + 169, + 310 + ], + [ + 181, + 343 + ], + [ + 167, + 339 + ], + [ + 186, + 360 + ], + [ + 172, + 360 + ] + ], + [ + [ + 173, + 323 + ], + [ + 184, + 294 + ], + [ + 162, + 329 + ], + [ + 182, + 317 + ], + [ + 150, + 360 + ], + [ + 171, + 344 + ] + ], + [ + [ + 192, + 317 + ], + [ + 188, + 305 + ], + [ + 171, + 301 + ], + [ + 214, + 334 + ], + [ + 176, + 313 + ], + [ + 217, + 345 + ] + ] + ] + }, + { + "task_id": "0000717", + "id": "RH20T_cfg2_task_0065_user_0007_scene_0012_cfg_0002_037522062165#409", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000717/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 212, + 263 + ], + [ + 225, + 283 + ], + [ + 191, + 279 + ], + [ + 233, + 247 + ], + [ + 178, + 259 + ], + [ + 220, + 227 + ] + ], + [ + [ + 232, + 264 + ], + [ + 251, + 250 + ], + [ + 248, + 285 + ], + [ + 216, + 244 + ], + [ + 228, + 302 + ], + [ + 195, + 259 + ] + ], + [ + [ + 218, + 266 + ], + [ + 213, + 254 + ], + [ + 197, + 253 + ], + [ + 240, + 281 + ], + [ + 202, + 265 + ], + [ + 243, + 292 + ] + ], + [ + [ + 194, + 278 + ], + [ + 209, + 259 + ], + [ + 182, + 286 + ], + [ + 208, + 270 + ], + [ + 166, + 307 + ], + [ + 191, + 292 + ] + ] + ] + }, + { + "task_id": "0000720", + "id": "RH20T_cfg2_task_0065_user_0007_scene_0012_cfg_0002_037522062165#557", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000720/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 240, + 278 + ], + [ + 237, + 265 + ], + [ + 220, + 263 + ], + [ + 260, + 294 + ], + [ + 224, + 275 + ], + [ + 262, + 306 + ] + ], + [ + [ + 224, + 253 + ], + [ + 213, + 279 + ], + [ + 198, + 245 + ], + [ + 250, + 262 + ], + [ + 209, + 221 + ], + [ + 260, + 237 + ] + ], + [ + [ + 247, + 256 + ], + [ + 268, + 262 + ], + [ + 251, + 229 + ], + [ + 243, + 283 + ], + [ + 228, + 222 + ], + [ + 219, + 278 + ] + ], + [ + [ + 240, + 302 + ], + [ + 231, + 296 + ], + [ + 224, + 281 + ], + [ + 256, + 323 + ], + [ + 233, + 287 + ], + [ + 266, + 331 + ] + ] + ] + }, + { + "task_id": "0000724", + "id": "RH20T_cfg6_task_0013_user_0014_scene_0007_cfg_0006_104122061330#121", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000724/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 254, + 151 + ], + [ + 246, + 144 + ], + [ + 243, + 174 + ], + [ + 264, + 128 + ], + [ + 251, + 178 + ], + [ + 270, + 135 + ] + ], + [ + [ + 254, + 152 + ], + [ + 261, + 161 + ], + [ + 231, + 162 + ], + [ + 277, + 142 + ], + [ + 222, + 152 + ], + [ + 270, + 131 + ] + ], + [ + [ + 253, + 173 + ], + [ + 254, + 158 + ], + [ + 244, + 191 + ], + [ + 263, + 153 + ], + [ + 243, + 207 + ], + [ + 263, + 168 + ] + ], + [ + [ + 254, + 154 + ], + [ + 255, + 176 + ], + [ + 229, + 155 + ], + [ + 279, + 153 + ], + [ + 228, + 134 + ], + [ + 277, + 132 + ] + ] + ] + }, + { + "task_id": "0000725", + "id": "RH20T_cfg7_task_0001_user_0014_scene_0007_cfg_0007_037522061512#124", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000725/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 250, + 111 + ], + [ + 249, + 134 + ], + [ + 246, + 113 + ], + [ + 255, + 109 + ], + [ + 247, + 92 + ], + [ + 257, + 85 + ] + ], + [ + [ + 261, + 123 + ], + [ + 261, + 145 + ], + [ + 259, + 120 + ], + [ + 264, + 126 + ], + [ + 260, + 99 + ], + [ + 264, + 103 + ] + ], + [ + [ + 246, + 109 + ], + [ + 238, + 94 + ], + [ + 257, + 93 + ], + [ + 234, + 126 + ], + [ + 264, + 107 + ], + [ + 242, + 139 + ] + ], + [ + [ + 254, + 98 + ], + [ + 233, + 92 + ], + [ + 249, + 121 + ], + [ + 259, + 75 + ], + [ + 269, + 126 + ], + [ + 279, + 81 + ] + ] + ] + }, + { + "task_id": "0000726", + "id": "RH20T_cfg5_task_0020_user_0010_scene_0007_cfg_0005_104122063678#303", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000726/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 297, + 187 + ], + [ + 278, + 188 + ], + [ + 303, + 167 + ], + [ + 292, + 206 + ], + [ + 321, + 166 + ], + [ + 310, + 203 + ] + ], + [ + [ + 312, + 177 + ], + [ + 324, + 164 + ], + [ + 314, + 196 + ], + [ + 310, + 159 + ], + [ + 302, + 208 + ], + [ + 298, + 172 + ] + ], + [ + [ + 321, + 187 + ], + [ + 298, + 188 + ], + [ + 326, + 199 + ], + [ + 316, + 173 + ], + [ + 348, + 198 + ], + [ + 341, + 171 + ] + ], + [ + [ + 338, + 184 + ], + [ + 356, + 193 + ], + [ + 327, + 205 + ], + [ + 348, + 163 + ], + [ + 308, + 197 + ], + [ + 330, + 154 + ] + ] + ] + }, + { + "task_id": "0000728", + "id": "RH20T_cfg4_task_0024_user_0014_scene_0007_cfg_0004_104122062295#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn off the power strip by pressing the button'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000728/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 211, + 166 + ], + [ + 227, + 140 + ], + [ + 171, + 152 + ], + [ + 248, + 180 + ], + [ + 159, + 177 + ], + [ + 232, + 203 + ] + ], + [ + [ + 220, + 157 + ], + [ + 224, + 194 + ], + [ + 214, + 155 + ], + [ + 225, + 158 + ], + [ + 210, + 114 + ], + [ + 221, + 124 + ] + ], + [ + [ + 203, + 178 + ], + [ + 182, + 176 + ], + [ + 195, + 143 + ], + [ + 211, + 214 + ], + [ + 213, + 148 + ], + [ + 228, + 214 + ] + ], + [ + [ + 176, + 194 + ], + [ + 184, + 166 + ], + [ + 174, + 164 + ], + [ + 178, + 221 + ], + [ + 165, + 196 + ], + [ + 171, + 253 + ] + ] + ] + }, + { + "task_id": "0000729", + "id": "RH20T_cfg6_task_0039_user_0014_scene_0003_cfg_0006_104122063633#164", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000729/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 220, + 196 + ], + [ + 222, + 175 + ], + [ + 224, + 180 + ], + [ + 216, + 215 + ], + [ + 222, + 199 + ], + [ + 214, + 235 + ] + ], + [ + [ + 206, + 199 + ], + [ + 185, + 194 + ], + [ + 217, + 176 + ], + [ + 195, + 222 + ], + [ + 237, + 182 + ], + [ + 215, + 225 + ] + ], + [ + [ + 245, + 194 + ], + [ + 225, + 182 + ], + [ + 232, + 217 + ], + [ + 259, + 171 + ], + [ + 253, + 230 + ], + [ + 280, + 182 + ] + ], + [ + [ + 243, + 183 + ], + [ + 232, + 159 + ], + [ + 220, + 190 + ], + [ + 266, + 176 + ], + [ + 231, + 212 + ], + [ + 276, + 200 + ] + ] + ] + }, + { + "task_id": "0000730", + "id": "RH20T_cfg1_task_0029_user_0001_scene_0004_cfg_0001_038522062547#219", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000730/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 461, + 113 + ], + [ + 471, + 123 + ], + [ + 469, + 133 + ], + [ + 453, + 94 + ], + [ + 457, + 123 + ], + [ + 442, + 82 + ] + ], + [ + [ + 471, + 90 + ], + [ + 489, + 75 + ], + [ + 481, + 112 + ], + [ + 460, + 68 + ], + [ + 463, + 125 + ], + [ + 444, + 83 + ] + ], + [ + [ + 424, + 108 + ], + [ + 406, + 127 + ], + [ + 407, + 90 + ], + [ + 442, + 126 + ], + [ + 426, + 71 + ], + [ + 460, + 107 + ] + ], + [ + [ + 448, + 93 + ], + [ + 461, + 77 + ], + [ + 453, + 115 + ], + [ + 443, + 73 + ], + [ + 441, + 129 + ], + [ + 432, + 88 + ] + ] + ] + }, + { + "task_id": "0000731", + "id": "RH20T_cfg1_task_0043_user_0016_scene_0005_cfg_0001_750612070853#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000731/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 535, + 248 + ], + [ + 551, + 261 + ], + [ + 450, + 245 + ], + [ + 630, + 252 + ], + [ + 451, + 237 + ], + [ + 603, + 242 + ] + ], + [ + [ + 605, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 520, + 360 + ], + [ + 632, + 312 + ], + [ + 457, + 360 + ] + ], + [ + [ + 571, + 360 + ], + [ + 598, + 325 + ], + [ + 486, + 330 + ], + [ + 640, + 360 + ], + [ + 475, + 360 + ], + [ + 640, + 360 + ] + ], + [ + [ + 557, + 360 + ], + [ + 557, + 343 + ], + [ + 554, + 360 + ], + [ + 559, + 265 + ], + [ + 553, + 360 + ], + [ + 559, + 273 + ] + ] + ] + }, + { + "task_id": "0000733", + "id": "RH20T_cfg2_task_0005_user_0013_scene_0002_cfg_0002_104122061850#193", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000733/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 426, + 158 + ], + [ + 411, + 162 + ], + [ + 433, + 181 + ], + [ + 418, + 135 + ], + [ + 450, + 178 + ], + [ + 434, + 130 + ] + ], + [ + [ + 430, + 141 + ], + [ + 444, + 121 + ], + [ + 445, + 146 + ], + [ + 414, + 137 + ], + [ + 431, + 165 + ], + [ + 401, + 157 + ] + ], + [ + [ + 428, + 153 + ], + [ + 426, + 139 + ], + [ + 410, + 170 + ], + [ + 445, + 137 + ], + [ + 412, + 182 + ], + [ + 446, + 150 + ] + ], + [ + [ + 432, + 126 + ], + [ + 434, + 106 + ], + [ + 443, + 112 + ], + [ + 423, + 139 + ], + [ + 441, + 134 + ], + [ + 421, + 160 + ] + ] + ] + }, + { + "task_id": "0000737", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_104422070044#79", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000737/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 437, + 177 + ], + [ + 436, + 160 + ], + [ + 455, + 185 + ], + [ + 419, + 169 + ], + [ + 455, + 200 + ], + [ + 420, + 184 + ] + ], + [ + [ + 426, + 163 + ], + [ + 415, + 155 + ], + [ + 433, + 180 + ], + [ + 419, + 146 + ], + [ + 443, + 187 + ], + [ + 429, + 154 + ] + ], + [ + [ + 413, + 174 + ], + [ + 421, + 163 + ], + [ + 434, + 179 + ], + [ + 393, + 170 + ], + [ + 426, + 192 + ], + [ + 384, + 182 + ] + ], + [ + [ + 438, + 168 + ], + [ + 454, + 177 + ], + [ + 444, + 148 + ], + [ + 431, + 187 + ], + [ + 428, + 140 + ], + [ + 416, + 178 + ] + ] + ] + }, + { + "task_id": "0000739", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0002_cfg_0003_104122062295#180", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000739/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 217, + 203 + ], + [ + 211, + 230 + ], + [ + 181, + 201 + ], + [ + 253, + 205 + ], + [ + 189, + 178 + ], + [ + 257, + 182 + ] + ], + [ + [ + 220, + 199 + ], + [ + 193, + 219 + ], + [ + 232, + 193 + ], + [ + 210, + 204 + ], + [ + 258, + 173 + ], + [ + 234, + 186 + ] + ], + [ + [ + 237, + 172 + ], + [ + 247, + 165 + ], + [ + 259, + 145 + ], + [ + 214, + 200 + ], + [ + 249, + 152 + ], + [ + 207, + 204 + ] + ], + [ + [ + 205, + 194 + ], + [ + 206, + 164 + ], + [ + 167, + 187 + ], + [ + 243, + 202 + ], + [ + 169, + 215 + ], + [ + 240, + 229 + ] + ] + ] + }, + { + "task_id": "0000740", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0002_cfg_0003_104122062295#220", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000740/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 211, + 207 + ], + [ + 182, + 229 + ], + [ + 186, + 176 + ], + [ + 235, + 237 + ], + [ + 213, + 157 + ], + [ + 259, + 216 + ] + ], + [ + [ + 197, + 220 + ], + [ + 198, + 188 + ], + [ + 159, + 210 + ], + [ + 236, + 230 + ], + [ + 160, + 239 + ], + [ + 234, + 259 + ] + ], + [ + [ + 206, + 200 + ], + [ + 220, + 167 + ], + [ + 166, + 197 + ], + [ + 247, + 204 + ], + [ + 148, + 233 + ], + [ + 233, + 243 + ] + ], + [ + [ + 205, + 215 + ], + [ + 201, + 258 + ], + [ + 202, + 220 + ], + [ + 207, + 211 + ], + [ + 206, + 172 + ], + [ + 210, + 171 + ] + ] + ] + }, + { + "task_id": "0000743", + "id": "RH20T_cfg4_task_0081_user_0007_scene_0004_cfg_0004_036422060909#157", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000743/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 320, + 104 + ], + [ + 324, + 112 + ], + [ + 344, + 109 + ], + [ + 297, + 100 + ], + [ + 341, + 101 + ], + [ + 292, + 92 + ] + ], + [ + [ + 344, + 86 + ], + [ + 346, + 93 + ], + [ + 366, + 96 + ], + [ + 321, + 76 + ], + [ + 364, + 90 + ], + [ + 321, + 71 + ] + ], + [ + [ + 320, + 104 + ], + [ + 322, + 87 + ], + [ + 333, + 89 + ], + [ + 306, + 120 + ], + [ + 330, + 104 + ], + [ + 304, + 134 + ] + ], + [ + [ + 311, + 118 + ], + [ + 306, + 98 + ], + [ + 293, + 129 + ], + [ + 330, + 106 + ], + [ + 298, + 150 + ], + [ + 336, + 128 + ] + ] + ] + }, + { + "task_id": "0000746", + "id": "RH20T_cfg1_task_0092_user_0010_scene_0009_cfg_0001_750612070853#147", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the hands of a clock'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000746/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 392, + 317 + ], + [ + 441, + 324 + ], + [ + 413, + 295 + ], + [ + 367, + 343 + ], + [ + 373, + 291 + ], + [ + 322, + 335 + ] + ], + [ + [ + 406, + 244 + ], + [ + 454, + 224 + ], + [ + 426, + 257 + ], + [ + 381, + 229 + ], + [ + 385, + 273 + ], + [ + 334, + 249 + ] + ], + [ + [ + 363, + 292 + ], + [ + 358, + 258 + ], + [ + 322, + 273 + ], + [ + 410, + 313 + ], + [ + 329, + 302 + ], + [ + 412, + 342 + ] + ], + [ + [ + 341, + 286 + ], + [ + 305, + 316 + ], + [ + 379, + 320 + ], + [ + 306, + 256 + ], + [ + 410, + 290 + ], + [ + 339, + 233 + ] + ] + ] + }, + { + "task_id": "0000751", + "id": "RH20T_cfg1_task_0089_user_0014_scene_0002_cfg_0001_750612070851#128", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push an object with obstacles'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000751/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 460, + 346 + ], + [ + 449, + 307 + ], + [ + 460, + 338 + ], + [ + 460, + 352 + ], + [ + 473, + 360 + ], + [ + 470, + 360 + ] + ], + [ + [ + 472, + 309 + ], + [ + 486, + 293 + ], + [ + 436, + 317 + ], + [ + 507, + 300 + ], + [ + 426, + 330 + ], + [ + 492, + 314 + ] + ], + [ + [ + 504, + 333 + ], + [ + 525, + 360 + ], + [ + 471, + 320 + ], + [ + 535, + 346 + ], + [ + 456, + 294 + ], + [ + 515, + 320 + ] + ], + [ + [ + 465, + 328 + ], + [ + 450, + 295 + ], + [ + 483, + 308 + ], + [ + 445, + 351 + ], + [ + 497, + 338 + ], + [ + 460, + 360 + ] + ] + ] + }, + { + "task_id": "0000753", + "id": "RH20T_cfg6_task_0129_user_0020_scene_0009_cfg_0006_104122061602#225", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000753/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 422, + 121 + ], + [ + 438, + 120 + ], + [ + 422, + 143 + ], + [ + 422, + 98 + ], + [ + 407, + 143 + ], + [ + 407, + 100 + ] + ], + [ + [ + 419, + 142 + ], + [ + 437, + 144 + ], + [ + 422, + 120 + ], + [ + 416, + 165 + ], + [ + 404, + 118 + ], + [ + 398, + 164 + ] + ], + [ + [ + 411, + 108 + ], + [ + 396, + 95 + ], + [ + 398, + 113 + ], + [ + 425, + 103 + ], + [ + 412, + 125 + ], + [ + 441, + 117 + ] + ], + [ + [ + 450, + 123 + ], + [ + 442, + 125 + ], + [ + 434, + 107 + ], + [ + 466, + 139 + ], + [ + 441, + 106 + ], + [ + 471, + 137 + ] + ] + ] + }, + { + "task_id": "0000754", + "id": "RH20T_cfg1_task_0051_user_0015_scene_0002_cfg_0001_750612070851#212", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the dish off the dish rack'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000754/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 255, + 188 + ], + [ + 226, + 184 + ], + [ + 232, + 224 + ], + [ + 280, + 151 + ], + [ + 258, + 225 + ], + [ + 303, + 157 + ] + ], + [ + [ + 280, + 232 + ], + [ + 245, + 231 + ], + [ + 259, + 210 + ], + [ + 299, + 251 + ], + [ + 299, + 210 + ], + [ + 336, + 253 + ] + ], + [ + [ + 219, + 183 + ], + [ + 186, + 160 + ], + [ + 214, + 146 + ], + [ + 223, + 214 + ], + [ + 253, + 170 + ], + [ + 256, + 238 + ] + ], + [ + [ + 287, + 239 + ], + [ + 317, + 269 + ], + [ + 300, + 245 + ], + [ + 271, + 231 + ], + [ + 274, + 219 + ], + [ + 241, + 201 + ] + ] + ] + }, + { + "task_id": "0000756", + "id": "RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_104122061018#168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000756/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 332, + 213 + ], + [ + 297, + 212 + ], + [ + 326, + 248 + ], + [ + 339, + 177 + ], + [ + 358, + 249 + ], + [ + 372, + 180 + ] + ], + [ + [ + 319, + 203 + ], + [ + 341, + 193 + ], + [ + 327, + 170 + ], + [ + 312, + 233 + ], + [ + 301, + 179 + ], + [ + 288, + 246 + ] + ], + [ + [ + 361, + 212 + ], + [ + 330, + 221 + ], + [ + 346, + 234 + ], + [ + 378, + 187 + ], + [ + 373, + 225 + ], + [ + 406, + 179 + ] + ], + [ + [ + 344, + 226 + ], + [ + 328, + 212 + ], + [ + 348, + 257 + ], + [ + 339, + 192 + ], + [ + 366, + 274 + ], + [ + 358, + 205 + ] + ] + ] + }, + { + "task_id": "0000758", + "id": "RH20T_cfg6_task_0038_user_0020_scene_0001_cfg_0006_104122061330#548", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000758/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 345, + 141 + ], + [ + 332, + 159 + ], + [ + 347, + 160 + ], + [ + 343, + 120 + ], + [ + 358, + 142 + ], + [ + 355, + 102 + ] + ], + [ + [ + 345, + 133 + ], + [ + 358, + 148 + ], + [ + 325, + 128 + ], + [ + 363, + 137 + ], + [ + 313, + 114 + ], + [ + 350, + 123 + ] + ], + [ + [ + 338, + 126 + ], + [ + 311, + 124 + ], + [ + 333, + 146 + ], + [ + 342, + 109 + ], + [ + 362, + 148 + ], + [ + 368, + 111 + ] + ], + [ + [ + 330, + 101 + ], + [ + 343, + 115 + ], + [ + 355, + 93 + ], + [ + 307, + 109 + ], + [ + 342, + 77 + ], + [ + 292, + 95 + ] + ] + ] + }, + { + "task_id": "0000762", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0005_cfg_0003_104122062823#360", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000762/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 180, + 84 + ], + [ + 196, + 64 + ], + [ + 202, + 107 + ], + [ + 156, + 59 + ], + [ + 187, + 129 + ], + [ + 138, + 81 + ] + ], + [ + [ + 214, + 85 + ], + [ + 217, + 93 + ], + [ + 235, + 61 + ], + [ + 193, + 109 + ], + [ + 231, + 56 + ], + [ + 191, + 101 + ] + ], + [ + [ + 219, + 36 + ], + [ + 249, + 41 + ], + [ + 233, + 11 + ], + [ + 207, + 59 + ], + [ + 199, + 4 + ], + [ + 176, + 55 + ] + ], + [ + [ + 193, + 53 + ], + [ + 187, + 24 + ], + [ + 222, + 59 + ], + [ + 165, + 47 + ], + [ + 226, + 86 + ], + [ + 172, + 73 + ] + ] + ] + }, + { + "task_id": "0000764", + "id": "RH20T_cfg4_task_0006_user_0010_scene_0008_cfg_0004_038522062288#369", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000764/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 122, + 195 + ], + [ + 88, + 191 + ], + [ + 125, + 165 + ], + [ + 119, + 226 + ], + [ + 156, + 169 + ], + [ + 150, + 228 + ] + ], + [ + [ + 148, + 225 + ], + [ + 149, + 250 + ], + [ + 147, + 210 + ], + [ + 149, + 241 + ], + [ + 145, + 185 + ], + [ + 147, + 213 + ] + ], + [ + [ + 128, + 203 + ], + [ + 128, + 220 + ], + [ + 156, + 216 + ], + [ + 100, + 190 + ], + [ + 157, + 197 + ], + [ + 98, + 171 + ] + ], + [ + [ + 136, + 213 + ], + [ + 124, + 196 + ], + [ + 118, + 197 + ], + [ + 154, + 229 + ], + [ + 129, + 213 + ], + [ + 164, + 244 + ] + ] + ] + }, + { + "task_id": "0000768", + "id": "RH20T_cfg5_task_0001_user_0014_scene_0002_cfg_0005_104122063678#258", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000768/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 356, + 188 + ], + [ + 363, + 206 + ], + [ + 330, + 192 + ], + [ + 383, + 185 + ], + [ + 325, + 175 + ], + [ + 376, + 169 + ] + ], + [ + [ + 357, + 186 + ], + [ + 363, + 169 + ], + [ + 364, + 206 + ], + [ + 351, + 168 + ], + [ + 358, + 221 + ], + [ + 346, + 184 + ] + ], + [ + [ + 344, + 161 + ], + [ + 366, + 157 + ], + [ + 330, + 150 + ], + [ + 358, + 171 + ], + [ + 310, + 154 + ], + [ + 338, + 174 + ] + ], + [ + [ + 371, + 192 + ], + [ + 395, + 199 + ], + [ + 371, + 198 + ], + [ + 371, + 186 + ], + [ + 349, + 192 + ], + [ + 346, + 179 + ] + ] + ] + }, + { + "task_id": "0000769", + "id": "RH20T_cfg6_task_0026_user_0014_scene_0007_cfg_0006_037522061512#115", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push and close the drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000769/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 219, + 164 + ], + [ + 217, + 139 + ], + [ + 242, + 166 + ], + [ + 194, + 162 + ], + [ + 244, + 192 + ], + [ + 195, + 189 + ] + ], + [ + [ + 218, + 137 + ], + [ + 207, + 154 + ], + [ + 205, + 148 + ], + [ + 230, + 127 + ], + [ + 216, + 129 + ], + [ + 241, + 109 + ] + ], + [ + [ + 219, + 143 + ], + [ + 213, + 118 + ], + [ + 239, + 130 + ], + [ + 198, + 156 + ], + [ + 244, + 153 + ], + [ + 205, + 179 + ] + ], + [ + [ + 198, + 157 + ], + [ + 196, + 130 + ], + [ + 177, + 159 + ], + [ + 221, + 155 + ], + [ + 180, + 186 + ], + [ + 223, + 184 + ] + ] + ] + }, + { + "task_id": "0000770", + "id": "RH20T_cfg1_task_0061_user_0014_scene_0008_cfg_0001_750612070851#97", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000770/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 422, + 163 + ], + [ + 430, + 140 + ], + [ + 403, + 185 + ], + [ + 439, + 142 + ], + [ + 396, + 204 + ], + [ + 431, + 163 + ] + ], + [ + [ + 410, + 148 + ], + [ + 440, + 154 + ], + [ + 409, + 119 + ], + [ + 411, + 176 + ], + [ + 380, + 115 + ], + [ + 384, + 170 + ] + ], + [ + [ + 459, + 157 + ], + [ + 481, + 128 + ], + [ + 430, + 135 + ], + [ + 488, + 179 + ], + [ + 408, + 164 + ], + [ + 466, + 208 + ] + ], + [ + [ + 423, + 142 + ], + [ + 402, + 169 + ], + [ + 415, + 146 + ], + [ + 429, + 140 + ], + [ + 438, + 115 + ], + [ + 449, + 113 + ] + ] + ] + }, + { + "task_id": "0000777", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0005_cfg_0004_036422060909#1168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000777/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 310, + 70 + ], + [ + 290, + 88 + ], + [ + 291, + 49 + ], + [ + 330, + 90 + ], + [ + 310, + 32 + ], + [ + 349, + 72 + ] + ], + [ + [ + 330, + 63 + ], + [ + 329, + 39 + ], + [ + 313, + 50 + ], + [ + 347, + 76 + ], + [ + 314, + 72 + ], + [ + 347, + 97 + ] + ], + [ + [ + 300, + 66 + ], + [ + 322, + 82 + ], + [ + 311, + 51 + ], + [ + 289, + 81 + ], + [ + 290, + 36 + ], + [ + 266, + 65 + ] + ], + [ + [ + 351, + 51 + ], + [ + 368, + 39 + ], + [ + 369, + 70 + ], + [ + 333, + 31 + ], + [ + 352, + 83 + ], + [ + 315, + 43 + ] + ] + ] + }, + { + "task_id": "0000778", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0005_cfg_0004_036422060909#3009", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000778/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 377, + 131 + ], + [ + 373, + 123 + ], + [ + 362, + 112 + ], + [ + 392, + 151 + ], + [ + 367, + 121 + ], + [ + 396, + 158 + ] + ], + [ + [ + 370, + 142 + ], + [ + 371, + 120 + ], + [ + 391, + 154 + ], + [ + 350, + 131 + ], + [ + 389, + 175 + ], + [ + 349, + 152 + ] + ], + [ + [ + 388, + 128 + ], + [ + 367, + 139 + ], + [ + 402, + 149 + ], + [ + 376, + 108 + ], + [ + 423, + 137 + ], + [ + 397, + 97 + ] + ], + [ + [ + 346, + 130 + ], + [ + 363, + 140 + ], + [ + 366, + 119 + ], + [ + 328, + 141 + ], + [ + 348, + 108 + ], + [ + 310, + 131 + ] + ] + ] + }, + { + "task_id": "0000780", + "id": "RH20T_cfg7_task_0044_user_0014_scene_0005_cfg_0007_104122061330#88", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000780/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 268, + 224 + ], + [ + 292, + 239 + ], + [ + 285, + 200 + ], + [ + 251, + 248 + ], + [ + 262, + 186 + ], + [ + 228, + 233 + ] + ], + [ + [ + 277, + 225 + ], + [ + 273, + 204 + ], + [ + 264, + 247 + ], + [ + 289, + 205 + ], + [ + 268, + 267 + ], + [ + 292, + 225 + ] + ], + [ + [ + 307, + 206 + ], + [ + 337, + 209 + ], + [ + 303, + 236 + ], + [ + 310, + 176 + ], + [ + 274, + 232 + ], + [ + 282, + 174 + ] + ], + [ + [ + 294, + 236 + ], + [ + 302, + 263 + ], + [ + 302, + 249 + ], + [ + 285, + 222 + ], + [ + 296, + 225 + ], + [ + 278, + 196 + ] + ] + ] + }, + { + "task_id": "0000782", + "id": "RH20T_cfg2_task_0077_user_0014_scene_0006_cfg_0002_037522062165#407", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000782/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 120, + 293 + ], + [ + 135, + 290 + ], + [ + 141, + 272 + ], + [ + 99, + 314 + ], + [ + 125, + 274 + ], + [ + 81, + 318 + ] + ], + [ + [ + 121, + 315 + ], + [ + 99, + 339 + ], + [ + 143, + 332 + ], + [ + 99, + 297 + ], + [ + 163, + 309 + ], + [ + 120, + 275 + ] + ], + [ + [ + 111, + 317 + ], + [ + 88, + 341 + ], + [ + 101, + 306 + ], + [ + 120, + 326 + ], + [ + 124, + 283 + ], + [ + 140, + 305 + ] + ], + [ + [ + 120, + 308 + ], + [ + 113, + 296 + ], + [ + 94, + 298 + ], + [ + 146, + 318 + ], + [ + 102, + 309 + ], + [ + 152, + 328 + ] + ] + ] + }, + { + "task_id": "0000783", + "id": "RH20T_cfg2_task_0077_user_0014_scene_0006_cfg_0002_037522062165#635", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000783/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 160, + 344 + ], + [ + 155, + 334 + ], + [ + 135, + 330 + ], + [ + 186, + 358 + ], + [ + 141, + 339 + ], + [ + 189, + 360 + ] + ], + [ + [ + 143, + 332 + ], + [ + 124, + 359 + ], + [ + 119, + 316 + ], + [ + 166, + 348 + ], + [ + 138, + 291 + ], + [ + 183, + 323 + ] + ], + [ + [ + 181, + 320 + ], + [ + 198, + 295 + ], + [ + 187, + 318 + ], + [ + 175, + 321 + ], + [ + 168, + 347 + ], + [ + 157, + 347 + ] + ], + [ + [ + 182, + 333 + ], + [ + 183, + 304 + ], + [ + 153, + 333 + ], + [ + 210, + 333 + ], + [ + 151, + 360 + ], + [ + 211, + 360 + ] + ] + ] + }, + { + "task_id": "0000784", + "id": "RH20T_cfg2_task_0077_user_0014_scene_0006_cfg_0002_037522062165#802", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000784/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 180, + 308 + ], + [ + 164, + 335 + ], + [ + 206, + 309 + ], + [ + 156, + 306 + ], + [ + 220, + 284 + ], + [ + 171, + 282 + ] + ], + [ + [ + 173, + 342 + ], + [ + 167, + 335 + ], + [ + 148, + 326 + ], + [ + 199, + 359 + ], + [ + 155, + 333 + ], + [ + 203, + 360 + ] + ], + [ + [ + 150, + 323 + ], + [ + 182, + 314 + ], + [ + 156, + 327 + ], + [ + 144, + 319 + ], + [ + 125, + 337 + ], + [ + 108, + 329 + ] + ], + [ + [ + 200, + 333 + ], + [ + 220, + 306 + ], + [ + 178, + 327 + ], + [ + 220, + 338 + ], + [ + 157, + 356 + ], + [ + 201, + 360 + ] + ] + ] + }, + { + "task_id": "0000785", + "id": "RH20T_cfg5_task_0023_user_0010_scene_0007_cfg_0005_104122063678#390", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000785/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 268, + 182 + ], + [ + 280, + 168 + ], + [ + 270, + 168 + ], + [ + 265, + 195 + ], + [ + 256, + 182 + ], + [ + 252, + 209 + ] + ], + [ + [ + 262, + 184 + ], + [ + 271, + 167 + ], + [ + 268, + 202 + ], + [ + 257, + 167 + ], + [ + 260, + 218 + ], + [ + 249, + 183 + ] + ], + [ + [ + 254, + 179 + ], + [ + 253, + 153 + ], + [ + 231, + 180 + ], + [ + 278, + 177 + ], + [ + 232, + 206 + ], + [ + 280, + 202 + ] + ], + [ + [ + 243, + 155 + ], + [ + 245, + 157 + ], + [ + 221, + 142 + ], + [ + 265, + 167 + ], + [ + 218, + 139 + ], + [ + 264, + 165 + ] + ] + ] + }, + { + "task_id": "0000789", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_819112070044#547", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000789/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 303, + 191 + ], + [ + 302, + 170 + ], + [ + 321, + 195 + ], + [ + 285, + 187 + ], + [ + 322, + 217 + ], + [ + 285, + 210 + ] + ], + [ + [ + 312, + 217 + ], + [ + 292, + 207 + ], + [ + 317, + 197 + ], + [ + 307, + 235 + ], + [ + 338, + 208 + ], + [ + 328, + 246 + ] + ], + [ + [ + 289, + 232 + ], + [ + 289, + 213 + ], + [ + 295, + 244 + ], + [ + 283, + 220 + ], + [ + 295, + 264 + ], + [ + 282, + 241 + ] + ], + [ + [ + 291, + 218 + ], + [ + 291, + 200 + ], + [ + 312, + 210 + ], + [ + 270, + 226 + ], + [ + 313, + 227 + ], + [ + 271, + 243 + ] + ] + ] + }, + { + "task_id": "0000791", + "id": "RH20T_cfg5_task_0085_user_0007_scene_0006_cfg_0005_036422060215#263", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the bubble ring from the assembled bubble ring and ball'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000791/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 208, + 205 + ], + [ + 185, + 208 + ], + [ + 200, + 195 + ], + [ + 217, + 215 + ], + [ + 221, + 192 + ], + [ + 239, + 212 + ] + ], + [ + [ + 208, + 174 + ], + [ + 219, + 160 + ], + [ + 204, + 157 + ], + [ + 211, + 191 + ], + [ + 193, + 172 + ], + [ + 201, + 206 + ] + ], + [ + [ + 210, + 195 + ], + [ + 212, + 178 + ], + [ + 223, + 210 + ], + [ + 198, + 180 + ], + [ + 221, + 225 + ], + [ + 196, + 196 + ] + ], + [ + [ + 192, + 206 + ], + [ + 181, + 215 + ], + [ + 171, + 197 + ], + [ + 212, + 215 + ], + [ + 182, + 189 + ], + [ + 221, + 206 + ] + ] + ] + }, + { + "task_id": "0000793", + "id": "RH20T_cfg5_task_0039_user_0007_scene_0003_cfg_0005_104122061850#241", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000793/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 365, + 99 + ], + [ + 361, + 81 + ], + [ + 352, + 118 + ], + [ + 378, + 80 + ], + [ + 357, + 133 + ], + [ + 382, + 97 + ] + ], + [ + [ + 368, + 86 + ], + [ + 357, + 64 + ], + [ + 349, + 98 + ], + [ + 389, + 74 + ], + [ + 360, + 119 + ], + [ + 401, + 97 + ] + ], + [ + [ + 357, + 121 + ], + [ + 345, + 105 + ], + [ + 357, + 137 + ], + [ + 358, + 103 + ], + [ + 369, + 154 + ], + [ + 372, + 121 + ] + ], + [ + [ + 369, + 75 + ], + [ + 367, + 92 + ], + [ + 356, + 91 + ], + [ + 384, + 58 + ], + [ + 359, + 75 + ], + [ + 386, + 42 + ] + ] + ] + }, + { + "task_id": "0000794", + "id": "RH20T_cfg5_task_0059_user_0007_scene_0003_cfg_0005_104122063678#600", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000794/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 259, + 192 + ], + [ + 267, + 217 + ], + [ + 245, + 195 + ], + [ + 272, + 190 + ], + [ + 238, + 169 + ], + [ + 265, + 166 + ] + ], + [ + [ + 288, + 208 + ], + [ + 272, + 196 + ], + [ + 289, + 229 + ], + [ + 287, + 187 + ], + [ + 305, + 242 + ], + [ + 304, + 199 + ] + ], + [ + [ + 275, + 202 + ], + [ + 279, + 185 + ], + [ + 272, + 223 + ], + [ + 277, + 182 + ], + [ + 269, + 238 + ], + [ + 274, + 197 + ] + ], + [ + [ + 289, + 199 + ], + [ + 291, + 185 + ], + [ + 296, + 181 + ], + [ + 283, + 220 + ], + [ + 294, + 194 + ], + [ + 281, + 232 + ] + ] + ] + }, + { + "task_id": "0000796", + "id": "RH20T_cfg4_task_0058_user_0014_scene_0008_cfg_0004_038522062288#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000796/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 85 + ], + [ + 303, + 73 + ], + [ + 281, + 86 + ], + [ + 332, + 84 + ], + [ + 285, + 97 + ], + [ + 334, + 96 + ] + ], + [ + [ + 322, + 83 + ], + [ + 321, + 56 + ], + [ + 296, + 84 + ], + [ + 347, + 82 + ], + [ + 297, + 111 + ], + [ + 346, + 107 + ] + ], + [ + [ + 289, + 93 + ], + [ + 288, + 69 + ], + [ + 314, + 97 + ], + [ + 263, + 89 + ], + [ + 316, + 122 + ], + [ + 264, + 115 + ] + ], + [ + [ + 287, + 107 + ], + [ + 278, + 122 + ], + [ + 306, + 124 + ], + [ + 267, + 89 + ], + [ + 313, + 110 + ], + [ + 276, + 76 + ] + ] + ] + }, + { + "task_id": "0000798", + "id": "RH20T_cfg4_task_0058_user_0014_scene_0008_cfg_0004_038522062288#226", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000798/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 272, + 89 + ], + [ + 284, + 110 + ], + [ + 291, + 80 + ], + [ + 252, + 98 + ], + [ + 279, + 60 + ], + [ + 240, + 76 + ] + ], + [ + [ + 278, + 64 + ], + [ + 277, + 52 + ], + [ + 254, + 72 + ], + [ + 301, + 56 + ], + [ + 256, + 82 + ], + [ + 301, + 67 + ] + ], + [ + [ + 285, + 71 + ], + [ + 296, + 47 + ], + [ + 304, + 81 + ], + [ + 267, + 62 + ], + [ + 292, + 106 + ], + [ + 256, + 86 + ] + ], + [ + [ + 272, + 76 + ], + [ + 250, + 76 + ], + [ + 261, + 73 + ], + [ + 282, + 78 + ], + [ + 285, + 74 + ], + [ + 303, + 79 + ] + ] + ] + }, + { + "task_id": "0000799", + "id": "RH20T_cfg6_task_0038_user_0014_scene_0010_cfg_0006_104122061602#244", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000799/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 395, + 125 + ], + [ + 398, + 104 + ], + [ + 409, + 140 + ], + [ + 382, + 112 + ], + [ + 406, + 161 + ], + [ + 381, + 132 + ] + ], + [ + [ + 388, + 133 + ], + [ + 406, + 134 + ], + [ + 371, + 136 + ], + [ + 407, + 129 + ], + [ + 353, + 135 + ], + [ + 388, + 128 + ] + ], + [ + [ + 421, + 117 + ], + [ + 401, + 129 + ], + [ + 403, + 99 + ], + [ + 439, + 136 + ], + [ + 422, + 87 + ], + [ + 458, + 123 + ] + ], + [ + [ + 375, + 128 + ], + [ + 366, + 105 + ], + [ + 396, + 122 + ], + [ + 353, + 134 + ], + [ + 405, + 145 + ], + [ + 363, + 158 + ] + ] + ] + }, + { + "task_id": "0000802", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_819112071037#234", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000802/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 186 + ], + [ + 303, + 155 + ], + [ + 284, + 234 + ], + [ + 356, + 133 + ], + [ + 298, + 277 + ], + [ + 382, + 166 + ] + ], + [ + [ + 349, + 210 + ], + [ + 304, + 226 + ], + [ + 313, + 151 + ], + [ + 384, + 266 + ], + [ + 365, + 124 + ], + [ + 441, + 252 + ] + ], + [ + [ + 283, + 190 + ], + [ + 283, + 126 + ], + [ + 217, + 195 + ], + [ + 343, + 185 + ], + [ + 220, + 256 + ], + [ + 341, + 242 + ] + ], + [ + [ + 300, + 147 + ], + [ + 265, + 198 + ], + [ + 258, + 113 + ], + [ + 347, + 185 + ], + [ + 292, + 61 + ], + [ + 387, + 128 + ] + ] + ] + }, + { + "task_id": "0000807", + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_104122061850#320", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000807/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 116 + ], + [ + 413, + 119 + ], + [ + 407, + 101 + ], + [ + 378, + 132 + ], + [ + 388, + 98 + ], + [ + 359, + 128 + ] + ], + [ + [ + 393, + 103 + ], + [ + 393, + 85 + ], + [ + 373, + 116 + ], + [ + 412, + 90 + ], + [ + 375, + 131 + ], + [ + 411, + 106 + ] + ], + [ + [ + 392, + 100 + ], + [ + 396, + 76 + ], + [ + 366, + 99 + ], + [ + 419, + 102 + ], + [ + 362, + 124 + ], + [ + 415, + 127 + ] + ], + [ + [ + 370, + 111 + ], + [ + 383, + 119 + ], + [ + 378, + 131 + ], + [ + 362, + 93 + ], + [ + 364, + 123 + ], + [ + 348, + 84 + ] + ] + ] + }, + { + "task_id": "0000808", + "id": "RH20T_cfg7_task_0015_user_0020_scene_0007_cfg_0007_104122064161#215", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000808/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 339, + 125 + ], + [ + 345, + 104 + ], + [ + 326, + 131 + ], + [ + 352, + 119 + ], + [ + 320, + 151 + ], + [ + 346, + 138 + ] + ], + [ + [ + 358, + 115 + ], + [ + 380, + 111 + ], + [ + 365, + 113 + ], + [ + 351, + 118 + ], + [ + 345, + 117 + ], + [ + 330, + 122 + ] + ], + [ + [ + 361, + 117 + ], + [ + 361, + 140 + ], + [ + 363, + 119 + ], + [ + 358, + 115 + ], + [ + 363, + 97 + ], + [ + 358, + 91 + ] + ], + [ + [ + 343, + 131 + ], + [ + 359, + 142 + ], + [ + 343, + 144 + ], + [ + 343, + 119 + ], + [ + 326, + 132 + ], + [ + 327, + 107 + ] + ] + ] + }, + { + "task_id": "0000809", + "id": "RH20T_cfg7_task_0029_user_0020_scene_0002_cfg_0007_104122064161#205", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000809/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 380, + 91 + ], + [ + 377, + 114 + ], + [ + 375, + 88 + ], + [ + 386, + 94 + ], + [ + 378, + 65 + ], + [ + 389, + 69 + ] + ], + [ + [ + 368, + 103 + ], + [ + 385, + 100 + ], + [ + 385, + 104 + ], + [ + 349, + 102 + ], + [ + 370, + 107 + ], + [ + 334, + 105 + ] + ], + [ + [ + 370, + 69 + ], + [ + 370, + 53 + ], + [ + 387, + 83 + ], + [ + 355, + 56 + ], + [ + 386, + 97 + ], + [ + 355, + 71 + ] + ], + [ + [ + 373, + 68 + ], + [ + 374, + 66 + ], + [ + 367, + 45 + ], + [ + 380, + 90 + ], + [ + 366, + 48 + ], + [ + 379, + 91 + ] + ] + ] + }, + { + "task_id": "0000812", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_036422060215#506", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000812/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 261, + 131 + ], + [ + 266, + 117 + ], + [ + 279, + 129 + ], + [ + 243, + 134 + ], + [ + 274, + 142 + ], + [ + 239, + 148 + ] + ], + [ + [ + 257, + 152 + ], + [ + 252, + 149 + ], + [ + 247, + 169 + ], + [ + 267, + 134 + ], + [ + 251, + 171 + ], + [ + 271, + 137 + ] + ], + [ + [ + 237, + 118 + ], + [ + 231, + 103 + ], + [ + 217, + 124 + ], + [ + 257, + 111 + ], + [ + 224, + 140 + ], + [ + 263, + 127 + ] + ], + [ + [ + 251, + 139 + ], + [ + 235, + 129 + ], + [ + 241, + 139 + ], + [ + 260, + 139 + ], + [ + 257, + 150 + ], + [ + 276, + 149 + ] + ] + ] + }, + { + "task_id": "0000813", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_036422060215#735", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000813/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 207, + 123 + ], + [ + 188, + 112 + ], + [ + 217, + 115 + ], + [ + 197, + 131 + ], + [ + 235, + 125 + ], + [ + 217, + 142 + ] + ], + [ + [ + 200, + 136 + ], + [ + 202, + 120 + ], + [ + 221, + 138 + ], + [ + 177, + 134 + ], + [ + 219, + 153 + ], + [ + 176, + 148 + ] + ], + [ + [ + 215, + 152 + ], + [ + 210, + 159 + ], + [ + 229, + 140 + ], + [ + 199, + 163 + ], + [ + 235, + 132 + ], + [ + 204, + 156 + ] + ], + [ + [ + 216, + 125 + ], + [ + 214, + 105 + ], + [ + 194, + 124 + ], + [ + 237, + 125 + ], + [ + 195, + 144 + ], + [ + 239, + 145 + ] + ] + ] + }, + { + "task_id": "0000814", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_036422060215#1106", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000814/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 155, + 172 + ], + [ + 167, + 190 + ], + [ + 165, + 172 + ], + [ + 143, + 171 + ], + [ + 153, + 155 + ], + [ + 130, + 152 + ] + ], + [ + [ + 174, + 175 + ], + [ + 176, + 159 + ], + [ + 198, + 176 + ], + [ + 148, + 175 + ], + [ + 195, + 191 + ], + [ + 147, + 191 + ] + ], + [ + [ + 187, + 179 + ], + [ + 182, + 162 + ], + [ + 207, + 182 + ], + [ + 164, + 176 + ], + [ + 213, + 198 + ], + [ + 169, + 193 + ] + ], + [ + [ + 168, + 201 + ], + [ + 180, + 196 + ], + [ + 170, + 179 + ], + [ + 165, + 224 + ], + [ + 159, + 185 + ], + [ + 154, + 229 + ] + ] + ] + }, + { + "task_id": "0000815", + "id": "RH20T_cfg7_task_0036_user_0020_scene_0001_cfg_0007_104122060811#150", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Shake the test tube'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000815/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 384, + 138 + ], + [ + 397, + 130 + ], + [ + 393, + 150 + ], + [ + 376, + 126 + ], + [ + 380, + 159 + ], + [ + 363, + 134 + ] + ], + [ + [ + 375, + 129 + ], + [ + 387, + 130 + ], + [ + 368, + 139 + ], + [ + 383, + 120 + ], + [ + 356, + 138 + ], + [ + 371, + 119 + ] + ], + [ + [ + 379, + 132 + ], + [ + 377, + 147 + ], + [ + 394, + 133 + ], + [ + 364, + 132 + ], + [ + 395, + 119 + ], + [ + 365, + 118 + ] + ], + [ + [ + 380, + 131 + ], + [ + 367, + 138 + ], + [ + 374, + 128 + ], + [ + 385, + 133 + ], + [ + 388, + 120 + ], + [ + 398, + 126 + ] + ] + ] + }, + { + "task_id": "0000818", + "id": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_035622060973#196", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000818/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 191 + ], + [ + 318, + 209 + ], + [ + 292, + 206 + ], + [ + 321, + 176 + ], + [ + 281, + 188 + ], + [ + 308, + 158 + ] + ], + [ + [ + 314, + 178 + ], + [ + 302, + 162 + ], + [ + 301, + 174 + ], + [ + 326, + 182 + ], + [ + 314, + 192 + ], + [ + 339, + 199 + ] + ], + [ + [ + 289, + 198 + ], + [ + 276, + 216 + ], + [ + 310, + 204 + ], + [ + 269, + 193 + ], + [ + 322, + 186 + ], + [ + 281, + 176 + ] + ], + [ + [ + 305, + 192 + ], + [ + 305, + 179 + ], + [ + 284, + 204 + ], + [ + 326, + 180 + ], + [ + 286, + 214 + ], + [ + 325, + 191 + ] + ] + ] + }, + { + "task_id": "0000819", + "id": "RH20T_cfg5_task_0016_user_0010_scene_0010_cfg_0005_036422060215#225", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000819/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 286, + 184 + ], + [ + 281, + 169 + ], + [ + 273, + 197 + ], + [ + 300, + 169 + ], + [ + 278, + 214 + ], + [ + 306, + 185 + ] + ], + [ + [ + 260, + 170 + ], + [ + 267, + 182 + ], + [ + 242, + 162 + ], + [ + 277, + 178 + ], + [ + 236, + 151 + ], + [ + 269, + 167 + ] + ], + [ + [ + 256, + 165 + ], + [ + 246, + 159 + ], + [ + 263, + 185 + ], + [ + 249, + 146 + ], + [ + 271, + 190 + ], + [ + 258, + 152 + ] + ], + [ + [ + 274, + 169 + ], + [ + 278, + 151 + ], + [ + 295, + 167 + ], + [ + 253, + 172 + ], + [ + 292, + 184 + ], + [ + 250, + 189 + ] + ] + ] + }, + { + "task_id": "0000820", + "id": "RH20T_cfg6_task_0062_user_0014_scene_0003_cfg_0006_037522061512#111", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the toilet paper on its holder'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000820/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 251, + 110 + ], + [ + 262, + 109 + ], + [ + 268, + 107 + ], + [ + 234, + 113 + ], + [ + 256, + 108 + ], + [ + 221, + 114 + ] + ], + [ + [ + 288, + 96 + ], + [ + 304, + 105 + ], + [ + 298, + 77 + ], + [ + 278, + 115 + ], + [ + 281, + 68 + ], + [ + 261, + 106 + ] + ], + [ + [ + 264, + 65 + ], + [ + 282, + 57 + ], + [ + 253, + 50 + ], + [ + 275, + 78 + ], + [ + 235, + 58 + ], + [ + 258, + 85 + ] + ], + [ + [ + 268, + 95 + ], + [ + 269, + 76 + ], + [ + 278, + 88 + ], + [ + 257, + 103 + ], + [ + 276, + 105 + ], + [ + 255, + 122 + ] + ] + ] + }, + { + "task_id": "0000822", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0009_cfg_0006_037522061512#141", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000822/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 290, + 151 + ], + [ + 300, + 131 + ], + [ + 300, + 155 + ], + [ + 280, + 147 + ], + [ + 290, + 175 + ], + [ + 270, + 169 + ] + ], + [ + [ + 295, + 147 + ], + [ + 278, + 146 + ], + [ + 308, + 143 + ], + [ + 280, + 152 + ], + [ + 325, + 144 + ], + [ + 298, + 153 + ] + ], + [ + [ + 323, + 142 + ], + [ + 320, + 144 + ], + [ + 345, + 136 + ], + [ + 301, + 147 + ], + [ + 346, + 135 + ], + [ + 305, + 145 + ] + ], + [ + [ + 295, + 170 + ], + [ + 306, + 175 + ], + [ + 306, + 188 + ], + [ + 285, + 154 + ], + [ + 295, + 184 + ], + [ + 273, + 149 + ] + ] + ] + }, + { + "task_id": "0000823", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0009_cfg_0006_037522061512#351", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000823/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 248, + 118 + ], + [ + 245, + 118 + ], + [ + 226, + 119 + ], + [ + 271, + 117 + ], + [ + 230, + 118 + ], + [ + 273, + 116 + ] + ], + [ + [ + 243, + 125 + ], + [ + 238, + 116 + ], + [ + 252, + 106 + ], + [ + 234, + 144 + ], + [ + 257, + 114 + ], + [ + 240, + 151 + ] + ], + [ + [ + 249, + 108 + ], + [ + 226, + 102 + ], + [ + 256, + 97 + ], + [ + 242, + 117 + ], + [ + 278, + 103 + ], + [ + 262, + 122 + ] + ], + [ + [ + 250, + 134 + ], + [ + 247, + 128 + ], + [ + 241, + 153 + ], + [ + 259, + 115 + ], + [ + 244, + 157 + ], + [ + 261, + 120 + ] + ] + ] + }, + { + "task_id": "0000825", + "id": "RH20T_cfg1_task_0005_user_0004_scene_0007_cfg_0001_819112071037#232", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000825/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 625, + 225 + ], + [ + 640, + 219 + ], + [ + 579, + 142 + ], + [ + 640, + 318 + ], + [ + 564, + 158 + ], + [ + 640, + 306 + ] + ], + [ + [ + 525, + 148 + ], + [ + 481, + 177 + ], + [ + 572, + 74 + ], + [ + 488, + 207 + ], + [ + 605, + 57 + ], + [ + 523, + 181 + ] + ], + [ + [ + 638, + 151 + ], + [ + 640, + 227 + ], + [ + 537, + 193 + ], + [ + 640, + 97 + ], + [ + 528, + 139 + ], + [ + 640, + 45 + ] + ], + [ + [ + 541, + 153 + ], + [ + 581, + 73 + ], + [ + 452, + 142 + ], + [ + 629, + 164 + ], + [ + 430, + 207 + ], + [ + 588, + 226 + ] + ] + ] + }, + { + "task_id": "0000826", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_104122061018#202", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000826/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 302, + 43 + ], + [ + 332, + 50 + ], + [ + 308, + 33 + ], + [ + 296, + 53 + ], + [ + 280, + 27 + ], + [ + 265, + 45 + ] + ], + [ + [ + 282, + 81 + ], + [ + 271, + 52 + ], + [ + 262, + 89 + ], + [ + 303, + 72 + ], + [ + 272, + 117 + ], + [ + 313, + 102 + ] + ], + [ + [ + 248, + 83 + ], + [ + 231, + 59 + ], + [ + 232, + 81 + ], + [ + 262, + 84 + ], + [ + 250, + 107 + ], + [ + 278, + 107 + ] + ], + [ + [ + 272, + 57 + ], + [ + 272, + 58 + ], + [ + 244, + 64 + ], + [ + 300, + 50 + ], + [ + 242, + 64 + ], + [ + 303, + 48 + ] + ] + ] + }, + { + "task_id": "0000827", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_104122061018#284", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000827/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 238, + 91 + ], + [ + 241, + 59 + ], + [ + 213, + 86 + ], + [ + 264, + 96 + ], + [ + 210, + 117 + ], + [ + 261, + 128 + ] + ], + [ + [ + 275, + 49 + ], + [ + 298, + 32 + ], + [ + 283, + 71 + ], + [ + 265, + 24 + ], + [ + 260, + 87 + ], + [ + 239, + 41 + ] + ], + [ + [ + 288, + 60 + ], + [ + 282, + 45 + ], + [ + 260, + 56 + ], + [ + 318, + 65 + ], + [ + 267, + 69 + ], + [ + 322, + 78 + ] + ], + [ + [ + 256, + 53 + ], + [ + 241, + 34 + ], + [ + 243, + 77 + ], + [ + 270, + 27 + ], + [ + 258, + 97 + ], + [ + 286, + 47 + ] + ] + ] + }, + { + "task_id": "0000828", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_104122061018#314", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000828/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 278, + 69 + ], + [ + 266, + 42 + ], + [ + 261, + 82 + ], + [ + 298, + 54 + ], + [ + 272, + 108 + ], + [ + 310, + 83 + ] + ], + [ + [ + 293, + 77 + ], + [ + 277, + 88 + ], + [ + 270, + 57 + ], + [ + 315, + 98 + ], + [ + 285, + 49 + ], + [ + 328, + 87 + ] + ], + [ + [ + 301, + 85 + ], + [ + 285, + 91 + ], + [ + 279, + 102 + ], + [ + 324, + 66 + ], + [ + 293, + 95 + ], + [ + 337, + 61 + ] + ], + [ + [ + 247, + 54 + ], + [ + 259, + 46 + ], + [ + 247, + 24 + ], + [ + 247, + 84 + ], + [ + 236, + 32 + ], + [ + 236, + 89 + ] + ] + ] + }, + { + "task_id": "0000830", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_104122061018#419", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000830/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 265, + 50 + ], + [ + 254, + 25 + ], + [ + 247, + 68 + ], + [ + 285, + 30 + ], + [ + 258, + 93 + ], + [ + 298, + 57 + ] + ], + [ + [ + 280, + 45 + ], + [ + 273, + 15 + ], + [ + 268, + 46 + ], + [ + 294, + 45 + ], + [ + 275, + 74 + ], + [ + 301, + 76 + ] + ], + [ + [ + 255, + 62 + ], + [ + 256, + 34 + ], + [ + 235, + 48 + ], + [ + 275, + 76 + ], + [ + 235, + 73 + ], + [ + 273, + 101 + ] + ], + [ + [ + 238, + 58 + ], + [ + 264, + 72 + ], + [ + 246, + 33 + ], + [ + 229, + 84 + ], + [ + 219, + 18 + ], + [ + 200, + 69 + ] + ] + ] + }, + { + "task_id": "0000831", + "id": "RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_104122060811#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Cover the box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000831/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 394, + 128 + ], + [ + 398, + 124 + ], + [ + 401, + 118 + ], + [ + 388, + 139 + ], + [ + 397, + 122 + ], + [ + 383, + 144 + ] + ], + [ + [ + 396, + 118 + ], + [ + 395, + 104 + ], + [ + 387, + 118 + ], + [ + 404, + 119 + ], + [ + 388, + 131 + ], + [ + 405, + 133 + ] + ], + [ + [ + 390, + 113 + ], + [ + 393, + 115 + ], + [ + 401, + 103 + ], + [ + 380, + 123 + ], + [ + 398, + 101 + ], + [ + 378, + 121 + ] + ], + [ + [ + 392, + 131 + ], + [ + 392, + 125 + ], + [ + 406, + 126 + ], + [ + 379, + 136 + ], + [ + 405, + 132 + ], + [ + 379, + 141 + ] + ] + ] + }, + { + "task_id": "0000834", + "id": "RH20T_cfg4_task_0016_user_0016_scene_0006_cfg_0004_038522062288#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000834/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 373, + 360 + ], + [ + 380, + 360 + ], + [ + 341, + 360 + ], + [ + 402, + 357 + ], + [ + 337, + 360 + ], + [ + 394, + 360 + ] + ], + [ + [ + 389, + 360 + ], + [ + 353, + 360 + ], + [ + 397, + 360 + ], + [ + 381, + 360 + ], + [ + 439, + 360 + ], + [ + 419, + 360 + ] + ], + [ + [ + 403, + 360 + ], + [ + 443, + 360 + ], + [ + 399, + 360 + ], + [ + 408, + 360 + ], + [ + 363, + 360 + ], + [ + 364, + 360 + ] + ], + [ + [ + 374, + 360 + ], + [ + 361, + 360 + ], + [ + 419, + 360 + ], + [ + 333, + 360 + ], + [ + 426, + 338 + ], + [ + 347, + 345 + ] + ] + ] + }, + { + "task_id": "0000835", + "id": "RH20T_cfg1_task_0070_user_0014_scene_0008_cfg_0001_819112071037#130", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000835/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 460, + 201 + ], + [ + 421, + 219 + ], + [ + 383, + 204 + ], + [ + 528, + 198 + ], + [ + 426, + 178 + ], + [ + 592, + 174 + ] + ], + [ + [ + 377, + 256 + ], + [ + 388, + 184 + ], + [ + 299, + 254 + ], + [ + 453, + 258 + ], + [ + 294, + 319 + ], + [ + 439, + 320 + ] + ], + [ + [ + 406, + 268 + ], + [ + 397, + 270 + ], + [ + 447, + 324 + ], + [ + 367, + 216 + ], + [ + 467, + 331 + ], + [ + 374, + 205 + ] + ], + [ + [ + 424, + 217 + ], + [ + 408, + 215 + ], + [ + 425, + 296 + ], + [ + 423, + 139 + ], + [ + 449, + 315 + ], + [ + 446, + 127 + ] + ] + ] + }, + { + "task_id": "0000836", + "id": "RH20T_cfg7_task_0038_user_0020_scene_0006_cfg_0007_104122060811#178", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000836/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 362, + 107 + ], + [ + 358, + 109 + ], + [ + 376, + 113 + ], + [ + 349, + 101 + ], + [ + 379, + 111 + ], + [ + 352, + 99 + ] + ], + [ + [ + 362, + 131 + ], + [ + 354, + 119 + ], + [ + 372, + 129 + ], + [ + 351, + 133 + ], + [ + 380, + 140 + ], + [ + 359, + 144 + ] + ], + [ + [ + 349, + 123 + ], + [ + 348, + 130 + ], + [ + 345, + 109 + ], + [ + 354, + 136 + ], + [ + 346, + 103 + ], + [ + 355, + 129 + ] + ], + [ + [ + 356, + 125 + ], + [ + 360, + 117 + ], + [ + 369, + 123 + ], + [ + 342, + 128 + ], + [ + 365, + 131 + ], + [ + 339, + 135 + ] + ] + ] + }, + { + "task_id": "0000838", + "id": "RH20T_cfg1_task_0064_user_0001_scene_0008_cfg_0001_750612070851#165", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000838/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 160, + 196 + ], + [ + 202, + 207 + ], + [ + 156, + 174 + ], + [ + 164, + 215 + ], + [ + 110, + 161 + ], + [ + 125, + 205 + ] + ], + [ + [ + 202, + 176 + ], + [ + 183, + 169 + ], + [ + 234, + 183 + ], + [ + 166, + 168 + ], + [ + 257, + 190 + ], + [ + 187, + 176 + ] + ], + [ + [ + 209, + 218 + ], + [ + 202, + 195 + ], + [ + 190, + 251 + ], + [ + 226, + 187 + ], + [ + 198, + 270 + ], + [ + 231, + 208 + ] + ], + [ + [ + 209, + 214 + ], + [ + 194, + 177 + ], + [ + 204, + 220 + ], + [ + 216, + 206 + ], + [ + 218, + 255 + ], + [ + 232, + 247 + ] + ] + ] + }, + { + "task_id": "0000842", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122061018#700", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000842/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 321, + 82 + ], + [ + 333, + 69 + ], + [ + 319, + 104 + ], + [ + 323, + 61 + ], + [ + 309, + 114 + ], + [ + 313, + 74 + ] + ], + [ + [ + 308, + 70 + ], + [ + 298, + 68 + ], + [ + 319, + 46 + ], + [ + 298, + 94 + ], + [ + 331, + 46 + ], + [ + 308, + 97 + ] + ], + [ + [ + 318, + 103 + ], + [ + 306, + 120 + ], + [ + 299, + 85 + ], + [ + 336, + 120 + ], + [ + 311, + 68 + ], + [ + 347, + 103 + ] + ], + [ + [ + 303, + 76 + ], + [ + 293, + 62 + ], + [ + 302, + 98 + ], + [ + 303, + 53 + ], + [ + 312, + 113 + ], + [ + 313, + 68 + ] + ] + ] + }, + { + "task_id": "0000843", + "id": "RH20T_cfg1_task_0015_user_0001_scene_0007_cfg_0001_750612070853#129", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000843/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 189, + 360 + ], + [ + 213, + 360 + ], + [ + 228, + 360 + ], + [ + 145, + 337 + ], + [ + 206, + 360 + ], + [ + 128, + 325 + ] + ], + [ + [ + 268, + 309 + ], + [ + 236, + 289 + ], + [ + 236, + 282 + ], + [ + 295, + 332 + ], + [ + 275, + 305 + ], + [ + 333, + 356 + ] + ], + [ + [ + 303, + 360 + ], + [ + 269, + 346 + ], + [ + 272, + 360 + ], + [ + 331, + 323 + ], + [ + 311, + 360 + ], + [ + 371, + 345 + ] + ], + [ + [ + 243, + 353 + ], + [ + 233, + 314 + ], + [ + 198, + 360 + ], + [ + 283, + 319 + ], + [ + 210, + 360 + ], + [ + 289, + 353 + ] + ] + ] + }, + { + "task_id": "0000844", + "id": "RH20T_cfg1_task_0065_user_0015_scene_0001_cfg_0001_035622060973#303", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000844/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 310, + 289 + ], + [ + 311, + 282 + ], + [ + 287, + 288 + ], + [ + 333, + 290 + ], + [ + 287, + 295 + ], + [ + 331, + 297 + ] + ], + [ + [ + 333, + 275 + ], + [ + 311, + 273 + ], + [ + 329, + 264 + ], + [ + 337, + 285 + ], + [ + 353, + 266 + ], + [ + 359, + 287 + ] + ], + [ + [ + 292, + 281 + ], + [ + 287, + 261 + ], + [ + 275, + 290 + ], + [ + 309, + 271 + ], + [ + 280, + 310 + ], + [ + 315, + 292 + ] + ], + [ + [ + 334, + 277 + ], + [ + 334, + 261 + ], + [ + 327, + 294 + ], + [ + 340, + 260 + ], + [ + 326, + 311 + ], + [ + 339, + 278 + ] + ] + ] + }, + { + "task_id": "0000845", + "id": "RH20T_cfg1_task_0065_user_0015_scene_0001_cfg_0001_035622060973#492", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000845/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 282, + 183 + ], + [ + 298, + 173 + ], + [ + 269, + 171 + ], + [ + 296, + 195 + ], + [ + 253, + 181 + ], + [ + 280, + 207 + ] + ], + [ + [ + 278, + 197 + ], + [ + 277, + 187 + ], + [ + 257, + 202 + ], + [ + 298, + 193 + ], + [ + 259, + 212 + ], + [ + 298, + 203 + ] + ], + [ + [ + 285, + 183 + ], + [ + 298, + 197 + ], + [ + 298, + 179 + ], + [ + 272, + 187 + ], + [ + 285, + 166 + ], + [ + 259, + 172 + ] + ], + [ + [ + 278, + 181 + ], + [ + 290, + 183 + ], + [ + 287, + 199 + ], + [ + 270, + 163 + ], + [ + 275, + 197 + ], + [ + 258, + 160 + ] + ] + ] + }, + { + "task_id": "0000846", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0002_cfg_0005_104122063678#169", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000846/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 302, + 157 + ], + [ + 325, + 165 + ], + [ + 295, + 169 + ], + [ + 310, + 143 + ], + [ + 272, + 160 + ], + [ + 286, + 133 + ] + ], + [ + [ + 295, + 144 + ], + [ + 316, + 146 + ], + [ + 307, + 138 + ], + [ + 281, + 152 + ], + [ + 287, + 136 + ], + [ + 260, + 150 + ] + ], + [ + [ + 323, + 134 + ], + [ + 329, + 115 + ], + [ + 314, + 148 + ], + [ + 332, + 120 + ], + [ + 308, + 165 + ], + [ + 325, + 137 + ] + ], + [ + [ + 305, + 135 + ], + [ + 311, + 157 + ], + [ + 284, + 134 + ], + [ + 326, + 136 + ], + [ + 279, + 112 + ], + [ + 320, + 116 + ] + ] + ] + }, + { + "task_id": "0000851", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_819112070044#88", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000851/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 345, + 209 + ], + [ + 332, + 196 + ], + [ + 340, + 195 + ], + [ + 350, + 223 + ], + [ + 352, + 207 + ], + [ + 362, + 235 + ] + ], + [ + [ + 339, + 218 + ], + [ + 321, + 209 + ], + [ + 332, + 228 + ], + [ + 347, + 208 + ], + [ + 350, + 237 + ], + [ + 365, + 217 + ] + ], + [ + [ + 345, + 217 + ], + [ + 347, + 201 + ], + [ + 355, + 205 + ], + [ + 335, + 230 + ], + [ + 352, + 219 + ], + [ + 333, + 245 + ] + ], + [ + [ + 333, + 226 + ], + [ + 343, + 237 + ], + [ + 341, + 208 + ], + [ + 324, + 244 + ], + [ + 331, + 196 + ], + [ + 313, + 232 + ] + ] + ] + }, + { + "task_id": "0000852", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_819112070044#188", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000852/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 311, + 195 + ], + [ + 318, + 211 + ], + [ + 321, + 181 + ], + [ + 302, + 209 + ], + [ + 314, + 166 + ], + [ + 295, + 194 + ] + ], + [ + [ + 323, + 211 + ], + [ + 325, + 195 + ], + [ + 340, + 206 + ], + [ + 304, + 217 + ], + [ + 338, + 220 + ], + [ + 303, + 232 + ] + ], + [ + [ + 311, + 231 + ], + [ + 329, + 234 + ], + [ + 300, + 223 + ], + [ + 321, + 238 + ], + [ + 282, + 221 + ], + [ + 303, + 236 + ] + ], + [ + [ + 317, + 226 + ], + [ + 303, + 230 + ], + [ + 316, + 205 + ], + [ + 317, + 248 + ], + [ + 329, + 203 + ], + [ + 330, + 244 + ] + ] + ] + }, + { + "task_id": "0000854", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_819112070044#467", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000854/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 202 + ], + [ + 277, + 195 + ], + [ + 301, + 190 + ], + [ + 268, + 214 + ], + [ + 308, + 197 + ], + [ + 276, + 221 + ] + ], + [ + [ + 307, + 205 + ], + [ + 307, + 191 + ], + [ + 295, + 192 + ], + [ + 320, + 218 + ], + [ + 296, + 206 + ], + [ + 320, + 230 + ] + ], + [ + [ + 301, + 203 + ], + [ + 308, + 209 + ], + [ + 289, + 190 + ], + [ + 314, + 216 + ], + [ + 281, + 183 + ], + [ + 307, + 210 + ] + ], + [ + [ + 320, + 219 + ], + [ + 300, + 211 + ], + [ + 321, + 216 + ], + [ + 318, + 223 + ], + [ + 342, + 224 + ], + [ + 337, + 231 + ] + ] + ] + }, + { + "task_id": "0000856", + "id": "RH20T_cfg6_task_0046_user_0020_scene_0010_cfg_0006_104122061018#311", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000856/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 298, + 125 + ], + [ + 307, + 102 + ], + [ + 329, + 129 + ], + [ + 269, + 122 + ], + [ + 320, + 154 + ], + [ + 259, + 146 + ] + ], + [ + [ + 271, + 84 + ], + [ + 297, + 70 + ], + [ + 291, + 100 + ], + [ + 249, + 67 + ], + [ + 267, + 112 + ], + [ + 224, + 82 + ] + ], + [ + [ + 265, + 109 + ], + [ + 255, + 97 + ], + [ + 272, + 136 + ], + [ + 256, + 80 + ], + [ + 283, + 151 + ], + [ + 267, + 93 + ] + ], + [ + [ + 285, + 102 + ], + [ + 274, + 108 + ], + [ + 296, + 73 + ], + [ + 275, + 133 + ], + [ + 309, + 65 + ], + [ + 286, + 129 + ] + ] + ] + }, + { + "task_id": "0000857", + "id": "RH20T_cfg6_task_0130_user_0014_scene_0004_cfg_0006_104122061330#195", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000857/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 362, + 173 + ], + [ + 358, + 155 + ], + [ + 385, + 162 + ], + [ + 339, + 184 + ], + [ + 390, + 181 + ], + [ + 343, + 203 + ] + ], + [ + [ + 360, + 147 + ], + [ + 346, + 144 + ], + [ + 354, + 173 + ], + [ + 365, + 120 + ], + [ + 367, + 174 + ], + [ + 377, + 124 + ] + ], + [ + [ + 355, + 168 + ], + [ + 338, + 187 + ], + [ + 370, + 184 + ], + [ + 341, + 154 + ], + [ + 388, + 164 + ], + [ + 357, + 135 + ] + ], + [ + [ + 372, + 160 + ], + [ + 366, + 135 + ], + [ + 347, + 166 + ], + [ + 397, + 155 + ], + [ + 354, + 192 + ], + [ + 404, + 180 + ] + ] + ] + }, + { + "task_id": "0000860", + "id": "RH20T_cfg1_task_0042_user_0016_scene_0007_cfg_0001_038522062547#298", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000860/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 461, + 17 + ], + [ + 454, + 20 + ], + [ + 437, + 6 + ], + [ + 485, + 29 + ], + [ + 443, + 2 + ], + [ + 493, + 26 + ] + ], + [ + [ + 448, + 6 + ], + [ + 455, + 0 + ], + [ + 467, + 22 + ], + [ + 431, + 0 + ], + [ + 460, + 41 + ], + [ + 426, + 11 + ] + ], + [ + [ + 439, + 9 + ], + [ + 416, + 0 + ], + [ + 448, + 0 + ], + [ + 430, + 36 + ], + [ + 470, + 0 + ], + [ + 452, + 44 + ] + ], + [ + [ + 459, + 14 + ], + [ + 461, + 38 + ], + [ + 483, + 6 + ], + [ + 434, + 22 + ], + [ + 482, + 0 + ], + [ + 432, + 0 + ] + ] + ] + }, + { + "task_id": "0000861", + "id": "RH20T_cfg1_task_0042_user_0016_scene_0007_cfg_0001_038522062547#337", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000861/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 440, + 29 + ], + [ + 414, + 30 + ], + [ + 449, + 42 + ], + [ + 431, + 18 + ], + [ + 474, + 41 + ], + [ + 455, + 17 + ] + ], + [ + [ + 454, + 5 + ], + [ + 461, + 0 + ], + [ + 473, + 21 + ], + [ + 437, + 0 + ], + [ + 466, + 41 + ], + [ + 432, + 11 + ] + ], + [ + [ + 461, + 21 + ], + [ + 440, + 32 + ], + [ + 477, + 39 + ], + [ + 446, + 4 + ], + [ + 498, + 28 + ], + [ + 466, + 0 + ] + ], + [ + [ + 422, + 2 + ], + [ + 428, + 27 + ], + [ + 432, + 0 + ], + [ + 413, + 13 + ], + [ + 425, + 0 + ], + [ + 408, + 0 + ] + ] + ] + }, + { + "task_id": "0000863", + "id": "RH20T_cfg5_task_0010_user_0010_scene_0002_cfg_0005_104122061850#306", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000863/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 246, + 159 + ], + [ + 225, + 173 + ], + [ + 252, + 154 + ], + [ + 241, + 164 + ], + [ + 274, + 141 + ], + [ + 261, + 152 + ] + ], + [ + [ + 268, + 184 + ], + [ + 258, + 171 + ], + [ + 260, + 207 + ], + [ + 276, + 163 + ], + [ + 271, + 219 + ], + [ + 285, + 176 + ] + ], + [ + [ + 263, + 176 + ], + [ + 284, + 174 + ], + [ + 265, + 152 + ], + [ + 262, + 200 + ], + [ + 245, + 154 + ], + [ + 242, + 200 + ] + ], + [ + [ + 264, + 151 + ], + [ + 237, + 152 + ], + [ + 261, + 134 + ], + [ + 267, + 170 + ], + [ + 286, + 134 + ], + [ + 294, + 169 + ] + ] + ] + }, + { + "task_id": "0000864", + "id": "RH20T_cfg5_task_0025_user_0010_scene_0006_cfg_0005_036422060215#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold a piece of paper'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000864/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 270, + 164 + ], + [ + 265, + 147 + ], + [ + 288, + 155 + ], + [ + 250, + 174 + ], + [ + 291, + 172 + ], + [ + 255, + 190 + ] + ], + [ + [ + 245, + 153 + ], + [ + 237, + 170 + ], + [ + 258, + 166 + ], + [ + 231, + 138 + ], + [ + 265, + 151 + ], + [ + 239, + 123 + ] + ], + [ + [ + 278, + 167 + ], + [ + 290, + 174 + ], + [ + 293, + 175 + ], + [ + 264, + 159 + ], + [ + 280, + 168 + ], + [ + 250, + 152 + ] + ], + [ + [ + 280, + 148 + ], + [ + 277, + 131 + ], + [ + 301, + 142 + ], + [ + 260, + 153 + ], + [ + 305, + 160 + ], + [ + 262, + 171 + ] + ] + ] + }, + { + "task_id": "0000865", + "id": "RH20T_cfg4_task_0011_user_0016_scene_0001_cfg_0004_038522062288#111", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000865/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 293, + 123 + ], + [ + 261, + 117 + ], + [ + 307, + 123 + ], + [ + 275, + 123 + ], + [ + 339, + 128 + ], + [ + 311, + 129 + ] + ], + [ + [ + 267, + 150 + ], + [ + 289, + 168 + ], + [ + 259, + 132 + ], + [ + 274, + 165 + ], + [ + 236, + 114 + ], + [ + 254, + 148 + ] + ], + [ + [ + 296, + 89 + ], + [ + 272, + 79 + ], + [ + 308, + 55 + ], + [ + 284, + 122 + ], + [ + 335, + 64 + ], + [ + 309, + 134 + ] + ], + [ + [ + 286, + 126 + ], + [ + 284, + 101 + ], + [ + 263, + 146 + ], + [ + 307, + 107 + ], + [ + 265, + 168 + ], + [ + 307, + 130 + ] + ] + ] + }, + { + "task_id": "0000870", + "id": "RH20T_cfg1_task_0045_user_0015_scene_0002_cfg_0001_750612070853#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the brush on the pen rack'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000870/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 503, + 327 + ], + [ + 510, + 269 + ], + [ + 444, + 313 + ], + [ + 564, + 341 + ], + [ + 435, + 360 + ], + [ + 558, + 360 + ] + ], + [ + [ + 462, + 348 + ], + [ + 500, + 360 + ], + [ + 496, + 360 + ], + [ + 426, + 323 + ], + [ + 464, + 339 + ], + [ + 399, + 292 + ] + ], + [ + [ + 441, + 296 + ], + [ + 457, + 353 + ], + [ + 396, + 299 + ], + [ + 482, + 294 + ], + [ + 384, + 245 + ], + [ + 467, + 246 + ] + ], + [ + [ + 443, + 323 + ], + [ + 445, + 288 + ], + [ + 386, + 327 + ], + [ + 501, + 319 + ], + [ + 388, + 358 + ], + [ + 495, + 351 + ] + ] + ] + }, + { + "task_id": "0000871", + "id": "RH20T_cfg5_task_0080_user_0007_scene_0010_cfg_0005_104122063678#275", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Flip over and spread out the paper that is laid flat on the table'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000871/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 167 + ], + [ + 309, + 151 + ], + [ + 284, + 169 + ], + [ + 327, + 164 + ], + [ + 282, + 183 + ], + [ + 323, + 178 + ] + ], + [ + [ + 293, + 165 + ], + [ + 298, + 183 + ], + [ + 314, + 155 + ], + [ + 273, + 174 + ], + [ + 308, + 138 + ], + [ + 269, + 156 + ] + ], + [ + [ + 311, + 181 + ], + [ + 321, + 198 + ], + [ + 290, + 187 + ], + [ + 331, + 174 + ], + [ + 281, + 170 + ], + [ + 322, + 159 + ] + ], + [ + [ + 313, + 170 + ], + [ + 326, + 159 + ], + [ + 330, + 178 + ], + [ + 295, + 161 + ], + [ + 318, + 188 + ], + [ + 284, + 171 + ] + ] + ] + }, + { + "task_id": "0000873", + "id": "RH20T_cfg7_task_0042_user_0020_scene_0003_cfg_0007_104122061018#161", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000873/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 264, + 86 + ], + [ + 288, + 89 + ], + [ + 277, + 63 + ], + [ + 252, + 108 + ], + [ + 250, + 58 + ], + [ + 226, + 105 + ] + ], + [ + [ + 288, + 91 + ], + [ + 314, + 80 + ], + [ + 281, + 85 + ], + [ + 296, + 97 + ], + [ + 256, + 95 + ], + [ + 268, + 109 + ] + ], + [ + [ + 280, + 115 + ], + [ + 301, + 130 + ], + [ + 289, + 125 + ], + [ + 269, + 103 + ], + [ + 270, + 111 + ], + [ + 249, + 88 + ] + ], + [ + [ + 267, + 104 + ], + [ + 268, + 78 + ], + [ + 255, + 89 + ], + [ + 281, + 120 + ], + [ + 255, + 113 + ], + [ + 279, + 145 + ] + ] + ] + }, + { + "task_id": "0000874", + "id": "RH20T_cfg2_task_0041_user_0007_scene_0009_cfg_0002_037522062165#466", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the tabletop with a sponge'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000874/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 118, + 185 + ], + [ + 111, + 172 + ], + [ + 98, + 173 + ], + [ + 138, + 197 + ], + [ + 106, + 186 + ], + [ + 143, + 209 + ] + ], + [ + [ + 103, + 183 + ], + [ + 84, + 199 + ], + [ + 90, + 184 + ], + [ + 117, + 182 + ], + [ + 108, + 169 + ], + [ + 136, + 167 + ] + ], + [ + [ + 135, + 182 + ], + [ + 151, + 164 + ], + [ + 155, + 197 + ], + [ + 114, + 166 + ], + [ + 140, + 215 + ], + [ + 97, + 185 + ] + ], + [ + [ + 106, + 165 + ], + [ + 123, + 162 + ], + [ + 96, + 185 + ], + [ + 116, + 144 + ], + [ + 77, + 189 + ], + [ + 97, + 147 + ] + ] + ] + }, + { + "task_id": "0000875", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0003_cfg_0004_104122062823#169", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000875/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 310, + 67 + ], + [ + 306, + 35 + ], + [ + 329, + 57 + ], + [ + 289, + 78 + ], + [ + 332, + 86 + ], + [ + 293, + 109 + ] + ], + [ + [ + 303, + 106 + ], + [ + 275, + 117 + ], + [ + 303, + 101 + ], + [ + 302, + 109 + ], + [ + 333, + 89 + ], + [ + 329, + 99 + ] + ], + [ + [ + 318, + 94 + ], + [ + 318, + 65 + ], + [ + 346, + 102 + ], + [ + 292, + 86 + ], + [ + 345, + 129 + ], + [ + 293, + 111 + ] + ], + [ + [ + 332, + 75 + ], + [ + 337, + 105 + ], + [ + 317, + 74 + ], + [ + 350, + 76 + ], + [ + 312, + 44 + ], + [ + 345, + 42 + ] + ] + ] + }, + { + "task_id": "0000876", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0003_cfg_0004_104122062823#396", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000876/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 76 + ], + [ + 303, + 51 + ], + [ + 328, + 87 + ], + [ + 283, + 66 + ], + [ + 328, + 111 + ], + [ + 285, + 90 + ] + ], + [ + [ + 299, + 55 + ], + [ + 328, + 66 + ], + [ + 288, + 67 + ], + [ + 309, + 45 + ], + [ + 258, + 56 + ], + [ + 282, + 35 + ] + ], + [ + [ + 312, + 79 + ], + [ + 295, + 96 + ], + [ + 301, + 56 + ], + [ + 322, + 99 + ], + [ + 318, + 39 + ], + [ + 338, + 83 + ] + ], + [ + [ + 290, + 102 + ], + [ + 290, + 131 + ], + [ + 293, + 109 + ], + [ + 287, + 94 + ], + [ + 293, + 82 + ], + [ + 287, + 64 + ] + ] + ] + }, + { + "task_id": "0000877", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0003_cfg_0004_104122062823#529", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000877/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 278, + 70 + ], + [ + 274, + 46 + ], + [ + 298, + 83 + ], + [ + 260, + 59 + ], + [ + 301, + 104 + ], + [ + 264, + 80 + ] + ], + [ + [ + 292, + 66 + ], + [ + 278, + 92 + ], + [ + 279, + 58 + ], + [ + 306, + 75 + ], + [ + 292, + 33 + ], + [ + 321, + 48 + ] + ], + [ + [ + 286, + 85 + ], + [ + 307, + 69 + ], + [ + 299, + 109 + ], + [ + 271, + 60 + ], + [ + 279, + 126 + ], + [ + 249, + 78 + ] + ], + [ + [ + 260, + 64 + ], + [ + 235, + 78 + ], + [ + 274, + 92 + ], + [ + 245, + 36 + ], + [ + 300, + 78 + ], + [ + 271, + 21 + ] + ] + ] + }, + { + "task_id": "0000878", + "id": "RH20T_cfg1_task_0015_user_0001_scene_0007_cfg_0001_035622060973#260", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000878/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 206, + 193 + ], + [ + 234, + 197 + ], + [ + 208, + 202 + ], + [ + 203, + 182 + ], + [ + 181, + 198 + ], + [ + 173, + 177 + ] + ], + [ + [ + 204, + 199 + ], + [ + 207, + 190 + ], + [ + 227, + 216 + ], + [ + 181, + 181 + ], + [ + 223, + 223 + ], + [ + 179, + 190 + ] + ], + [ + [ + 163, + 213 + ], + [ + 134, + 221 + ], + [ + 154, + 188 + ], + [ + 172, + 238 + ], + [ + 181, + 180 + ], + [ + 199, + 229 + ] + ], + [ + [ + 187, + 214 + ], + [ + 189, + 203 + ], + [ + 157, + 215 + ], + [ + 215, + 212 + ], + [ + 157, + 225 + ], + [ + 212, + 222 + ] + ] + ] + }, + { + "task_id": "0000879", + "id": "RH20T_cfg7_task_0046_user_0014_scene_0001_cfg_0007_104122060811#102", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000879/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 354, + 134 + ], + [ + 349, + 126 + ], + [ + 357, + 145 + ], + [ + 352, + 122 + ], + [ + 362, + 154 + ], + [ + 357, + 131 + ] + ], + [ + [ + 349, + 112 + ], + [ + 354, + 123 + ], + [ + 358, + 117 + ], + [ + 340, + 106 + ], + [ + 353, + 106 + ], + [ + 335, + 94 + ] + ], + [ + [ + 353, + 131 + ], + [ + 353, + 123 + ], + [ + 367, + 127 + ], + [ + 338, + 136 + ], + [ + 367, + 135 + ], + [ + 338, + 144 + ] + ], + [ + [ + 343, + 136 + ], + [ + 332, + 143 + ], + [ + 349, + 131 + ], + [ + 337, + 141 + ], + [ + 361, + 124 + ], + [ + 349, + 133 + ] + ] + ] + }, + { + "task_id": "0000881", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0010_cfg_0001_819112071037#227", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000881/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 387, + 83 + ], + [ + 393, + 18 + ], + [ + 329, + 84 + ], + [ + 442, + 82 + ], + [ + 326, + 145 + ], + [ + 435, + 139 + ] + ], + [ + [ + 384, + 53 + ], + [ + 364, + 28 + ], + [ + 447, + 50 + ], + [ + 316, + 57 + ], + [ + 478, + 79 + ], + [ + 333, + 90 + ] + ], + [ + [ + 392, + 25 + ], + [ + 422, + 0 + ], + [ + 338, + 2 + ], + [ + 451, + 49 + ], + [ + 302, + 34 + ], + [ + 420, + 87 + ] + ], + [ + [ + 400, + 22 + ], + [ + 378, + 19 + ], + [ + 382, + 80 + ], + [ + 420, + 0 + ], + [ + 408, + 92 + ], + [ + 452, + 0 + ] + ] + ] + }, + { + "task_id": "0000885", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061018#126", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000885/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 303, + 168 + ], + [ + 329, + 146 + ], + [ + 302, + 159 + ], + [ + 304, + 175 + ], + [ + 273, + 184 + ], + [ + 279, + 196 + ] + ], + [ + [ + 336, + 135 + ], + [ + 357, + 118 + ], + [ + 311, + 115 + ], + [ + 362, + 155 + ], + [ + 288, + 132 + ], + [ + 339, + 174 + ] + ], + [ + [ + 312, + 145 + ], + [ + 310, + 126 + ], + [ + 281, + 139 + ], + [ + 343, + 152 + ], + [ + 284, + 155 + ], + [ + 343, + 168 + ] + ], + [ + [ + 321, + 154 + ], + [ + 327, + 131 + ], + [ + 308, + 130 + ], + [ + 335, + 179 + ], + [ + 303, + 152 + ], + [ + 328, + 199 + ] + ] + ] + }, + { + "task_id": "0000891", + "id": "RH20T_cfg3_task_0023_user_0010_scene_0004_cfg_0003_104122062823#316", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000891/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 138 + ], + [ + 315, + 141 + ], + [ + 271, + 158 + ], + [ + 339, + 119 + ], + [ + 266, + 154 + ], + [ + 329, + 118 + ] + ], + [ + [ + 320, + 194 + ], + [ + 309, + 206 + ], + [ + 305, + 158 + ], + [ + 334, + 229 + ], + [ + 316, + 151 + ], + [ + 342, + 217 + ] + ], + [ + [ + 305, + 186 + ], + [ + 328, + 211 + ], + [ + 279, + 183 + ], + [ + 328, + 190 + ], + [ + 257, + 160 + ], + [ + 307, + 169 + ] + ], + [ + [ + 308, + 164 + ], + [ + 311, + 133 + ], + [ + 337, + 152 + ], + [ + 277, + 176 + ], + [ + 334, + 179 + ], + [ + 276, + 204 + ] + ] + ] + }, + { + "task_id": "0000896", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122061018#130", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000896/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 157 + ], + [ + 286, + 168 + ], + [ + 322, + 186 + ], + [ + 309, + 125 + ], + [ + 349, + 175 + ], + [ + 338, + 115 + ] + ], + [ + [ + 305, + 151 + ], + [ + 309, + 128 + ], + [ + 291, + 128 + ], + [ + 319, + 177 + ], + [ + 288, + 149 + ], + [ + 314, + 197 + ] + ], + [ + [ + 295, + 123 + ], + [ + 317, + 140 + ], + [ + 288, + 100 + ], + [ + 300, + 143 + ], + [ + 266, + 84 + ], + [ + 281, + 127 + ] + ], + [ + [ + 309, + 124 + ], + [ + 286, + 139 + ], + [ + 318, + 155 + ], + [ + 300, + 94 + ], + [ + 344, + 139 + ], + [ + 324, + 76 + ] + ] + ] + }, + { + "task_id": "0000897", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122061018#295", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000897/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 250, + 176 + ], + [ + 245, + 195 + ], + [ + 245, + 200 + ], + [ + 256, + 149 + ], + [ + 250, + 182 + ], + [ + 261, + 133 + ] + ], + [ + [ + 276, + 188 + ], + [ + 246, + 197 + ], + [ + 263, + 181 + ], + [ + 290, + 195 + ], + [ + 289, + 173 + ], + [ + 319, + 186 + ] + ], + [ + [ + 285, + 177 + ], + [ + 288, + 154 + ], + [ + 272, + 155 + ], + [ + 300, + 201 + ], + [ + 270, + 175 + ], + [ + 296, + 220 + ] + ], + [ + [ + 306, + 156 + ], + [ + 301, + 162 + ], + [ + 304, + 187 + ], + [ + 309, + 126 + ], + [ + 310, + 181 + ], + [ + 316, + 116 + ] + ] + ] + }, + { + "task_id": "0000898", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122061018#428", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000898/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 303, + 148 + ], + [ + 314, + 167 + ], + [ + 313, + 164 + ], + [ + 292, + 129 + ], + [ + 303, + 147 + ], + [ + 282, + 111 + ] + ], + [ + [ + 304, + 150 + ], + [ + 306, + 121 + ], + [ + 289, + 138 + ], + [ + 322, + 163 + ], + [ + 288, + 164 + ], + [ + 320, + 191 + ] + ], + [ + [ + 299, + 171 + ], + [ + 292, + 195 + ], + [ + 284, + 186 + ], + [ + 313, + 158 + ], + [ + 291, + 160 + ], + [ + 321, + 133 + ] + ], + [ + [ + 307, + 137 + ], + [ + 314, + 160 + ], + [ + 308, + 154 + ], + [ + 306, + 117 + ], + [ + 302, + 133 + ], + [ + 300, + 95 + ] + ] + ] + }, + { + "task_id": "0000900", + "id": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_104122063550#41", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000900/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 258, + 245 + ], + [ + 245, + 252 + ], + [ + 281, + 232 + ], + [ + 234, + 259 + ], + [ + 296, + 224 + ], + [ + 248, + 252 + ] + ], + [ + [ + 245, + 278 + ], + [ + 223, + 299 + ], + [ + 258, + 293 + ], + [ + 234, + 265 + ], + [ + 281, + 271 + ], + [ + 255, + 245 + ] + ], + [ + [ + 227, + 239 + ], + [ + 199, + 244 + ], + [ + 223, + 247 + ], + [ + 231, + 231 + ], + [ + 253, + 242 + ], + [ + 257, + 227 + ] + ], + [ + [ + 253, + 248 + ], + [ + 248, + 240 + ], + [ + 270, + 270 + ], + [ + 236, + 226 + ], + [ + 273, + 276 + ], + [ + 241, + 233 + ] + ] + ] + }, + { + "task_id": "0000901", + "id": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_104122063550#121", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000901/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 288, + 267 + ], + [ + 305, + 263 + ], + [ + 269, + 246 + ], + [ + 308, + 289 + ], + [ + 250, + 248 + ], + [ + 290, + 295 + ] + ], + [ + [ + 300, + 244 + ], + [ + 308, + 221 + ], + [ + 275, + 245 + ], + [ + 324, + 242 + ], + [ + 268, + 268 + ], + [ + 317, + 263 + ] + ], + [ + [ + 282, + 246 + ], + [ + 280, + 239 + ], + [ + 297, + 271 + ], + [ + 268, + 222 + ], + [ + 298, + 276 + ], + [ + 270, + 230 + ] + ], + [ + [ + 273, + 234 + ], + [ + 296, + 244 + ], + [ + 259, + 258 + ], + [ + 287, + 211 + ], + [ + 238, + 248 + ], + [ + 266, + 203 + ] + ] + ] + }, + { + "task_id": "0000903", + "id": "RH20T_cfg7_task_0063_user_0020_scene_0008_cfg_0007_104122061018#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a shovel to scoop up an object'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000903/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 349, + 231 + ], + [ + 357, + 217 + ], + [ + 344, + 257 + ], + [ + 354, + 204 + ], + [ + 335, + 275 + ], + [ + 345, + 219 + ] + ], + [ + [ + 364, + 216 + ], + [ + 355, + 185 + ], + [ + 346, + 223 + ], + [ + 384, + 209 + ], + [ + 354, + 253 + ], + [ + 394, + 243 + ] + ], + [ + [ + 332, + 240 + ], + [ + 354, + 228 + ], + [ + 310, + 233 + ], + [ + 356, + 247 + ], + [ + 287, + 245 + ], + [ + 332, + 261 + ] + ], + [ + [ + 399, + 236 + ], + [ + 371, + 248 + ], + [ + 385, + 208 + ], + [ + 414, + 264 + ], + [ + 414, + 195 + ], + [ + 444, + 252 + ] + ] + ] + }, + { + "task_id": "0000909", + "id": "RH20T_cfg2_task_0004_user_0006_scene_0003_cfg_0002_104122061850#305", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000909/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 228 + ], + [ + 293, + 213 + ], + [ + 322, + 214 + ], + [ + 276, + 241 + ], + [ + 327, + 228 + ], + [ + 283, + 255 + ] + ], + [ + [ + 283, + 244 + ], + [ + 274, + 230 + ], + [ + 258, + 249 + ], + [ + 309, + 238 + ], + [ + 267, + 261 + ], + [ + 316, + 251 + ] + ], + [ + [ + 304, + 251 + ], + [ + 304, + 229 + ], + [ + 328, + 243 + ], + [ + 281, + 260 + ], + [ + 328, + 267 + ], + [ + 280, + 284 + ] + ], + [ + [ + 268, + 240 + ], + [ + 253, + 222 + ], + [ + 282, + 237 + ], + [ + 252, + 242 + ], + [ + 297, + 255 + ], + [ + 268, + 262 + ] + ] + ] + }, + { + "task_id": "0000910", + "id": "RH20T_cfg4_task_0003_user_0010_scene_0010_cfg_0004_104122062295#101", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000910/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 103, + 94 + ], + [ + 143, + 106 + ], + [ + 107, + 126 + ], + [ + 99, + 57 + ], + [ + 69, + 115 + ], + [ + 57, + 44 + ] + ], + [ + [ + 116, + 126 + ], + [ + 92, + 160 + ], + [ + 79, + 105 + ], + [ + 149, + 145 + ], + [ + 105, + 69 + ], + [ + 171, + 114 + ] + ], + [ + [ + 59, + 51 + ], + [ + 78, + 38 + ], + [ + 101, + 75 + ], + [ + 15, + 25 + ], + [ + 83, + 92 + ], + [ + 0, + 38 + ] + ], + [ + [ + 78, + 95 + ], + [ + 81, + 55 + ], + [ + 39, + 78 + ], + [ + 118, + 111 + ], + [ + 38, + 115 + ], + [ + 113, + 146 + ] + ] + ] + }, + { + "task_id": "0000912", + "id": "RH20T_cfg4_task_0003_user_0010_scene_0010_cfg_0004_104122062295#216", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000912/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 307, + 267 + ], + [ + 319, + 210 + ], + [ + 329, + 314 + ], + [ + 290, + 230 + ], + [ + 316, + 360 + ], + [ + 281, + 277 + ] + ], + [ + [ + 265, + 253 + ], + [ + 308, + 268 + ], + [ + 303, + 267 + ], + [ + 219, + 236 + ], + [ + 267, + 255 + ], + [ + 184, + 224 + ] + ], + [ + [ + 308, + 315 + ], + [ + 318, + 258 + ], + [ + 250, + 298 + ], + [ + 371, + 333 + ], + [ + 243, + 350 + ], + [ + 359, + 360 + ] + ], + [ + [ + 360, + 358 + ], + [ + 410, + 360 + ], + [ + 391, + 360 + ], + [ + 338, + 310 + ], + [ + 325, + 360 + ], + [ + 287, + 284 + ] + ] + ] + }, + { + "task_id": "0000914", + "id": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_104122062823#218", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000914/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 70 + ], + [ + 332, + 42 + ], + [ + 361, + 81 + ], + [ + 309, + 60 + ], + [ + 361, + 107 + ], + [ + 312, + 85 + ] + ], + [ + [ + 306, + 87 + ], + [ + 322, + 101 + ], + [ + 314, + 59 + ], + [ + 298, + 115 + ], + [ + 297, + 42 + ], + [ + 279, + 100 + ] + ], + [ + [ + 347, + 94 + ], + [ + 327, + 121 + ], + [ + 324, + 80 + ], + [ + 371, + 110 + ], + [ + 343, + 54 + ], + [ + 392, + 82 + ] + ], + [ + [ + 300, + 47 + ], + [ + 326, + 42 + ], + [ + 311, + 79 + ], + [ + 289, + 15 + ], + [ + 283, + 86 + ], + [ + 260, + 20 + ] + ] + ] + }, + { + "task_id": "0000915", + "id": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_104122062823#292", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000915/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 298, + 98 + ], + [ + 317, + 121 + ], + [ + 299, + 117 + ], + [ + 298, + 81 + ], + [ + 277, + 92 + ], + [ + 278, + 58 + ] + ], + [ + [ + 353, + 116 + ], + [ + 359, + 98 + ], + [ + 364, + 87 + ], + [ + 343, + 143 + ], + [ + 358, + 107 + ], + [ + 336, + 164 + ] + ], + [ + [ + 318, + 159 + ], + [ + 328, + 187 + ], + [ + 312, + 145 + ], + [ + 325, + 174 + ], + [ + 301, + 116 + ], + [ + 314, + 142 + ] + ], + [ + [ + 322, + 120 + ], + [ + 320, + 91 + ], + [ + 357, + 123 + ], + [ + 287, + 116 + ], + [ + 356, + 149 + ], + [ + 290, + 141 + ] + ] + ] + }, + { + "task_id": "0000916", + "id": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_104122062823#464", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000916/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 196, + 91 + ], + [ + 172, + 110 + ], + [ + 196, + 103 + ], + [ + 196, + 80 + ], + [ + 222, + 83 + ], + [ + 220, + 62 + ] + ], + [ + [ + 217, + 108 + ], + [ + 213, + 81 + ], + [ + 248, + 114 + ], + [ + 186, + 102 + ], + [ + 250, + 139 + ], + [ + 191, + 127 + ] + ], + [ + [ + 197, + 105 + ], + [ + 182, + 122 + ], + [ + 169, + 104 + ], + [ + 226, + 106 + ], + [ + 184, + 90 + ], + [ + 239, + 90 + ] + ], + [ + [ + 231, + 111 + ], + [ + 220, + 131 + ], + [ + 231, + 85 + ], + [ + 231, + 135 + ], + [ + 242, + 67 + ], + [ + 241, + 116 + ] + ] + ] + }, + { + "task_id": "0000917", + "id": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_104122062823#594", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000917/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 54 + ], + [ + 375, + 56 + ], + [ + 359, + 80 + ], + [ + 337, + 30 + ], + [ + 329, + 78 + ], + [ + 308, + 26 + ] + ], + [ + [ + 397, + 95 + ], + [ + 379, + 117 + ], + [ + 368, + 82 + ], + [ + 425, + 107 + ], + [ + 387, + 57 + ], + [ + 447, + 85 + ] + ], + [ + [ + 372, + 88 + ], + [ + 375, + 61 + ], + [ + 400, + 79 + ], + [ + 343, + 98 + ], + [ + 396, + 104 + ], + [ + 342, + 122 + ] + ], + [ + [ + 389, + 73 + ], + [ + 387, + 41 + ], + [ + 389, + 65 + ], + [ + 388, + 80 + ], + [ + 391, + 98 + ], + [ + 390, + 109 + ] + ] + ] + }, + { + "task_id": "0000918", + "id": "RH20T_cfg4_task_0050_user_0014_scene_0010_cfg_0004_104122062823#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the dish on the dish rack'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000918/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 289, + 56 + ], + [ + 279, + 74 + ], + [ + 321, + 58 + ], + [ + 257, + 54 + ], + [ + 329, + 42 + ], + [ + 267, + 38 + ] + ], + [ + [ + 276, + 80 + ], + [ + 276, + 52 + ], + [ + 308, + 85 + ], + [ + 244, + 75 + ], + [ + 306, + 109 + ], + [ + 245, + 100 + ] + ], + [ + [ + 259, + 108 + ], + [ + 252, + 132 + ], + [ + 271, + 89 + ], + [ + 248, + 126 + ], + [ + 277, + 67 + ], + [ + 255, + 105 + ] + ], + [ + [ + 264, + 94 + ], + [ + 286, + 115 + ], + [ + 277, + 71 + ], + [ + 252, + 116 + ], + [ + 254, + 50 + ], + [ + 232, + 96 + ] + ] + ] + }, + { + "task_id": "0000921", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_036422060215#853", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000921/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 389, + 146 + ], + [ + 410, + 151 + ], + [ + 399, + 154 + ], + [ + 380, + 139 + ], + [ + 377, + 150 + ], + [ + 360, + 135 + ] + ], + [ + [ + 357, + 120 + ], + [ + 341, + 105 + ], + [ + 365, + 126 + ], + [ + 350, + 115 + ], + [ + 381, + 141 + ], + [ + 365, + 128 + ] + ], + [ + [ + 406, + 154 + ], + [ + 396, + 148 + ], + [ + 385, + 164 + ], + [ + 426, + 144 + ], + [ + 395, + 171 + ], + [ + 438, + 150 + ] + ], + [ + [ + 383, + 135 + ], + [ + 382, + 112 + ], + [ + 400, + 129 + ], + [ + 366, + 141 + ], + [ + 401, + 151 + ], + [ + 367, + 164 + ] + ] + ] + }, + { + "task_id": "0000922", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_036422060215#1005", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000922/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 417, + 100 + ], + [ + 428, + 81 + ], + [ + 411, + 115 + ], + [ + 423, + 87 + ], + [ + 400, + 133 + ], + [ + 413, + 105 + ] + ], + [ + [ + 386, + 110 + ], + [ + 390, + 85 + ], + [ + 404, + 109 + ], + [ + 367, + 112 + ], + [ + 400, + 132 + ], + [ + 363, + 136 + ] + ], + [ + [ + 369, + 96 + ], + [ + 371, + 74 + ], + [ + 356, + 85 + ], + [ + 383, + 107 + ], + [ + 354, + 105 + ], + [ + 380, + 128 + ] + ], + [ + [ + 364, + 94 + ], + [ + 379, + 92 + ], + [ + 365, + 120 + ], + [ + 363, + 69 + ], + [ + 351, + 121 + ], + [ + 350, + 72 + ] + ] + ] + }, + { + "task_id": "0000924", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_036422060215#2115", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000924/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 442, + 60 + ], + [ + 449, + 44 + ], + [ + 461, + 63 + ], + [ + 422, + 58 + ], + [ + 454, + 78 + ], + [ + 416, + 73 + ] + ], + [ + [ + 429, + 43 + ], + [ + 437, + 60 + ], + [ + 419, + 57 + ], + [ + 440, + 28 + ], + [ + 412, + 42 + ], + [ + 433, + 12 + ] + ], + [ + [ + 421, + 79 + ], + [ + 405, + 77 + ], + [ + 434, + 78 + ], + [ + 407, + 81 + ], + [ + 450, + 80 + ], + [ + 424, + 83 + ] + ], + [ + [ + 442, + 46 + ], + [ + 436, + 46 + ], + [ + 453, + 27 + ], + [ + 430, + 64 + ], + [ + 458, + 27 + ], + [ + 435, + 62 + ] + ] + ] + }, + { + "task_id": "0000925", + "id": "RH20T_cfg3_task_0035_user_0010_scene_0001_cfg_0003_104122062295#157", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up one small block'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000925/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 121 + ], + [ + 253, + 130 + ], + [ + 270, + 113 + ], + [ + 295, + 128 + ], + [ + 304, + 102 + ], + [ + 325, + 119 + ] + ], + [ + [ + 285, + 146 + ], + [ + 293, + 129 + ], + [ + 309, + 128 + ], + [ + 264, + 162 + ], + [ + 302, + 146 + ], + [ + 255, + 182 + ] + ], + [ + [ + 269, + 138 + ], + [ + 268, + 109 + ], + [ + 254, + 122 + ], + [ + 285, + 156 + ], + [ + 256, + 149 + ], + [ + 285, + 183 + ] + ], + [ + [ + 255, + 151 + ], + [ + 283, + 141 + ], + [ + 241, + 126 + ], + [ + 269, + 178 + ], + [ + 211, + 135 + ], + [ + 238, + 189 + ] + ] + ] + }, + { + "task_id": "0000930", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122064161#498", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000930/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 361, + 208 + ], + [ + 373, + 189 + ], + [ + 352, + 213 + ], + [ + 369, + 205 + ], + [ + 340, + 232 + ], + [ + 358, + 222 + ] + ], + [ + [ + 365, + 226 + ], + [ + 368, + 203 + ], + [ + 384, + 234 + ], + [ + 347, + 219 + ], + [ + 380, + 257 + ], + [ + 344, + 241 + ] + ], + [ + [ + 387, + 222 + ], + [ + 380, + 243 + ], + [ + 382, + 210 + ], + [ + 393, + 234 + ], + [ + 389, + 190 + ], + [ + 399, + 214 + ] + ], + [ + [ + 379, + 227 + ], + [ + 380, + 232 + ], + [ + 374, + 250 + ], + [ + 383, + 203 + ], + [ + 374, + 244 + ], + [ + 382, + 199 + ] + ] + ] + }, + { + "task_id": "0000931", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122064161#755", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000931/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 404, + 258 + ], + [ + 380, + 257 + ], + [ + 404, + 279 + ], + [ + 405, + 238 + ], + [ + 430, + 280 + ], + [ + 429, + 239 + ] + ], + [ + [ + 425, + 238 + ], + [ + 427, + 215 + ], + [ + 426, + 251 + ], + [ + 425, + 226 + ], + [ + 424, + 274 + ], + [ + 423, + 247 + ] + ], + [ + [ + 408, + 234 + ], + [ + 383, + 229 + ], + [ + 403, + 256 + ], + [ + 414, + 214 + ], + [ + 429, + 262 + ], + [ + 439, + 219 + ] + ], + [ + [ + 406, + 229 + ], + [ + 424, + 229 + ], + [ + 395, + 248 + ], + [ + 418, + 209 + ], + [ + 376, + 249 + ], + [ + 399, + 209 + ] + ] + ] + }, + { + "task_id": "0000932", + "id": "RH20T_cfg1_task_0066_user_0010_scene_0004_cfg_0001_750612070853#391", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000932/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 490, + 347 + ], + [ + 520, + 304 + ], + [ + 559, + 360 + ], + [ + 434, + 323 + ], + [ + 525, + 360 + ], + [ + 398, + 360 + ] + ], + [ + [ + 527, + 349 + ], + [ + 539, + 314 + ], + [ + 468, + 318 + ], + [ + 595, + 360 + ], + [ + 463, + 348 + ], + [ + 579, + 360 + ] + ], + [ + [ + 582, + 360 + ], + [ + 581, + 324 + ], + [ + 640, + 360 + ], + [ + 518, + 360 + ], + [ + 640, + 360 + ], + [ + 523, + 360 + ] + ], + [ + [ + 472, + 360 + ], + [ + 444, + 360 + ], + [ + 445, + 360 + ], + [ + 496, + 332 + ], + [ + 479, + 360 + ], + [ + 532, + 323 + ] + ] + ] + }, + { + "task_id": "0000933", + "id": "RH20T_cfg1_task_0066_user_0010_scene_0004_cfg_0001_750612070853#548", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000933/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 640, + 360 + ], + [ + 640, + 357 + ], + [ + 640, + 360 + ], + [ + 640, + 326 + ], + [ + 640, + 360 + ], + [ + 630, + 349 + ] + ], + [ + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ] + ], + [ + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ] + ], + [ + [ + 640, + 360 + ], + [ + 640, + 345 + ], + [ + 640, + 360 + ], + [ + 640, + 356 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ] + ] + ] + }, + { + "task_id": "0000934", + "id": "RH20T_cfg1_task_0066_user_0010_scene_0004_cfg_0001_750612070853#728", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000934/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 627, + 295 + ], + [ + 640, + 353 + ] + ], + [ + [ + 640, + 360 + ], + [ + 629, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 360 + ], + [ + 640, + 359 + ] + ], + [ + [ + 634, + 360 + ], + [ + 640, + 337 + ], + [ + 566, + 360 + ], + [ + 640, + 360 + ], + [ + 550, + 360 + ], + [ + 640, + 360 + ] + ], + [ + [ + 570, + 360 + ], + [ + 594, + 360 + ], + [ + 623, + 360 + ], + [ + 523, + 360 + ], + [ + 596, + 308 + ], + [ + 509, + 360 + ] + ] + ] + }, + { + "task_id": "0000937", + "id": "RH20T_cfg6_task_0013_user_0014_scene_0007_cfg_0006_104122060811#118", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000937/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 394, + 196 + ], + [ + 402, + 208 + ], + [ + 392, + 203 + ], + [ + 397, + 189 + ], + [ + 384, + 191 + ], + [ + 388, + 177 + ] + ], + [ + [ + 384, + 177 + ], + [ + 391, + 167 + ], + [ + 396, + 176 + ], + [ + 372, + 178 + ], + [ + 389, + 186 + ], + [ + 365, + 188 + ] + ], + [ + [ + 371, + 188 + ], + [ + 362, + 191 + ], + [ + 359, + 187 + ], + [ + 384, + 189 + ], + [ + 369, + 185 + ], + [ + 393, + 186 + ] + ], + [ + [ + 377, + 178 + ], + [ + 373, + 183 + ], + [ + 362, + 176 + ], + [ + 392, + 180 + ], + [ + 367, + 172 + ], + [ + 395, + 175 + ] + ] + ] + }, + { + "task_id": "0000940", + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_104122063678#179", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000940/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 303 + ], + [ + 396, + 285 + ], + [ + 354, + 276 + ], + [ + 390, + 330 + ], + [ + 329, + 294 + ], + [ + 364, + 348 + ] + ], + [ + [ + 347, + 295 + ], + [ + 350, + 277 + ], + [ + 358, + 320 + ], + [ + 338, + 273 + ], + [ + 355, + 337 + ], + [ + 336, + 289 + ] + ], + [ + [ + 325, + 288 + ], + [ + 317, + 285 + ], + [ + 313, + 261 + ], + [ + 338, + 316 + ], + [ + 320, + 265 + ], + [ + 345, + 317 + ] + ], + [ + [ + 336, + 285 + ], + [ + 362, + 288 + ], + [ + 324, + 291 + ], + [ + 347, + 280 + ], + [ + 298, + 287 + ], + [ + 323, + 277 + ] + ] + ] + }, + { + "task_id": "0000942", + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_104122063678#339", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000942/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 295, + 190 + ], + [ + 291, + 188 + ], + [ + 319, + 193 + ], + [ + 272, + 186 + ], + [ + 322, + 194 + ], + [ + 277, + 187 + ] + ], + [ + [ + 300, + 171 + ], + [ + 293, + 193 + ], + [ + 290, + 163 + ], + [ + 312, + 181 + ], + [ + 297, + 142 + ], + [ + 320, + 158 + ] + ], + [ + [ + 294, + 185 + ], + [ + 294, + 168 + ], + [ + 292, + 202 + ], + [ + 297, + 170 + ], + [ + 292, + 218 + ], + [ + 297, + 186 + ] + ], + [ + [ + 276, + 196 + ], + [ + 279, + 211 + ], + [ + 298, + 191 + ], + [ + 253, + 200 + ], + [ + 295, + 175 + ], + [ + 249, + 183 + ] + ] + ] + }, + { + "task_id": "0000944", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0010_cfg_0007_104122061602#437", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000944/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 462, + 143 + ], + [ + 463, + 120 + ], + [ + 459, + 156 + ], + [ + 465, + 131 + ], + [ + 458, + 179 + ], + [ + 464, + 152 + ] + ], + [ + [ + 472, + 160 + ], + [ + 446, + 163 + ], + [ + 470, + 146 + ], + [ + 474, + 176 + ], + [ + 495, + 142 + ], + [ + 502, + 173 + ] + ], + [ + [ + 477, + 164 + ], + [ + 490, + 156 + ], + [ + 499, + 164 + ], + [ + 455, + 164 + ], + [ + 486, + 171 + ], + [ + 444, + 171 + ] + ], + [ + [ + 459, + 153 + ], + [ + 459, + 176 + ], + [ + 442, + 149 + ], + [ + 477, + 158 + ], + [ + 442, + 126 + ], + [ + 476, + 133 + ] + ] + ] + }, + { + "task_id": "0000945", + "id": "RH20T_cfg1_task_0015_user_0008_scene_0002_cfg_0001_819112070044#93", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000945/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 386, + 220 + ], + [ + 390, + 230 + ], + [ + 390, + 201 + ], + [ + 382, + 239 + ], + [ + 385, + 189 + ], + [ + 377, + 228 + ] + ], + [ + [ + 395, + 224 + ], + [ + 402, + 233 + ], + [ + 387, + 242 + ], + [ + 403, + 205 + ], + [ + 379, + 233 + ], + [ + 396, + 195 + ] + ], + [ + [ + 365, + 203 + ], + [ + 358, + 186 + ], + [ + 357, + 196 + ], + [ + 373, + 211 + ], + [ + 365, + 214 + ], + [ + 380, + 228 + ] + ], + [ + [ + 371, + 220 + ], + [ + 374, + 204 + ], + [ + 388, + 213 + ], + [ + 353, + 227 + ], + [ + 385, + 228 + ], + [ + 351, + 242 + ] + ] + ] + }, + { + "task_id": "0000948", + "id": "RH20T_cfg5_task_0082_user_0007_scene_0002_cfg_0005_104422070044#112", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000948/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 404, + 182 + ], + [ + 392, + 197 + ], + [ + 411, + 178 + ], + [ + 397, + 187 + ], + [ + 423, + 164 + ], + [ + 409, + 171 + ] + ], + [ + [ + 398, + 177 + ], + [ + 396, + 164 + ], + [ + 412, + 188 + ], + [ + 385, + 166 + ], + [ + 413, + 199 + ], + [ + 388, + 177 + ] + ], + [ + [ + 392, + 166 + ], + [ + 373, + 169 + ], + [ + 398, + 175 + ], + [ + 386, + 158 + ], + [ + 418, + 172 + ], + [ + 404, + 155 + ] + ], + [ + [ + 394, + 168 + ], + [ + 384, + 173 + ], + [ + 402, + 153 + ], + [ + 384, + 183 + ], + [ + 413, + 147 + ], + [ + 394, + 178 + ] + ] + ] + }, + { + "task_id": "0000949", + "id": "RH20T_cfg5_task_0020_user_0010_scene_0007_cfg_0005_104422070044#142", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000949/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 173 + ], + [ + 370, + 167 + ], + [ + 358, + 157 + ], + [ + 368, + 189 + ], + [ + 351, + 163 + ], + [ + 361, + 196 + ] + ], + [ + [ + 369, + 167 + ], + [ + 361, + 172 + ], + [ + 354, + 166 + ], + [ + 386, + 168 + ], + [ + 362, + 161 + ], + [ + 394, + 163 + ] + ], + [ + [ + 361, + 171 + ], + [ + 365, + 156 + ], + [ + 375, + 179 + ], + [ + 348, + 163 + ], + [ + 372, + 193 + ], + [ + 345, + 177 + ] + ], + [ + [ + 358, + 163 + ], + [ + 374, + 171 + ], + [ + 363, + 149 + ], + [ + 354, + 177 + ], + [ + 347, + 142 + ], + [ + 339, + 170 + ] + ] + ] + }, + { + "task_id": "0000951", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0002_cfg_0001_039422060546#238", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000951/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 390, + 127 + ], + [ + 414, + 148 + ], + [ + 372, + 127 + ], + [ + 406, + 128 + ], + [ + 349, + 106 + ], + [ + 384, + 110 + ] + ], + [ + [ + 371, + 116 + ], + [ + 373, + 93 + ], + [ + 363, + 139 + ], + [ + 378, + 96 + ], + [ + 362, + 160 + ], + [ + 376, + 117 + ] + ], + [ + [ + 349, + 124 + ], + [ + 320, + 131 + ], + [ + 335, + 132 + ], + [ + 365, + 115 + ], + [ + 361, + 125 + ], + [ + 393, + 109 + ] + ], + [ + [ + 354, + 114 + ], + [ + 359, + 109 + ], + [ + 358, + 145 + ], + [ + 351, + 83 + ], + [ + 353, + 147 + ], + [ + 347, + 89 + ] + ] + ] + }, + { + "task_id": "0000952", + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_038522063145#155", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000952/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 498, + 261 + ], + [ + 495, + 239 + ], + [ + 483, + 273 + ], + [ + 514, + 248 + ], + [ + 485, + 296 + ], + [ + 518, + 273 + ] + ], + [ + [ + 486, + 246 + ], + [ + 490, + 220 + ], + [ + 466, + 245 + ], + [ + 504, + 246 + ], + [ + 462, + 272 + ], + [ + 500, + 272 + ] + ], + [ + [ + 518, + 256 + ], + [ + 529, + 240 + ], + [ + 544, + 254 + ], + [ + 492, + 258 + ], + [ + 533, + 269 + ], + [ + 483, + 272 + ] + ], + [ + [ + 500, + 276 + ], + [ + 489, + 252 + ], + [ + 480, + 288 + ], + [ + 521, + 263 + ], + [ + 491, + 313 + ], + [ + 534, + 289 + ] + ] + ] + }, + { + "task_id": "0000956", + "id": "RH20T_cfg5_task_0020_user_0010_scene_0007_cfg_0005_037522062165#237", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000956/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 329, + 274 + ], + [ + 332, + 245 + ], + [ + 359, + 278 + ], + [ + 299, + 270 + ], + [ + 356, + 309 + ], + [ + 294, + 300 + ] + ], + [ + [ + 295, + 243 + ], + [ + 269, + 246 + ], + [ + 281, + 231 + ], + [ + 310, + 258 + ], + [ + 305, + 229 + ], + [ + 335, + 254 + ] + ], + [ + [ + 317, + 266 + ], + [ + 311, + 261 + ], + [ + 293, + 251 + ], + [ + 341, + 281 + ], + [ + 299, + 256 + ], + [ + 345, + 284 + ] + ], + [ + [ + 330, + 248 + ], + [ + 341, + 221 + ], + [ + 312, + 247 + ], + [ + 345, + 248 + ], + [ + 300, + 275 + ], + [ + 334, + 273 + ] + ] + ] + }, + { + "task_id": "0000957", + "id": "RH20T_cfg4_task_0024_user_0014_scene_0007_cfg_0004_038522062288#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn off the power strip by pressing the button'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000957/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 218, + 153 + ], + [ + 208, + 129 + ], + [ + 240, + 138 + ], + [ + 197, + 167 + ], + [ + 251, + 163 + ], + [ + 207, + 191 + ] + ], + [ + [ + 247, + 181 + ], + [ + 256, + 154 + ], + [ + 217, + 170 + ], + [ + 276, + 192 + ], + [ + 207, + 199 + ], + [ + 268, + 220 + ] + ], + [ + [ + 266, + 185 + ], + [ + 259, + 157 + ], + [ + 293, + 180 + ], + [ + 239, + 189 + ], + [ + 299, + 208 + ], + [ + 246, + 216 + ] + ], + [ + [ + 244, + 166 + ], + [ + 249, + 153 + ], + [ + 214, + 165 + ], + [ + 272, + 168 + ], + [ + 211, + 177 + ], + [ + 266, + 180 + ] + ] + ] + }, + { + "task_id": "0000960", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0009_cfg_0004_104122062823#324", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000960/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 255, + 65 + ], + [ + 249, + 40 + ], + [ + 275, + 74 + ], + [ + 236, + 57 + ], + [ + 280, + 97 + ], + [ + 242, + 80 + ] + ], + [ + [ + 264, + 85 + ], + [ + 263, + 110 + ], + [ + 236, + 86 + ], + [ + 292, + 85 + ], + [ + 237, + 62 + ], + [ + 292, + 61 + ] + ], + [ + [ + 245, + 57 + ], + [ + 218, + 60 + ], + [ + 254, + 78 + ], + [ + 235, + 38 + ], + [ + 280, + 75 + ], + [ + 260, + 36 + ] + ], + [ + [ + 269, + 101 + ], + [ + 262, + 126 + ], + [ + 252, + 89 + ], + [ + 287, + 114 + ], + [ + 258, + 63 + ], + [ + 295, + 88 + ] + ] + ] + }, + { + "task_id": "0000962", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0009_cfg_0004_104122062823#559", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000962/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 293, + 75 + ], + [ + 289, + 49 + ], + [ + 317, + 83 + ], + [ + 270, + 67 + ], + [ + 320, + 107 + ], + [ + 275, + 91 + ] + ], + [ + [ + 301, + 69 + ], + [ + 300, + 91 + ], + [ + 330, + 75 + ], + [ + 272, + 63 + ], + [ + 330, + 54 + ], + [ + 274, + 42 + ] + ], + [ + [ + 314, + 55 + ], + [ + 287, + 67 + ], + [ + 318, + 56 + ], + [ + 311, + 54 + ], + [ + 347, + 44 + ], + [ + 337, + 43 + ] + ], + [ + [ + 279, + 61 + ], + [ + 292, + 50 + ], + [ + 303, + 62 + ], + [ + 255, + 60 + ], + [ + 289, + 75 + ], + [ + 240, + 72 + ] + ] + ] + }, + { + "task_id": "0000964", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0009_cfg_0004_104122062823#773", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000964/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 341, + 52 + ], + [ + 356, + 37 + ], + [ + 358, + 78 + ], + [ + 323, + 25 + ], + [ + 343, + 95 + ], + [ + 306, + 40 + ] + ], + [ + [ + 340, + 102 + ], + [ + 322, + 91 + ], + [ + 316, + 117 + ], + [ + 362, + 87 + ], + [ + 335, + 130 + ], + [ + 382, + 98 + ] + ], + [ + [ + 336, + 80 + ], + [ + 334, + 53 + ], + [ + 363, + 86 + ], + [ + 310, + 74 + ], + [ + 364, + 111 + ], + [ + 312, + 98 + ] + ], + [ + [ + 329, + 84 + ], + [ + 306, + 78 + ], + [ + 345, + 61 + ], + [ + 314, + 104 + ], + [ + 366, + 67 + ], + [ + 335, + 108 + ] + ] + ] + }, + { + "task_id": "0000966", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0009_cfg_0004_104122062823#1215", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000966/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 403, + 125 + ], + [ + 400, + 156 + ], + [ + 388, + 130 + ], + [ + 419, + 119 + ], + [ + 390, + 102 + ], + [ + 421, + 87 + ] + ], + [ + [ + 392, + 140 + ], + [ + 378, + 140 + ], + [ + 366, + 154 + ], + [ + 420, + 126 + ], + [ + 378, + 154 + ], + [ + 430, + 127 + ] + ], + [ + [ + 366, + 94 + ], + [ + 390, + 113 + ], + [ + 385, + 70 + ], + [ + 347, + 118 + ], + [ + 361, + 50 + ], + [ + 323, + 99 + ] + ], + [ + [ + 386, + 98 + ], + [ + 389, + 70 + ], + [ + 416, + 108 + ], + [ + 359, + 88 + ], + [ + 411, + 133 + ], + [ + 357, + 112 + ] + ] + ] + }, + { + "task_id": "0000969", + "id": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_104122060811#120", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000969/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 343, + 156 + ], + [ + 330, + 152 + ], + [ + 339, + 151 + ], + [ + 346, + 161 + ], + [ + 353, + 156 + ], + [ + 360, + 166 + ] + ], + [ + [ + 347, + 163 + ], + [ + 341, + 167 + ], + [ + 358, + 156 + ], + [ + 336, + 170 + ], + [ + 365, + 151 + ], + [ + 342, + 166 + ] + ], + [ + [ + 337, + 141 + ], + [ + 333, + 136 + ], + [ + 351, + 135 + ], + [ + 323, + 147 + ], + [ + 354, + 139 + ], + [ + 327, + 151 + ] + ], + [ + [ + 348, + 153 + ], + [ + 356, + 145 + ], + [ + 360, + 152 + ], + [ + 336, + 154 + ], + [ + 353, + 160 + ], + [ + 329, + 162 + ] + ] + ] + }, + { + "task_id": "0000970", + "id": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_104122060811#231", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000970/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 368, + 122 + ], + [ + 359, + 126 + ], + [ + 368, + 135 + ], + [ + 369, + 108 + ], + [ + 376, + 131 + ], + [ + 378, + 105 + ] + ], + [ + [ + 377, + 129 + ], + [ + 391, + 124 + ], + [ + 377, + 135 + ], + [ + 377, + 122 + ], + [ + 364, + 140 + ], + [ + 363, + 126 + ] + ], + [ + [ + 372, + 130 + ], + [ + 373, + 119 + ], + [ + 359, + 126 + ], + [ + 385, + 134 + ], + [ + 358, + 136 + ], + [ + 385, + 145 + ] + ], + [ + [ + 392, + 142 + ], + [ + 399, + 153 + ], + [ + 380, + 149 + ], + [ + 405, + 135 + ], + [ + 373, + 138 + ], + [ + 397, + 124 + ] + ] + ] + }, + { + "task_id": "0000971", + "id": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_104122060811#377", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000971/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 115 + ], + [ + 336, + 126 + ], + [ + 329, + 125 + ], + [ + 339, + 105 + ], + [ + 327, + 115 + ], + [ + 337, + 95 + ] + ], + [ + [ + 326, + 123 + ], + [ + 330, + 135 + ], + [ + 317, + 120 + ], + [ + 335, + 127 + ], + [ + 314, + 108 + ], + [ + 332, + 115 + ] + ], + [ + [ + 324, + 124 + ], + [ + 337, + 121 + ], + [ + 322, + 131 + ], + [ + 325, + 116 + ], + [ + 309, + 133 + ], + [ + 311, + 118 + ] + ], + [ + [ + 315, + 121 + ], + [ + 306, + 112 + ], + [ + 309, + 132 + ], + [ + 322, + 111 + ], + [ + 319, + 140 + ], + [ + 331, + 119 + ] + ] + ] + }, + { + "task_id": "0000972", + "id": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_104122060811#518", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000972/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 323, + 143 + ], + [ + 329, + 151 + ], + [ + 312, + 138 + ], + [ + 334, + 148 + ], + [ + 306, + 130 + ], + [ + 329, + 140 + ] + ], + [ + [ + 318, + 148 + ], + [ + 328, + 141 + ], + [ + 328, + 150 + ], + [ + 307, + 147 + ], + [ + 318, + 157 + ], + [ + 298, + 154 + ] + ], + [ + [ + 310, + 146 + ], + [ + 297, + 153 + ], + [ + 317, + 157 + ], + [ + 302, + 134 + ], + [ + 330, + 150 + ], + [ + 315, + 126 + ] + ], + [ + [ + 327, + 141 + ], + [ + 317, + 141 + ], + [ + 323, + 155 + ], + [ + 332, + 127 + ], + [ + 332, + 155 + ], + [ + 341, + 128 + ] + ] + ] + }, + { + "task_id": "0000973", + "id": "RH20T_cfg1_task_0005_user_0008_scene_0006_cfg_0001_819112071037#172", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000973/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 137, + 117 + ], + [ + 153, + 36 + ], + [ + 45, + 109 + ], + [ + 210, + 123 + ], + [ + 34, + 192 + ], + [ + 194, + 187 + ] + ], + [ + [ + 113, + 213 + ], + [ + 157, + 258 + ], + [ + 40, + 256 + ], + [ + 174, + 177 + ], + [ + 2, + 210 + ], + [ + 134, + 143 + ] + ], + [ + [ + 91, + 95 + ], + [ + 100, + 101 + ], + [ + 99, + 3 + ], + [ + 84, + 182 + ], + [ + 91, + 12 + ], + [ + 78, + 165 + ] + ], + [ + [ + 76, + 192 + ], + [ + 34, + 109 + ], + [ + 0, + 219 + ], + [ + 163, + 167 + ], + [ + 26, + 289 + ], + [ + 191, + 238 + ] + ] + ] + }, + { + "task_id": "0000974", + "id": "RH20T_cfg2_task_0201_user_0007_scene_0008_cfg_0002_104122063678#335", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Insert the tip of a medium pipette into the holder for medium pipette tips'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000974/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 418, + 212 + ], + [ + 433, + 204 + ], + [ + 429, + 202 + ], + [ + 407, + 221 + ], + [ + 412, + 210 + ], + [ + 391, + 229 + ] + ], + [ + [ + 400, + 188 + ], + [ + 391, + 167 + ], + [ + 404, + 190 + ], + [ + 396, + 185 + ], + [ + 413, + 212 + ], + [ + 405, + 205 + ] + ], + [ + [ + 415, + 173 + ], + [ + 397, + 185 + ], + [ + 412, + 161 + ], + [ + 418, + 186 + ], + [ + 429, + 149 + ], + [ + 437, + 174 + ] + ], + [ + [ + 399, + 192 + ], + [ + 399, + 171 + ], + [ + 377, + 193 + ], + [ + 421, + 192 + ], + [ + 376, + 213 + ], + [ + 420, + 213 + ] + ] + ] + }, + { + "task_id": "0000976", + "id": "RH20T_cfg1_task_0210_user_0015_scene_0008_cfg_0001_819112071037#164", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Chop the onions'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000976/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 640, + 124 + ], + [ + 640, + 180 + ], + [ + 598, + 135 + ], + [ + 640, + 113 + ], + [ + 591, + 76 + ], + [ + 640, + 56 + ] + ], + [ + [ + 640, + 114 + ], + [ + 640, + 61 + ], + [ + 598, + 113 + ], + [ + 640, + 115 + ], + [ + 583, + 162 + ], + [ + 640, + 162 + ] + ], + [ + [ + 640, + 103 + ], + [ + 633, + 70 + ], + [ + 640, + 115 + ], + [ + 599, + 90 + ], + [ + 640, + 153 + ], + [ + 610, + 126 + ] + ], + [ + [ + 627, + 138 + ], + [ + 640, + 140 + ], + [ + 591, + 121 + ], + [ + 640, + 153 + ], + [ + 549, + 120 + ], + [ + 615, + 149 + ] + ] + ] + }, + { + "task_id": "0000979", + "id": "RH20T_cfg3_task_0032_user_0011_scene_0001_cfg_0003_036422060909#362", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000979/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 311, + 116 + ], + [ + 333, + 123 + ], + [ + 304, + 136 + ], + [ + 317, + 94 + ], + [ + 282, + 129 + ], + [ + 294, + 86 + ] + ], + [ + [ + 289, + 148 + ], + [ + 274, + 129 + ], + [ + 277, + 158 + ], + [ + 300, + 138 + ], + [ + 293, + 178 + ], + [ + 314, + 157 + ] + ], + [ + [ + 298, + 120 + ], + [ + 304, + 101 + ], + [ + 316, + 114 + ], + [ + 279, + 126 + ], + [ + 310, + 132 + ], + [ + 274, + 144 + ] + ], + [ + [ + 285, + 128 + ], + [ + 268, + 117 + ], + [ + 302, + 120 + ], + [ + 267, + 136 + ], + [ + 319, + 130 + ], + [ + 285, + 148 + ] + ] + ] + }, + { + "task_id": "0000980", + "id": "RH20T_cfg2_task_0080_user_0014_scene_0006_cfg_0002_104422070044#103", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Flip over and spread out the paper that is laid flat on the table'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000980/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 369, + 163 + ], + [ + 376, + 150 + ], + [ + 383, + 172 + ], + [ + 355, + 154 + ], + [ + 376, + 185 + ], + [ + 349, + 167 + ] + ], + [ + [ + 352, + 186 + ], + [ + 361, + 174 + ], + [ + 349, + 175 + ], + [ + 355, + 198 + ], + [ + 341, + 186 + ], + [ + 346, + 209 + ] + ], + [ + [ + 355, + 168 + ], + [ + 340, + 162 + ], + [ + 363, + 157 + ], + [ + 346, + 180 + ], + [ + 378, + 162 + ], + [ + 362, + 186 + ] + ], + [ + [ + 364, + 176 + ], + [ + 363, + 164 + ], + [ + 358, + 188 + ], + [ + 369, + 164 + ], + [ + 359, + 199 + ], + [ + 369, + 176 + ] + ] + ] + }, + { + "task_id": "0000981", + "id": "RH20T_cfg6_task_0039_user_0014_scene_0005_cfg_0006_104122063633#96", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000981/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 253, + 183 + ], + [ + 275, + 191 + ], + [ + 247, + 206 + ], + [ + 259, + 162 + ], + [ + 224, + 198 + ], + [ + 237, + 153 + ] + ], + [ + [ + 237, + 176 + ], + [ + 235, + 158 + ], + [ + 241, + 158 + ], + [ + 232, + 196 + ], + [ + 242, + 174 + ], + [ + 234, + 212 + ] + ], + [ + [ + 234, + 195 + ], + [ + 231, + 207 + ], + [ + 256, + 208 + ], + [ + 212, + 183 + ], + [ + 258, + 197 + ], + [ + 216, + 173 + ] + ], + [ + [ + 212, + 175 + ], + [ + 196, + 162 + ], + [ + 222, + 192 + ], + [ + 203, + 159 + ], + [ + 237, + 204 + ], + [ + 218, + 172 + ] + ] + ] + }, + { + "task_id": "0000982", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0005_cfg_0004_038522062288#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000982/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 246, + 184 + ], + [ + 254, + 155 + ], + [ + 278, + 193 + ], + [ + 212, + 176 + ], + [ + 269, + 220 + ], + [ + 205, + 204 + ] + ], + [ + [ + 219, + 223 + ], + [ + 252, + 235 + ], + [ + 220, + 228 + ], + [ + 219, + 219 + ], + [ + 183, + 215 + ], + [ + 187, + 208 + ] + ], + [ + [ + 214, + 179 + ], + [ + 202, + 154 + ], + [ + 197, + 161 + ], + [ + 231, + 199 + ], + [ + 208, + 184 + ], + [ + 241, + 222 + ] + ], + [ + [ + 210, + 206 + ], + [ + 205, + 173 + ], + [ + 216, + 224 + ], + [ + 205, + 190 + ], + [ + 221, + 256 + ], + [ + 209, + 220 + ] + ] + ] + }, + { + "task_id": "0000984", + "id": "RH20T_cfg5_task_0010_user_0010_scene_0009_cfg_0005_037522062165#306", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000984/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 279, + 293 + ], + [ + 272, + 322 + ], + [ + 301, + 302 + ], + [ + 256, + 283 + ], + [ + 308, + 275 + ], + [ + 265, + 254 + ] + ], + [ + [ + 297, + 289 + ], + [ + 311, + 269 + ], + [ + 276, + 293 + ], + [ + 319, + 285 + ], + [ + 260, + 315 + ], + [ + 304, + 309 + ] + ], + [ + [ + 300, + 303 + ], + [ + 302, + 290 + ], + [ + 276, + 286 + ], + [ + 324, + 321 + ], + [ + 275, + 298 + ], + [ + 320, + 331 + ] + ], + [ + [ + 312, + 295 + ], + [ + 326, + 291 + ], + [ + 329, + 275 + ], + [ + 297, + 314 + ], + [ + 315, + 280 + ], + [ + 281, + 321 + ] + ] + ] + }, + { + "task_id": "0000986", + "id": "RH20T_cfg4_task_0020_user_0016_scene_0002_cfg_0004_104122062823#184", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000986/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 189, + 50 + ], + [ + 215, + 64 + ], + [ + 185, + 24 + ], + [ + 193, + 78 + ], + [ + 156, + 8 + ], + [ + 162, + 64 + ] + ], + [ + [ + 129, + 74 + ], + [ + 133, + 83 + ], + [ + 131, + 41 + ], + [ + 127, + 108 + ], + [ + 127, + 28 + ], + [ + 123, + 101 + ] + ], + [ + [ + 159, + 82 + ], + [ + 162, + 56 + ], + [ + 193, + 83 + ], + [ + 121, + 81 + ], + [ + 189, + 105 + ], + [ + 120, + 104 + ] + ], + [ + [ + 132, + 106 + ], + [ + 135, + 80 + ], + [ + 167, + 107 + ], + [ + 96, + 104 + ], + [ + 165, + 134 + ], + [ + 91, + 132 + ] + ] + ] + }, + { + "task_id": "0000987", + "id": "RH20T_cfg1_task_0041_user_0001_scene_0006_cfg_0001_819112071037#152", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the tabletop with a sponge'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000987/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 214 + ], + [ + 369, + 252 + ], + [ + 339, + 269 + ], + [ + 389, + 153 + ], + [ + 337, + 235 + ], + [ + 380, + 130 + ] + ], + [ + [ + 378, + 215 + ], + [ + 366, + 205 + ], + [ + 414, + 168 + ], + [ + 341, + 261 + ], + [ + 435, + 173 + ], + [ + 351, + 280 + ] + ], + [ + [ + 448, + 230 + ], + [ + 442, + 176 + ], + [ + 394, + 228 + ], + [ + 498, + 232 + ], + [ + 398, + 289 + ], + [ + 505, + 289 + ] + ], + [ + [ + 409, + 204 + ], + [ + 414, + 141 + ], + [ + 345, + 201 + ], + [ + 473, + 206 + ], + [ + 342, + 258 + ], + [ + 465, + 263 + ] + ] + ] + }, + { + "task_id": "0000989", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122060811#259", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000989/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 138 + ], + [ + 340, + 134 + ], + [ + 321, + 125 + ], + [ + 331, + 152 + ], + [ + 307, + 129 + ], + [ + 317, + 156 + ] + ], + [ + [ + 335, + 133 + ], + [ + 338, + 126 + ], + [ + 348, + 130 + ], + [ + 321, + 135 + ], + [ + 345, + 136 + ], + [ + 319, + 142 + ] + ], + [ + [ + 332, + 142 + ], + [ + 328, + 155 + ], + [ + 317, + 139 + ], + [ + 346, + 144 + ], + [ + 320, + 126 + ], + [ + 349, + 131 + ] + ], + [ + [ + 349, + 137 + ], + [ + 345, + 124 + ], + [ + 349, + 142 + ], + [ + 349, + 132 + ], + [ + 353, + 154 + ], + [ + 353, + 145 + ] + ] + ] + }, + { + "task_id": "0000990", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122060811#375", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000990/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 353, + 134 + ], + [ + 363, + 126 + ], + [ + 345, + 123 + ], + [ + 361, + 146 + ], + [ + 335, + 131 + ], + [ + 350, + 154 + ] + ], + [ + [ + 354, + 142 + ], + [ + 350, + 133 + ], + [ + 362, + 131 + ], + [ + 347, + 154 + ], + [ + 366, + 140 + ], + [ + 351, + 163 + ] + ], + [ + [ + 332, + 143 + ], + [ + 322, + 140 + ], + [ + 333, + 129 + ], + [ + 331, + 157 + ], + [ + 344, + 132 + ], + [ + 342, + 161 + ] + ], + [ + [ + 350, + 145 + ], + [ + 362, + 138 + ], + [ + 357, + 148 + ], + [ + 342, + 142 + ], + [ + 346, + 154 + ], + [ + 330, + 149 + ] + ] + ] + }, + { + "task_id": "0000994", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122063633#305", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0000994/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 181, + 207 + ], + [ + 187, + 187 + ], + [ + 183, + 221 + ], + [ + 179, + 192 + ], + [ + 177, + 242 + ], + [ + 172, + 214 + ] + ], + [ + [ + 193, + 215 + ], + [ + 180, + 235 + ], + [ + 178, + 198 + ], + [ + 206, + 231 + ], + [ + 191, + 178 + ], + [ + 217, + 211 + ] + ], + [ + [ + 206, + 213 + ], + [ + 223, + 197 + ], + [ + 194, + 214 + ], + [ + 216, + 211 + ], + [ + 176, + 230 + ], + [ + 200, + 226 + ] + ], + [ + [ + 202, + 241 + ], + [ + 183, + 238 + ], + [ + 197, + 266 + ], + [ + 207, + 216 + ], + [ + 217, + 270 + ], + [ + 227, + 219 + ] + ] + ] + }, + { + "task_id": "0001000", + "id": "RH20T_cfg1_task_0070_user_0014_scene_0008_cfg_0001_038522062547#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001000/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 459, + 162 + ], + [ + 453, + 142 + ], + [ + 446, + 177 + ], + [ + 473, + 146 + ], + [ + 451, + 199 + ], + [ + 480, + 168 + ] + ], + [ + [ + 440, + 146 + ], + [ + 461, + 138 + ], + [ + 450, + 170 + ], + [ + 430, + 122 + ], + [ + 428, + 178 + ], + [ + 409, + 129 + ] + ], + [ + [ + 435, + 165 + ], + [ + 461, + 166 + ], + [ + 433, + 145 + ], + [ + 437, + 183 + ], + [ + 407, + 145 + ], + [ + 412, + 182 + ] + ], + [ + [ + 441, + 169 + ], + [ + 446, + 156 + ], + [ + 446, + 192 + ], + [ + 436, + 147 + ], + [ + 441, + 204 + ], + [ + 432, + 160 + ] + ] + ] + }, + { + "task_id": "0001002", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_104122063678#487", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001002/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 280, + 227 + ], + [ + 295, + 201 + ], + [ + 290, + 249 + ], + [ + 271, + 208 + ], + [ + 275, + 276 + ], + [ + 258, + 233 + ] + ], + [ + [ + 329, + 231 + ], + [ + 338, + 200 + ], + [ + 305, + 231 + ], + [ + 355, + 230 + ], + [ + 295, + 262 + ], + [ + 346, + 265 + ] + ], + [ + [ + 265, + 228 + ], + [ + 280, + 226 + ], + [ + 293, + 250 + ], + [ + 239, + 207 + ], + [ + 278, + 254 + ], + [ + 221, + 207 + ] + ], + [ + [ + 292, + 196 + ], + [ + 320, + 198 + ], + [ + 279, + 184 + ], + [ + 308, + 210 + ], + [ + 250, + 182 + ], + [ + 276, + 209 + ] + ] + ] + }, + { + "task_id": "0001003", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_104122063678#851", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001003/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 242, + 287 + ], + [ + 245, + 296 + ], + [ + 219, + 260 + ], + [ + 266, + 316 + ], + [ + 214, + 248 + ], + [ + 264, + 308 + ] + ], + [ + [ + 317, + 261 + ], + [ + 313, + 279 + ], + [ + 332, + 233 + ], + [ + 301, + 292 + ], + [ + 337, + 212 + ], + [ + 305, + 273 + ] + ], + [ + [ + 284, + 255 + ], + [ + 304, + 235 + ], + [ + 290, + 286 + ], + [ + 278, + 227 + ], + [ + 269, + 305 + ], + [ + 260, + 246 + ] + ], + [ + [ + 269, + 253 + ], + [ + 260, + 234 + ], + [ + 272, + 219 + ], + [ + 265, + 286 + ], + [ + 283, + 239 + ], + [ + 275, + 310 + ] + ] + ] + }, + { + "task_id": "0001005", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_104122063678#1497", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001005/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 395, + 250 + ], + [ + 415, + 265 + ], + [ + 381, + 265 + ], + [ + 409, + 234 + ], + [ + 361, + 251 + ], + [ + 388, + 218 + ] + ], + [ + [ + 401, + 230 + ], + [ + 410, + 214 + ], + [ + 379, + 231 + ], + [ + 423, + 230 + ], + [ + 371, + 247 + ], + [ + 414, + 245 + ] + ], + [ + [ + 382, + 203 + ], + [ + 391, + 186 + ], + [ + 394, + 193 + ], + [ + 372, + 213 + ], + [ + 384, + 211 + ], + [ + 362, + 231 + ] + ], + [ + [ + 427, + 219 + ], + [ + 414, + 225 + ], + [ + 418, + 240 + ], + [ + 435, + 198 + ], + [ + 430, + 234 + ], + [ + 447, + 193 + ] + ] + ] + }, + { + "task_id": "0001006", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_104122063678#2015", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001006/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 235, + 344 + ], + [ + 254, + 335 + ], + [ + 235, + 360 + ], + [ + 235, + 309 + ], + [ + 213, + 360 + ], + [ + 214, + 317 + ] + ], + [ + [ + 267, + 360 + ], + [ + 296, + 352 + ], + [ + 273, + 360 + ], + [ + 262, + 344 + ], + [ + 242, + 360 + ], + [ + 235, + 360 + ] + ], + [ + [ + 241, + 360 + ], + [ + 246, + 360 + ], + [ + 277, + 360 + ], + [ + 203, + 360 + ], + [ + 272, + 360 + ], + [ + 198, + 360 + ] + ], + [ + [ + 266, + 349 + ], + [ + 248, + 334 + ], + [ + 262, + 320 + ], + [ + 270, + 360 + ], + [ + 278, + 334 + ], + [ + 288, + 360 + ] + ] + ] + }, + { + "task_id": "0001007", + "id": "RH20T_cfg5_task_0083_user_0007_scene_0005_cfg_0005_037522062165#263", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Swat with a flyswatter'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001007/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 337, + 357 + ], + [ + 369, + 360 + ], + [ + 332, + 360 + ], + [ + 343, + 324 + ], + [ + 300, + 360 + ], + [ + 311, + 319 + ] + ], + [ + [ + 328, + 360 + ], + [ + 332, + 360 + ], + [ + 346, + 356 + ], + [ + 309, + 360 + ], + [ + 342, + 360 + ], + [ + 307, + 360 + ] + ], + [ + [ + 346, + 360 + ], + [ + 341, + 360 + ], + [ + 318, + 360 + ], + [ + 375, + 360 + ], + [ + 324, + 360 + ], + [ + 378, + 360 + ] + ], + [ + [ + 351, + 360 + ], + [ + 340, + 347 + ], + [ + 382, + 358 + ], + [ + 321, + 360 + ], + [ + 391, + 360 + ], + [ + 332, + 360 + ] + ] + ] + }, + { + "task_id": "0001008", + "id": "RH20T_cfg4_task_0014_user_0010_scene_0002_cfg_0004_104422070011#79", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the object from the scale'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001008/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 164 + ], + [ + 343, + 148 + ], + [ + 326, + 180 + ], + [ + 370, + 147 + ], + [ + 329, + 198 + ], + [ + 376, + 165 + ] + ], + [ + [ + 323, + 190 + ], + [ + 346, + 208 + ], + [ + 321, + 189 + ], + [ + 325, + 190 + ], + [ + 299, + 172 + ], + [ + 301, + 171 + ] + ], + [ + [ + 340, + 172 + ], + [ + 341, + 147 + ], + [ + 344, + 158 + ], + [ + 336, + 185 + ], + [ + 342, + 186 + ], + [ + 335, + 210 + ] + ], + [ + [ + 332, + 163 + ], + [ + 344, + 151 + ], + [ + 307, + 167 + ], + [ + 355, + 159 + ], + [ + 297, + 178 + ], + [ + 343, + 170 + ] + ] + ] + }, + { + "task_id": "0001010", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_819112070044#411", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001010/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 318, + 225 + ], + [ + 331, + 205 + ], + [ + 318, + 223 + ], + [ + 319, + 228 + ], + [ + 305, + 245 + ], + [ + 307, + 248 + ] + ], + [ + [ + 348, + 224 + ], + [ + 360, + 236 + ], + [ + 368, + 211 + ], + [ + 329, + 236 + ], + [ + 356, + 200 + ], + [ + 319, + 224 + ] + ], + [ + [ + 337, + 244 + ], + [ + 340, + 230 + ], + [ + 354, + 233 + ], + [ + 319, + 256 + ], + [ + 351, + 247 + ], + [ + 317, + 270 + ] + ], + [ + [ + 321, + 222 + ], + [ + 312, + 236 + ], + [ + 307, + 234 + ], + [ + 335, + 210 + ], + [ + 315, + 221 + ], + [ + 343, + 198 + ] + ] + ] + }, + { + "task_id": "0001013", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0006_cfg_0001_038522062547#150", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001013/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 443, + 181 + ], + [ + 463, + 168 + ], + [ + 427, + 159 + ], + [ + 460, + 203 + ], + [ + 410, + 172 + ], + [ + 441, + 214 + ] + ], + [ + [ + 433, + 180 + ], + [ + 437, + 159 + ], + [ + 438, + 200 + ], + [ + 429, + 162 + ], + [ + 434, + 219 + ], + [ + 426, + 181 + ] + ], + [ + [ + 422, + 201 + ], + [ + 413, + 227 + ], + [ + 396, + 192 + ], + [ + 449, + 211 + ], + [ + 405, + 167 + ], + [ + 458, + 185 + ] + ], + [ + [ + 449, + 163 + ], + [ + 445, + 137 + ], + [ + 469, + 152 + ], + [ + 432, + 172 + ], + [ + 474, + 181 + ], + [ + 436, + 198 + ] + ] + ] + }, + { + "task_id": "0001017", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_750612070851#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001017/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 607, + 256 + ], + [ + 607, + 268 + ], + [ + 597, + 218 + ], + [ + 618, + 297 + ], + [ + 597, + 201 + ], + [ + 619, + 285 + ] + ], + [ + [ + 551, + 262 + ], + [ + 580, + 240 + ], + [ + 520, + 261 + ], + [ + 579, + 262 + ], + [ + 495, + 281 + ], + [ + 553, + 280 + ] + ], + [ + [ + 513, + 288 + ], + [ + 551, + 307 + ], + [ + 534, + 257 + ], + [ + 492, + 319 + ], + [ + 500, + 243 + ], + [ + 459, + 300 + ] + ], + [ + [ + 552, + 268 + ], + [ + 543, + 237 + ], + [ + 563, + 288 + ], + [ + 541, + 245 + ], + [ + 574, + 322 + ], + [ + 551, + 280 + ] + ] + ] + }, + { + "task_id": "0001018", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_750612070851#442", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001018/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 526, + 360 + ], + [ + 543, + 360 + ], + [ + 478, + 360 + ], + [ + 573, + 360 + ], + [ + 467, + 360 + ], + [ + 555, + 360 + ] + ], + [ + [ + 507, + 360 + ], + [ + 504, + 360 + ], + [ + 500, + 360 + ], + [ + 514, + 360 + ], + [ + 503, + 319 + ], + [ + 516, + 349 + ] + ], + [ + [ + 541, + 360 + ], + [ + 521, + 360 + ], + [ + 584, + 360 + ], + [ + 497, + 360 + ], + [ + 612, + 360 + ], + [ + 518, + 360 + ] + ], + [ + [ + 526, + 337 + ], + [ + 549, + 360 + ], + [ + 490, + 360 + ], + [ + 565, + 307 + ], + [ + 475, + 330 + ], + [ + 542, + 274 + ] + ] + ] + }, + { + "task_id": "0001019", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_750612070851#908", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001019/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 500, + 360 + ], + [ + 546, + 360 + ], + [ + 498, + 360 + ], + [ + 503, + 360 + ], + [ + 460, + 360 + ], + [ + 462, + 357 + ] + ], + [ + [ + 459, + 360 + ], + [ + 471, + 357 + ], + [ + 411, + 360 + ], + [ + 508, + 360 + ], + [ + 405, + 360 + ], + [ + 493, + 360 + ] + ], + [ + [ + 466, + 360 + ], + [ + 467, + 360 + ], + [ + 504, + 360 + ], + [ + 425, + 360 + ], + [ + 501, + 354 + ], + [ + 426, + 334 + ] + ], + [ + [ + 477, + 360 + ], + [ + 451, + 360 + ], + [ + 490, + 355 + ], + [ + 465, + 360 + ], + [ + 515, + 318 + ], + [ + 488, + 331 + ] + ] + ] + }, + { + "task_id": "0001020", + "id": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_038522062547#118", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Insert the tip of a medium pipette into the holder for medium pipette tips'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001020/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 421, + 212 + ], + [ + 435, + 204 + ], + [ + 439, + 200 + ], + [ + 402, + 224 + ], + [ + 426, + 207 + ], + [ + 389, + 231 + ] + ], + [ + [ + 393, + 187 + ], + [ + 381, + 197 + ], + [ + 415, + 196 + ], + [ + 371, + 177 + ], + [ + 429, + 186 + ], + [ + 384, + 165 + ] + ], + [ + [ + 401, + 212 + ], + [ + 375, + 222 + ], + [ + 397, + 221 + ], + [ + 405, + 201 + ], + [ + 420, + 211 + ], + [ + 432, + 191 + ] + ], + [ + [ + 413, + 200 + ], + [ + 420, + 190 + ], + [ + 408, + 224 + ], + [ + 418, + 176 + ], + [ + 402, + 232 + ], + [ + 412, + 186 + ] + ] + ] + }, + { + "task_id": "0001023", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_819112071037#336", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001023/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 561, + 360 + ], + [ + 640, + 335 + ], + [ + 492, + 282 + ], + [ + 640, + 360 + ], + [ + 392, + 337 + ], + [ + 523, + 360 + ] + ], + [ + [ + 492, + 288 + ], + [ + 460, + 178 + ], + [ + 506, + 340 + ], + [ + 483, + 256 + ], + [ + 547, + 360 + ], + [ + 509, + 343 + ] + ], + [ + [ + 516, + 252 + ], + [ + 391, + 317 + ], + [ + 467, + 204 + ], + [ + 545, + 281 + ], + [ + 640, + 85 + ], + [ + 640, + 220 + ] + ], + [ + [ + 389, + 263 + ], + [ + 282, + 255 + ], + [ + 353, + 161 + ], + [ + 416, + 339 + ], + [ + 516, + 155 + ], + [ + 539, + 358 + ] + ] + ] + }, + { + "task_id": "0001026", + "id": "RH20T_cfg1_task_0079_user_0015_scene_0003_cfg_0001_035622060973#126", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open the microwave door'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001026/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 154 + ], + [ + 315, + 159 + ], + [ + 323, + 165 + ], + [ + 285, + 142 + ], + [ + 313, + 160 + ], + [ + 277, + 137 + ] + ], + [ + [ + 305, + 163 + ], + [ + 303, + 149 + ], + [ + 295, + 144 + ], + [ + 315, + 183 + ], + [ + 297, + 157 + ], + [ + 316, + 195 + ] + ], + [ + [ + 317, + 135 + ], + [ + 341, + 147 + ], + [ + 306, + 157 + ], + [ + 328, + 114 + ], + [ + 283, + 146 + ], + [ + 305, + 102 + ] + ], + [ + [ + 319, + 165 + ], + [ + 296, + 168 + ], + [ + 315, + 141 + ], + [ + 323, + 188 + ], + [ + 336, + 138 + ], + [ + 345, + 185 + ] + ] + ] + }, + { + "task_id": "0001029", + "id": "RH20T_cfg5_task_0118_user_0007_scene_0009_cfg_0005_104122063678#244", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the water tap'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001029/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 293 + ], + [ + 314, + 279 + ], + [ + 304, + 323 + ], + [ + 309, + 266 + ], + [ + 297, + 335 + ], + [ + 302, + 280 + ] + ], + [ + [ + 314, + 302 + ], + [ + 335, + 324 + ], + [ + 320, + 294 + ], + [ + 308, + 311 + ], + [ + 300, + 273 + ], + [ + 287, + 288 + ] + ], + [ + [ + 321, + 326 + ], + [ + 349, + 319 + ], + [ + 334, + 335 + ], + [ + 310, + 319 + ], + [ + 304, + 344 + ], + [ + 282, + 326 + ] + ], + [ + [ + 279, + 306 + ], + [ + 249, + 297 + ], + [ + 275, + 321 + ], + [ + 284, + 289 + ], + [ + 305, + 330 + ], + [ + 317, + 299 + ] + ] + ] + }, + { + "task_id": "0001030", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104122062295#125", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001030/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 208, + 137 + ], + [ + 201, + 168 + ], + [ + 206, + 113 + ], + [ + 211, + 165 + ], + [ + 214, + 80 + ], + [ + 220, + 130 + ] + ], + [ + [ + 183, + 185 + ], + [ + 172, + 208 + ], + [ + 184, + 218 + ], + [ + 182, + 149 + ], + [ + 192, + 197 + ], + [ + 191, + 132 + ] + ], + [ + [ + 195, + 176 + ], + [ + 202, + 139 + ], + [ + 150, + 171 + ], + [ + 237, + 180 + ], + [ + 146, + 205 + ], + [ + 229, + 212 + ] + ], + [ + [ + 227, + 181 + ], + [ + 217, + 146 + ], + [ + 190, + 184 + ], + [ + 259, + 179 + ], + [ + 199, + 224 + ], + [ + 269, + 214 + ] + ] + ] + }, + { + "task_id": "0001031", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104122062295#528", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001031/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 188, + 145 + ], + [ + 197, + 110 + ], + [ + 147, + 130 + ], + [ + 227, + 160 + ], + [ + 140, + 163 + ], + [ + 217, + 190 + ] + ], + [ + [ + 226, + 117 + ], + [ + 213, + 146 + ], + [ + 255, + 142 + ], + [ + 194, + 89 + ], + [ + 265, + 116 + ], + [ + 208, + 64 + ] + ], + [ + [ + 197, + 114 + ], + [ + 213, + 124 + ], + [ + 223, + 81 + ], + [ + 172, + 147 + ], + [ + 207, + 75 + ], + [ + 160, + 136 + ] + ], + [ + [ + 198, + 160 + ], + [ + 222, + 141 + ], + [ + 212, + 196 + ], + [ + 185, + 125 + ], + [ + 189, + 211 + ], + [ + 164, + 143 + ] + ] + ] + }, + { + "task_id": "0001037", + "id": "RH20T_cfg1_task_0004_user_0001_scene_0005_cfg_0001_039422060546#485", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001037/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 213, + 304 + ], + [ + 214, + 281 + ], + [ + 235, + 274 + ], + [ + 191, + 333 + ], + [ + 236, + 298 + ], + [ + 187, + 360 + ] + ], + [ + [ + 216, + 270 + ], + [ + 203, + 256 + ], + [ + 198, + 304 + ], + [ + 233, + 239 + ], + [ + 211, + 314 + ], + [ + 243, + 253 + ] + ], + [ + [ + 210, + 248 + ], + [ + 242, + 255 + ], + [ + 202, + 228 + ], + [ + 217, + 265 + ], + [ + 169, + 221 + ], + [ + 188, + 258 + ] + ], + [ + [ + 201, + 270 + ], + [ + 169, + 254 + ], + [ + 213, + 283 + ], + [ + 191, + 259 + ], + [ + 245, + 298 + ], + [ + 220, + 273 + ] + ] + ] + }, + { + "task_id": "0001038", + "id": "RH20T_cfg1_task_0065_user_0015_scene_0001_cfg_0001_819112071037#358", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001038/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 640, + 145 + ], + [ + 640, + 83 + ], + [ + 640, + 168 + ], + [ + 640, + 126 + ], + [ + 640, + 229 + ], + [ + 639, + 176 + ] + ], + [ + [ + 640, + 170 + ], + [ + 640, + 191 + ], + [ + 617, + 203 + ], + [ + 640, + 131 + ], + [ + 575, + 183 + ], + [ + 640, + 107 + ] + ], + [ + [ + 640, + 147 + ], + [ + 609, + 160 + ], + [ + 640, + 197 + ], + [ + 640, + 97 + ], + [ + 640, + 187 + ], + [ + 640, + 74 + ] + ], + [ + [ + 640, + 166 + ], + [ + 640, + 163 + ], + [ + 640, + 170 + ], + [ + 640, + 162 + ], + [ + 640, + 172 + ], + [ + 614, + 166 + ] + ] + ] + }, + { + "task_id": "0001040", + "id": "RH20T_cfg4_task_0067_user_0015_scene_0006_cfg_0004_038522062288#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button from top to bottom with obstacles'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001040/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 194, + 96 + ], + [ + 204, + 97 + ], + [ + 206, + 122 + ], + [ + 182, + 68 + ], + [ + 197, + 120 + ], + [ + 173, + 69 + ] + ], + [ + [ + 142, + 113 + ], + [ + 147, + 93 + ], + [ + 153, + 136 + ], + [ + 129, + 87 + ], + [ + 147, + 157 + ], + [ + 121, + 109 + ] + ], + [ + [ + 166, + 117 + ], + [ + 180, + 114 + ], + [ + 144, + 131 + ], + [ + 187, + 103 + ], + [ + 132, + 133 + ], + [ + 174, + 106 + ] + ], + [ + [ + 164, + 110 + ], + [ + 160, + 85 + ], + [ + 137, + 103 + ], + [ + 192, + 117 + ], + [ + 142, + 126 + ], + [ + 195, + 140 + ] + ] + ] + }, + { + "task_id": "0001041", + "id": "RH20T_cfg1_task_0001_user_0001_scene_0002_cfg_0001_035622060973#229", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001041/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 318, + 181 + ], + [ + 323, + 159 + ], + [ + 338, + 170 + ], + [ + 296, + 193 + ], + [ + 332, + 190 + ], + [ + 292, + 213 + ] + ], + [ + [ + 290, + 168 + ], + [ + 297, + 154 + ], + [ + 285, + 143 + ], + [ + 295, + 195 + ], + [ + 279, + 156 + ], + [ + 288, + 206 + ] + ], + [ + [ + 273, + 160 + ], + [ + 300, + 168 + ], + [ + 272, + 140 + ], + [ + 274, + 179 + ], + [ + 245, + 131 + ], + [ + 249, + 171 + ] + ], + [ + [ + 303, + 191 + ], + [ + 323, + 170 + ], + [ + 312, + 206 + ], + [ + 295, + 176 + ], + [ + 291, + 227 + ], + [ + 277, + 196 + ] + ] + ] + }, + { + "task_id": "0001042", + "id": "RH20T_cfg5_task_0001_user_0007_scene_0008_cfg_0005_104422070044#57", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001042/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 571, + 156 + ], + [ + 583, + 156 + ], + [ + 549, + 163 + ], + [ + 594, + 149 + ], + [ + 536, + 163 + ], + [ + 581, + 150 + ] + ], + [ + [ + 532, + 179 + ], + [ + 549, + 168 + ], + [ + 544, + 197 + ], + [ + 520, + 161 + ], + [ + 527, + 209 + ], + [ + 503, + 172 + ] + ], + [ + [ + 543, + 185 + ], + [ + 523, + 184 + ], + [ + 530, + 193 + ], + [ + 557, + 177 + ], + [ + 548, + 194 + ], + [ + 578, + 179 + ] + ], + [ + [ + 554, + 170 + ], + [ + 555, + 152 + ], + [ + 576, + 178 + ], + [ + 534, + 162 + ], + [ + 575, + 195 + ], + [ + 534, + 179 + ] + ] + ] + }, + { + "task_id": "0001043", + "id": "RH20T_cfg1_task_0009_user_0014_scene_0007_cfg_0001_038522063145#341", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001043/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 366, + 131 + ], + [ + 362, + 136 + ], + [ + 346, + 129 + ], + [ + 387, + 132 + ], + [ + 350, + 124 + ], + [ + 389, + 127 + ] + ], + [ + [ + 351, + 108 + ], + [ + 366, + 120 + ], + [ + 361, + 91 + ], + [ + 341, + 125 + ], + [ + 346, + 78 + ], + [ + 325, + 112 + ] + ], + [ + [ + 365, + 111 + ], + [ + 350, + 123 + ], + [ + 353, + 94 + ], + [ + 376, + 129 + ], + [ + 369, + 82 + ], + [ + 392, + 117 + ] + ], + [ + [ + 354, + 99 + ], + [ + 357, + 82 + ], + [ + 375, + 103 + ], + [ + 334, + 96 + ], + [ + 371, + 119 + ], + [ + 332, + 112 + ] + ] + ] + }, + { + "task_id": "0001045", + "id": "RH20T_cfg7_task_0029_user_0020_scene_0002_cfg_0007_104122061018#216", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001045/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 340, + 128 + ], + [ + 374, + 113 + ], + [ + 332, + 148 + ], + [ + 350, + 105 + ], + [ + 298, + 163 + ], + [ + 311, + 122 + ] + ], + [ + [ + 356, + 184 + ], + [ + 367, + 216 + ], + [ + 331, + 172 + ], + [ + 384, + 197 + ], + [ + 319, + 138 + ], + [ + 372, + 159 + ] + ], + [ + [ + 354, + 149 + ], + [ + 340, + 157 + ], + [ + 371, + 185 + ], + [ + 337, + 114 + ], + [ + 389, + 179 + ], + [ + 352, + 102 + ] + ], + [ + [ + 384, + 137 + ], + [ + 383, + 153 + ], + [ + 392, + 100 + ], + [ + 377, + 171 + ], + [ + 393, + 89 + ], + [ + 379, + 156 + ] + ] + ] + }, + { + "task_id": "0001047", + "id": "RH20T_cfg6_task_0006_user_0014_scene_0005_cfg_0006_104122061330#180", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001047/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 287, + 243 + ], + [ + 311, + 257 + ], + [ + 298, + 227 + ], + [ + 277, + 258 + ], + [ + 273, + 212 + ], + [ + 255, + 244 + ] + ], + [ + [ + 293, + 230 + ], + [ + 285, + 222 + ], + [ + 277, + 254 + ], + [ + 307, + 207 + ], + [ + 285, + 260 + ], + [ + 313, + 215 + ] + ], + [ + [ + 325, + 224 + ], + [ + 348, + 236 + ], + [ + 330, + 236 + ], + [ + 320, + 210 + ], + [ + 309, + 224 + ], + [ + 298, + 198 + ] + ], + [ + [ + 257, + 233 + ], + [ + 278, + 247 + ], + [ + 262, + 210 + ], + [ + 252, + 254 + ], + [ + 241, + 196 + ], + [ + 232, + 240 + ] + ] + ] + }, + { + "task_id": "0001049", + "id": "RH20T_cfg4_task_0003_user_0010_scene_0010_cfg_0004_104422070011#86", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001049/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 232, + 214 + ], + [ + 239, + 197 + ], + [ + 209, + 237 + ], + [ + 253, + 193 + ], + [ + 204, + 252 + ], + [ + 247, + 209 + ] + ], + [ + [ + 215, + 189 + ], + [ + 192, + 164 + ], + [ + 203, + 202 + ], + [ + 228, + 173 + ], + [ + 224, + 226 + ], + [ + 251, + 200 + ] + ], + [ + [ + 235, + 197 + ], + [ + 243, + 174 + ], + [ + 255, + 187 + ], + [ + 215, + 206 + ], + [ + 246, + 213 + ], + [ + 205, + 230 + ] + ], + [ + [ + 244, + 165 + ], + [ + 215, + 146 + ], + [ + 226, + 182 + ], + [ + 260, + 149 + ], + [ + 256, + 202 + ], + [ + 287, + 167 + ] + ] + ] + }, + { + "task_id": "0001052", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0006_cfg_0004_104122062823#110", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001052/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 264, + 28 + ], + [ + 265, + 0 + ], + [ + 298, + 41 + ], + [ + 232, + 16 + ], + [ + 296, + 75 + ], + [ + 233, + 49 + ] + ], + [ + [ + 241, + 0 + ], + [ + 279, + 0 + ], + [ + 260, + 6 + ], + [ + 218, + 0 + ], + [ + 226, + 12 + ], + [ + 179, + 0 + ] + ], + [ + [ + 273, + 43 + ], + [ + 285, + 36 + ], + [ + 309, + 37 + ], + [ + 240, + 50 + ], + [ + 297, + 45 + ], + [ + 224, + 58 + ] + ], + [ + [ + 308, + 9 + ], + [ + 292, + 44 + ], + [ + 310, + 0 + ], + [ + 306, + 25 + ], + [ + 327, + 0 + ], + [ + 320, + 0 + ] + ] + ] + }, + { + "task_id": "0001054", + "id": "RH20T_cfg6_task_0042_user_0014_scene_0004_cfg_0006_104122061330#110", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001054/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 282, + 190 + ], + [ + 310, + 199 + ], + [ + 268, + 216 + ], + [ + 295, + 165 + ], + [ + 242, + 206 + ], + [ + 269, + 156 + ] + ], + [ + [ + 276, + 195 + ], + [ + 272, + 204 + ], + [ + 286, + 225 + ], + [ + 267, + 166 + ], + [ + 291, + 216 + ], + [ + 270, + 153 + ] + ], + [ + [ + 264, + 167 + ], + [ + 257, + 149 + ], + [ + 250, + 192 + ], + [ + 278, + 144 + ], + [ + 257, + 208 + ], + [ + 283, + 161 + ] + ], + [ + [ + 283, + 169 + ], + [ + 292, + 148 + ], + [ + 277, + 146 + ], + [ + 288, + 194 + ], + [ + 270, + 165 + ], + [ + 280, + 212 + ] + ] + ] + }, + { + "task_id": "0001055", + "id": "RH20T_cfg6_task_0042_user_0014_scene_0004_cfg_0006_104122061330#192", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001055/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 226, + 173 + ], + [ + 229, + 178 + ], + [ + 227, + 204 + ], + [ + 224, + 142 + ], + [ + 222, + 201 + ], + [ + 220, + 134 + ] + ], + [ + [ + 257, + 184 + ], + [ + 288, + 183 + ], + [ + 256, + 205 + ], + [ + 259, + 160 + ], + [ + 226, + 206 + ], + [ + 226, + 160 + ] + ], + [ + [ + 249, + 167 + ], + [ + 274, + 175 + ], + [ + 263, + 157 + ], + [ + 233, + 178 + ], + [ + 240, + 150 + ], + [ + 208, + 170 + ] + ], + [ + [ + 259, + 168 + ], + [ + 257, + 145 + ], + [ + 242, + 188 + ], + [ + 276, + 151 + ], + [ + 245, + 210 + ], + [ + 277, + 172 + ] + ] + ] + }, + { + "task_id": "0001056", + "id": "RH20T_cfg6_task_0042_user_0014_scene_0004_cfg_0006_104122061330#277", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001056/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 170 + ], + [ + 269, + 187 + ], + [ + 286, + 149 + ], + [ + 262, + 190 + ], + [ + 290, + 135 + ], + [ + 267, + 175 + ] + ], + [ + [ + 278, + 167 + ], + [ + 268, + 161 + ], + [ + 300, + 149 + ], + [ + 255, + 185 + ], + [ + 311, + 155 + ], + [ + 264, + 193 + ] + ], + [ + [ + 266, + 189 + ], + [ + 264, + 168 + ], + [ + 249, + 208 + ], + [ + 281, + 171 + ], + [ + 251, + 228 + ], + [ + 282, + 191 + ] + ], + [ + [ + 264, + 204 + ], + [ + 261, + 233 + ], + [ + 255, + 203 + ], + [ + 275, + 205 + ], + [ + 258, + 176 + ], + [ + 278, + 175 + ] + ] + ] + }, + { + "task_id": "0001058", + "id": "RH20T_cfg6_task_0042_user_0014_scene_0004_cfg_0006_104122061330#492", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001058/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 259, + 165 + ], + [ + 251, + 139 + ], + [ + 270, + 174 + ], + [ + 250, + 157 + ], + [ + 277, + 200 + ], + [ + 257, + 181 + ] + ], + [ + [ + 255, + 190 + ], + [ + 253, + 161 + ], + [ + 235, + 187 + ], + [ + 277, + 192 + ], + [ + 237, + 214 + ], + [ + 279, + 221 + ] + ], + [ + [ + 270, + 196 + ], + [ + 253, + 180 + ], + [ + 284, + 199 + ], + [ + 255, + 193 + ], + [ + 301, + 215 + ], + [ + 274, + 211 + ] + ], + [ + [ + 257, + 187 + ], + [ + 256, + 164 + ], + [ + 241, + 204 + ], + [ + 273, + 171 + ], + [ + 243, + 225 + ], + [ + 273, + 192 + ] + ] + ] + }, + { + "task_id": "0001061", + "id": "RH20T_cfg3_task_0043_user_0010_scene_0006_cfg_0003_104422070011#172", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001061/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 379, + 225 + ], + [ + 391, + 253 + ], + [ + 349, + 238 + ], + [ + 409, + 212 + ], + [ + 337, + 209 + ], + [ + 397, + 183 + ] + ], + [ + [ + 331, + 245 + ], + [ + 310, + 259 + ], + [ + 311, + 223 + ], + [ + 352, + 268 + ], + [ + 330, + 211 + ], + [ + 370, + 253 + ] + ], + [ + [ + 347, + 235 + ], + [ + 362, + 227 + ], + [ + 337, + 264 + ], + [ + 356, + 208 + ], + [ + 324, + 269 + ], + [ + 342, + 216 + ] + ], + [ + [ + 331, + 243 + ], + [ + 340, + 277 + ], + [ + 333, + 237 + ], + [ + 330, + 248 + ], + [ + 324, + 203 + ], + [ + 322, + 218 + ] + ] + ] + }, + { + "task_id": "0001062", + "id": "RH20T_cfg6_task_0130_user_0014_scene_0004_cfg_0006_104122061602#239", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001062/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 466, + 140 + ], + [ + 491, + 134 + ], + [ + 463, + 150 + ], + [ + 469, + 130 + ], + [ + 440, + 155 + ], + [ + 443, + 136 + ] + ], + [ + [ + 447, + 145 + ], + [ + 458, + 139 + ], + [ + 459, + 166 + ], + [ + 436, + 124 + ], + [ + 449, + 171 + ], + [ + 427, + 131 + ] + ], + [ + [ + 426, + 155 + ], + [ + 412, + 158 + ], + [ + 412, + 169 + ], + [ + 442, + 140 + ], + [ + 425, + 166 + ], + [ + 456, + 137 + ] + ], + [ + [ + 455, + 132 + ], + [ + 429, + 129 + ], + [ + 459, + 126 + ], + [ + 450, + 138 + ], + [ + 486, + 129 + ], + [ + 475, + 141 + ] + ] + ] + }, + { + "task_id": "0001067", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0010_cfg_0001_038522063145#565", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001067/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 119 + ], + [ + 399, + 109 + ], + [ + 391, + 139 + ], + [ + 395, + 98 + ], + [ + 384, + 150 + ], + [ + 388, + 108 + ] + ], + [ + [ + 380, + 112 + ], + [ + 386, + 91 + ], + [ + 399, + 112 + ], + [ + 360, + 113 + ], + [ + 393, + 131 + ], + [ + 355, + 133 + ] + ], + [ + [ + 396, + 121 + ], + [ + 401, + 100 + ], + [ + 410, + 115 + ], + [ + 384, + 126 + ], + [ + 405, + 138 + ], + [ + 379, + 147 + ] + ], + [ + [ + 391, + 116 + ], + [ + 393, + 98 + ], + [ + 412, + 126 + ], + [ + 371, + 106 + ], + [ + 409, + 142 + ], + [ + 370, + 123 + ] + ] + ] + }, + { + "task_id": "0001068", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0009_cfg_0004_036422060909#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001068/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 271, + 63 + ], + [ + 269, + 54 + ], + [ + 262, + 38 + ], + [ + 280, + 88 + ], + [ + 264, + 47 + ], + [ + 281, + 94 + ] + ], + [ + [ + 287, + 39 + ], + [ + 287, + 17 + ], + [ + 298, + 23 + ], + [ + 274, + 56 + ], + [ + 298, + 44 + ], + [ + 275, + 77 + ] + ], + [ + [ + 308, + 28 + ], + [ + 304, + 0 + ], + [ + 292, + 26 + ], + [ + 325, + 29 + ], + [ + 297, + 52 + ], + [ + 329, + 58 + ] + ], + [ + [ + 275, + 13 + ], + [ + 253, + 0 + ], + [ + 291, + 0 + ], + [ + 261, + 38 + ], + [ + 313, + 4 + ], + [ + 282, + 53 + ] + ] + ] + }, + { + "task_id": "0001070", + "id": "RH20T_cfg5_task_0059_user_0007_scene_0003_cfg_0005_105422061350#665", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001070/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 301, + 141 + ], + [ + 304, + 123 + ], + [ + 323, + 128 + ], + [ + 277, + 155 + ], + [ + 319, + 145 + ], + [ + 275, + 171 + ] + ], + [ + [ + 328, + 145 + ], + [ + 302, + 134 + ], + [ + 343, + 130 + ], + [ + 314, + 158 + ], + [ + 369, + 141 + ], + [ + 339, + 167 + ] + ], + [ + [ + 335, + 134 + ], + [ + 322, + 151 + ], + [ + 311, + 136 + ], + [ + 361, + 133 + ], + [ + 323, + 121 + ], + [ + 373, + 117 + ] + ], + [ + [ + 293, + 152 + ], + [ + 302, + 134 + ], + [ + 318, + 167 + ], + [ + 267, + 137 + ], + [ + 309, + 187 + ], + [ + 256, + 157 + ] + ] + ] + }, + { + "task_id": "0001073", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0004_cfg_0004_104122062295#419", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001073/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 248, + 40 + ], + [ + 249, + 81 + ], + [ + 232, + 24 + ], + [ + 265, + 58 + ], + [ + 230, + 0 + ], + [ + 265, + 9 + ] + ], + [ + [ + 265, + 59 + ], + [ + 268, + 15 + ], + [ + 219, + 59 + ], + [ + 310, + 60 + ], + [ + 220, + 99 + ], + [ + 305, + 98 + ] + ], + [ + [ + 304, + 102 + ], + [ + 313, + 58 + ], + [ + 332, + 115 + ], + [ + 271, + 85 + ], + [ + 324, + 157 + ], + [ + 260, + 134 + ] + ], + [ + [ + 258, + 29 + ], + [ + 274, + 66 + ], + [ + 262, + 65 + ], + [ + 252, + 0 + ], + [ + 248, + 33 + ], + [ + 237, + 0 + ] + ] + ] + }, + { + "task_id": "0001077", + "id": "RH20T_cfg1_task_0018_user_0008_scene_0007_cfg_0001_750612070851#164", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001077/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 428, + 247 + ], + [ + 399, + 263 + ], + [ + 419, + 226 + ], + [ + 437, + 266 + ], + [ + 450, + 210 + ], + [ + 466, + 251 + ] + ], + [ + [ + 438, + 253 + ], + [ + 448, + 234 + ], + [ + 445, + 285 + ], + [ + 431, + 224 + ], + [ + 435, + 300 + ], + [ + 422, + 241 + ] + ], + [ + [ + 466, + 255 + ], + [ + 465, + 247 + ], + [ + 431, + 246 + ], + [ + 502, + 264 + ], + [ + 434, + 254 + ], + [ + 501, + 270 + ] + ], + [ + [ + 458, + 230 + ], + [ + 461, + 252 + ], + [ + 443, + 255 + ], + [ + 475, + 203 + ], + [ + 442, + 235 + ], + [ + 472, + 185 + ] + ] + ] + }, + { + "task_id": "0001078", + "id": "RH20T_cfg1_task_0120_user_0014_scene_0002_cfg_0001_750612070853#112", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001078/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 394, + 191 + ], + [ + 417, + 154 + ], + [ + 344, + 176 + ], + [ + 443, + 205 + ], + [ + 328, + 209 + ], + [ + 420, + 234 + ] + ], + [ + [ + 390, + 189 + ], + [ + 401, + 159 + ], + [ + 360, + 209 + ], + [ + 425, + 167 + ], + [ + 346, + 242 + ], + [ + 413, + 202 + ] + ], + [ + [ + 357, + 207 + ], + [ + 391, + 173 + ], + [ + 315, + 178 + ], + [ + 403, + 239 + ], + [ + 277, + 212 + ], + [ + 365, + 281 + ] + ], + [ + [ + 407, + 192 + ], + [ + 410, + 158 + ], + [ + 403, + 222 + ], + [ + 412, + 157 + ], + [ + 400, + 260 + ], + [ + 409, + 197 + ] + ] + ] + }, + { + "task_id": "0001081", + "id": "RH20T_cfg5_task_0075_user_0007_scene_0004_cfg_0005_105422061350#634", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Finish setting up the starting position of a chessboard that is almost arranged'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001081/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 125, + 116 + ], + [ + 109, + 103 + ], + [ + 117, + 142 + ], + [ + 134, + 89 + ], + [ + 132, + 156 + ], + [ + 151, + 102 + ] + ], + [ + [ + 144, + 132 + ], + [ + 142, + 160 + ], + [ + 167, + 136 + ], + [ + 123, + 128 + ], + [ + 169, + 106 + ], + [ + 125, + 99 + ] + ], + [ + [ + 161, + 134 + ], + [ + 187, + 125 + ], + [ + 161, + 148 + ], + [ + 161, + 118 + ], + [ + 135, + 158 + ], + [ + 132, + 127 + ] + ], + [ + [ + 138, + 106 + ], + [ + 135, + 86 + ], + [ + 164, + 96 + ], + [ + 111, + 117 + ], + [ + 166, + 114 + ], + [ + 115, + 135 + ] + ] + ] + }, + { + "task_id": "0001082", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0009_cfg_0007_104122060811#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001082/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 119 + ], + [ + 357, + 122 + ], + [ + 339, + 131 + ], + [ + 357, + 108 + ], + [ + 331, + 128 + ], + [ + 348, + 105 + ] + ], + [ + [ + 355, + 132 + ], + [ + 369, + 128 + ], + [ + 357, + 146 + ], + [ + 353, + 119 + ], + [ + 344, + 150 + ], + [ + 340, + 123 + ] + ], + [ + [ + 356, + 126 + ], + [ + 357, + 115 + ], + [ + 364, + 119 + ], + [ + 349, + 133 + ], + [ + 363, + 131 + ], + [ + 348, + 145 + ] + ], + [ + [ + 353, + 135 + ], + [ + 362, + 138 + ], + [ + 343, + 144 + ], + [ + 363, + 127 + ], + [ + 333, + 141 + ], + [ + 352, + 124 + ] + ] + ] + }, + { + "task_id": "0001086", + "id": "RH20T_cfg2_task_0210_user_0007_scene_0005_cfg_0002_104422070044#110", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Chop the onions'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001086/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 281, + 180 + ], + [ + 297, + 177 + ], + [ + 286, + 193 + ], + [ + 275, + 168 + ], + [ + 269, + 196 + ], + [ + 259, + 171 + ] + ], + [ + [ + 272, + 200 + ], + [ + 270, + 217 + ], + [ + 275, + 194 + ], + [ + 270, + 207 + ], + [ + 277, + 178 + ], + [ + 272, + 192 + ] + ], + [ + [ + 279, + 183 + ], + [ + 276, + 172 + ], + [ + 289, + 193 + ], + [ + 270, + 174 + ], + [ + 292, + 203 + ], + [ + 273, + 184 + ] + ], + [ + [ + 289, + 164 + ], + [ + 297, + 153 + ], + [ + 277, + 152 + ], + [ + 300, + 176 + ], + [ + 267, + 165 + ], + [ + 292, + 188 + ] + ] + ] + }, + { + "task_id": "0001087", + "id": "RH20T_cfg1_task_0079_user_0015_scene_0003_cfg_0001_039422060546#103", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open the microwave door'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001087/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 275, + 33 + ], + [ + 246, + 21 + ], + [ + 285, + 35 + ], + [ + 263, + 32 + ], + [ + 312, + 47 + ], + [ + 294, + 46 + ] + ], + [ + [ + 264, + 24 + ], + [ + 261, + 0 + ], + [ + 247, + 45 + ], + [ + 279, + 5 + ], + [ + 250, + 71 + ], + [ + 280, + 32 + ] + ], + [ + [ + 252, + 25 + ], + [ + 241, + 24 + ], + [ + 241, + 0 + ], + [ + 262, + 56 + ], + [ + 253, + 0 + ], + [ + 275, + 60 + ] + ], + [ + [ + 224, + 7 + ], + [ + 194, + 2 + ], + [ + 243, + 13 + ], + [ + 202, + 0 + ], + [ + 273, + 18 + ], + [ + 235, + 5 + ] + ] + ] + }, + { + "task_id": "0001091", + "id": "RH20T_cfg1_task_0034_user_0001_scene_0006_cfg_0001_038522062547#548", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001091/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 528, + 146 + ], + [ + 518, + 123 + ], + [ + 521, + 146 + ], + [ + 533, + 147 + ], + [ + 531, + 171 + ], + [ + 543, + 170 + ] + ], + [ + [ + 529, + 122 + ], + [ + 545, + 140 + ], + [ + 532, + 105 + ], + [ + 527, + 138 + ], + [ + 517, + 89 + ], + [ + 513, + 122 + ] + ], + [ + [ + 564, + 130 + ], + [ + 546, + 124 + ], + [ + 577, + 105 + ], + [ + 551, + 154 + ], + [ + 593, + 112 + ], + [ + 567, + 159 + ] + ], + [ + [ + 546, + 144 + ], + [ + 555, + 132 + ], + [ + 550, + 168 + ], + [ + 543, + 121 + ], + [ + 541, + 178 + ], + [ + 535, + 133 + ] + ] + ] + }, + { + "task_id": "0001092", + "id": "RH20T_cfg1_task_0034_user_0001_scene_0006_cfg_0001_038522062547#839", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001092/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 450, + 97 + ], + [ + 446, + 101 + ], + [ + 453, + 120 + ], + [ + 447, + 72 + ], + [ + 457, + 115 + ], + [ + 451, + 69 + ] + ], + [ + [ + 467, + 89 + ], + [ + 472, + 74 + ], + [ + 473, + 110 + ], + [ + 461, + 70 + ], + [ + 468, + 123 + ], + [ + 457, + 84 + ] + ], + [ + [ + 474, + 113 + ], + [ + 459, + 118 + ], + [ + 454, + 116 + ], + [ + 496, + 110 + ], + [ + 468, + 111 + ], + [ + 510, + 106 + ] + ], + [ + [ + 459, + 81 + ], + [ + 474, + 60 + ], + [ + 473, + 94 + ], + [ + 445, + 68 + ], + [ + 458, + 114 + ], + [ + 431, + 88 + ] + ] + ] + }, + { + "task_id": "0001094", + "id": "RH20T_cfg1_task_0034_user_0001_scene_0006_cfg_0001_038522062547#1139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001094/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 494, + 75 + ], + [ + 501, + 51 + ], + [ + 496, + 83 + ], + [ + 492, + 68 + ], + [ + 489, + 107 + ], + [ + 486, + 90 + ] + ], + [ + [ + 452, + 110 + ], + [ + 474, + 97 + ], + [ + 461, + 128 + ], + [ + 444, + 92 + ], + [ + 440, + 140 + ], + [ + 424, + 105 + ] + ], + [ + [ + 469, + 74 + ], + [ + 474, + 59 + ], + [ + 479, + 93 + ], + [ + 460, + 57 + ], + [ + 474, + 107 + ], + [ + 456, + 71 + ] + ], + [ + [ + 470, + 59 + ], + [ + 488, + 40 + ], + [ + 452, + 46 + ], + [ + 488, + 73 + ], + [ + 436, + 65 + ], + [ + 471, + 91 + ] + ] + ] + }, + { + "task_id": "0001096", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0001_cfg_0006_104122060811#189", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001096/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 136 + ], + [ + 340, + 134 + ], + [ + 343, + 148 + ], + [ + 325, + 124 + ], + [ + 336, + 150 + ], + [ + 318, + 126 + ] + ], + [ + [ + 323, + 132 + ], + [ + 324, + 125 + ], + [ + 337, + 128 + ], + [ + 309, + 136 + ], + [ + 335, + 135 + ], + [ + 308, + 143 + ] + ], + [ + [ + 323, + 145 + ], + [ + 316, + 133 + ], + [ + 329, + 145 + ], + [ + 318, + 145 + ], + [ + 336, + 157 + ], + [ + 326, + 158 + ] + ], + [ + [ + 340, + 131 + ], + [ + 337, + 145 + ], + [ + 355, + 134 + ], + [ + 325, + 128 + ], + [ + 358, + 120 + ], + [ + 328, + 114 + ] + ] + ] + }, + { + "task_id": "0001097", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0001_cfg_0006_104122060811#468", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001097/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 105 + ], + [ + 350, + 96 + ], + [ + 360, + 101 + ], + [ + 334, + 110 + ], + [ + 357, + 109 + ], + [ + 332, + 118 + ] + ], + [ + [ + 345, + 115 + ], + [ + 347, + 130 + ], + [ + 357, + 112 + ], + [ + 333, + 118 + ], + [ + 355, + 97 + ], + [ + 331, + 103 + ] + ], + [ + [ + 350, + 96 + ], + [ + 364, + 92 + ], + [ + 347, + 82 + ], + [ + 353, + 110 + ], + [ + 332, + 85 + ], + [ + 339, + 113 + ] + ], + [ + [ + 342, + 102 + ], + [ + 347, + 112 + ], + [ + 329, + 101 + ], + [ + 354, + 104 + ], + [ + 323, + 91 + ], + [ + 349, + 94 + ] + ] + ] + }, + { + "task_id": "0001098", + "id": "RH20T_cfg1_task_0111_user_0010_scene_0006_cfg_0001_750612070853#126", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object with the corresponding shape into the corresponding hole'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001098/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 249, + 360 + ], + [ + 252, + 317 + ], + [ + 282, + 360 + ], + [ + 221, + 360 + ], + [ + 279, + 360 + ], + [ + 218, + 360 + ] + ], + [ + [ + 281, + 359 + ], + [ + 276, + 306 + ], + [ + 284, + 345 + ], + [ + 277, + 360 + ], + [ + 290, + 360 + ], + [ + 282, + 360 + ] + ], + [ + [ + 259, + 331 + ], + [ + 250, + 296 + ], + [ + 240, + 295 + ], + [ + 282, + 360 + ], + [ + 249, + 325 + ], + [ + 289, + 360 + ] + ], + [ + [ + 278, + 331 + ], + [ + 231, + 360 + ], + [ + 253, + 298 + ], + [ + 301, + 360 + ], + [ + 302, + 265 + ], + [ + 343, + 330 + ] + ] + ] + }, + { + "task_id": "0001099", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0008_cfg_0005_036422060215#427", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001099/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 216, + 129 + ], + [ + 217, + 111 + ], + [ + 237, + 127 + ], + [ + 195, + 131 + ], + [ + 236, + 143 + ], + [ + 195, + 147 + ] + ], + [ + [ + 234, + 142 + ], + [ + 216, + 134 + ], + [ + 224, + 155 + ], + [ + 244, + 131 + ], + [ + 243, + 164 + ], + [ + 261, + 139 + ] + ], + [ + [ + 205, + 128 + ], + [ + 201, + 138 + ], + [ + 194, + 111 + ], + [ + 215, + 145 + ], + [ + 197, + 102 + ], + [ + 218, + 135 + ] + ], + [ + [ + 201, + 131 + ], + [ + 214, + 123 + ], + [ + 184, + 118 + ], + [ + 216, + 144 + ], + [ + 172, + 127 + ], + [ + 203, + 151 + ] + ] + ] + }, + { + "task_id": "0001106", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0005_cfg_0001_750612070853#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001106/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 281, + 163 + ], + [ + 253, + 178 + ], + [ + 253, + 139 + ], + [ + 311, + 187 + ], + [ + 279, + 127 + ], + [ + 334, + 172 + ] + ], + [ + [ + 259, + 138 + ], + [ + 257, + 100 + ], + [ + 232, + 124 + ], + [ + 290, + 154 + ], + [ + 235, + 158 + ], + [ + 290, + 189 + ] + ], + [ + [ + 265, + 130 + ], + [ + 235, + 98 + ], + [ + 290, + 118 + ], + [ + 244, + 141 + ], + [ + 319, + 151 + ], + [ + 271, + 169 + ] + ], + [ + [ + 233, + 175 + ], + [ + 250, + 141 + ], + [ + 206, + 150 + ], + [ + 256, + 197 + ], + [ + 185, + 187 + ], + [ + 240, + 231 + ] + ] + ] + }, + { + "task_id": "0001107", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0005_cfg_0001_750612070853#516", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001107/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 249, + 199 + ], + [ + 275, + 208 + ], + [ + 282, + 208 + ], + [ + 220, + 191 + ], + [ + 254, + 197 + ], + [ + 190, + 180 + ] + ], + [ + [ + 234, + 233 + ], + [ + 233, + 282 + ], + [ + 210, + 228 + ], + [ + 254, + 238 + ], + [ + 212, + 177 + ], + [ + 254, + 196 + ] + ], + [ + [ + 231, + 192 + ], + [ + 229, + 154 + ], + [ + 211, + 170 + ], + [ + 254, + 216 + ], + [ + 214, + 203 + ], + [ + 255, + 250 + ] + ], + [ + [ + 181, + 178 + ], + [ + 224, + 157 + ], + [ + 199, + 200 + ], + [ + 167, + 160 + ], + [ + 150, + 225 + ], + [ + 123, + 180 + ] + ] + ] + }, + { + "task_id": "0001108", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_037522062165#396", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001108/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 312, + 245 + ], + [ + 337, + 256 + ], + [ + 300, + 263 + ], + [ + 324, + 228 + ], + [ + 275, + 251 + ], + [ + 301, + 217 + ] + ], + [ + [ + 294, + 213 + ], + [ + 292, + 205 + ], + [ + 319, + 205 + ], + [ + 269, + 221 + ], + [ + 323, + 213 + ], + [ + 269, + 231 + ] + ], + [ + [ + 285, + 240 + ], + [ + 285, + 214 + ], + [ + 299, + 238 + ], + [ + 270, + 243 + ], + [ + 298, + 263 + ], + [ + 269, + 271 + ] + ], + [ + [ + 307, + 230 + ], + [ + 309, + 216 + ], + [ + 319, + 210 + ], + [ + 295, + 252 + ], + [ + 317, + 223 + ], + [ + 294, + 265 + ] + ] + ] + }, + { + "task_id": "0001110", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_037522062165#569", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001110/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 256, + 236 + ], + [ + 281, + 224 + ], + [ + 242, + 218 + ], + [ + 270, + 253 + ], + [ + 217, + 232 + ], + [ + 246, + 265 + ] + ], + [ + [ + 270, + 218 + ], + [ + 269, + 203 + ], + [ + 287, + 200 + ], + [ + 252, + 236 + ], + [ + 287, + 214 + ], + [ + 254, + 250 + ] + ], + [ + [ + 260, + 207 + ], + [ + 258, + 232 + ], + [ + 275, + 201 + ], + [ + 247, + 212 + ], + [ + 276, + 176 + ], + [ + 248, + 189 + ] + ], + [ + [ + 257, + 226 + ], + [ + 282, + 237 + ], + [ + 267, + 203 + ], + [ + 246, + 251 + ], + [ + 242, + 192 + ], + [ + 219, + 240 + ] + ] + ] + }, + { + "task_id": "0001112", + "id": "RH20T_cfg5_task_0015_user_0007_scene_0006_cfg_0005_105422061350#270", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001112/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 202, + 31 + ], + [ + 222, + 57 + ], + [ + 177, + 50 + ], + [ + 227, + 13 + ], + [ + 157, + 24 + ], + [ + 207, + 0 + ] + ], + [ + [ + 227, + 46 + ], + [ + 226, + 23 + ], + [ + 257, + 44 + ], + [ + 196, + 48 + ], + [ + 256, + 64 + ], + [ + 199, + 68 + ] + ], + [ + [ + 236, + 37 + ], + [ + 239, + 57 + ], + [ + 248, + 10 + ], + [ + 226, + 62 + ], + [ + 245, + 0 + ], + [ + 224, + 44 + ] + ], + [ + [ + 204, + 5 + ], + [ + 205, + 28 + ], + [ + 236, + 8 + ], + [ + 172, + 1 + ], + [ + 234, + 0 + ], + [ + 172, + 0 + ] + ] + ] + }, + { + "task_id": "0001114", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122061330#225", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001114/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 199 + ], + [ + 319, + 210 + ], + [ + 287, + 216 + ], + [ + 331, + 182 + ], + [ + 279, + 205 + ], + [ + 321, + 173 + ] + ], + [ + [ + 310, + 223 + ], + [ + 314, + 206 + ], + [ + 298, + 243 + ], + [ + 321, + 205 + ], + [ + 295, + 259 + ], + [ + 318, + 221 + ] + ], + [ + [ + 303, + 212 + ], + [ + 327, + 223 + ], + [ + 307, + 190 + ], + [ + 300, + 232 + ], + [ + 283, + 179 + ], + [ + 278, + 221 + ] + ], + [ + [ + 280, + 230 + ], + [ + 258, + 215 + ], + [ + 288, + 217 + ], + [ + 274, + 243 + ], + [ + 312, + 233 + ], + [ + 296, + 257 + ] + ] + ] + }, + { + "task_id": "0001115", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122061330#297", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001115/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 301, + 224 + ], + [ + 296, + 216 + ], + [ + 323, + 238 + ], + [ + 281, + 210 + ], + [ + 327, + 244 + ], + [ + 286, + 217 + ] + ], + [ + [ + 320, + 229 + ], + [ + 320, + 211 + ], + [ + 307, + 249 + ], + [ + 332, + 209 + ], + [ + 307, + 265 + ], + [ + 331, + 225 + ] + ], + [ + [ + 287, + 223 + ], + [ + 274, + 242 + ], + [ + 282, + 240 + ], + [ + 291, + 208 + ], + [ + 297, + 219 + ], + [ + 305, + 188 + ] + ], + [ + [ + 323, + 233 + ], + [ + 301, + 218 + ], + [ + 311, + 239 + ], + [ + 337, + 227 + ], + [ + 331, + 252 + ], + [ + 360, + 242 + ] + ] + ] + }, + { + "task_id": "0001120", + "id": "RH20T_cfg1_task_0023_user_0014_scene_0002_cfg_0001_038522063145#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001120/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 400, + 139 + ], + [ + 416, + 157 + ], + [ + 417, + 132 + ], + [ + 383, + 146 + ], + [ + 402, + 115 + ], + [ + 368, + 128 + ] + ], + [ + [ + 390, + 130 + ], + [ + 386, + 138 + ], + [ + 412, + 122 + ], + [ + 368, + 138 + ], + [ + 415, + 116 + ], + [ + 373, + 132 + ] + ], + [ + [ + 389, + 141 + ], + [ + 395, + 123 + ], + [ + 409, + 152 + ], + [ + 370, + 131 + ], + [ + 403, + 169 + ], + [ + 365, + 148 + ] + ], + [ + [ + 394, + 145 + ], + [ + 380, + 144 + ], + [ + 407, + 128 + ], + [ + 381, + 161 + ], + [ + 420, + 130 + ], + [ + 395, + 162 + ] + ] + ] + }, + { + "task_id": "0001121", + "id": "RH20T_cfg1_task_0018_user_0008_scene_0007_cfg_0001_819112070044#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001121/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 321, + 210 + ], + [ + 313, + 231 + ], + [ + 307, + 205 + ], + [ + 335, + 217 + ], + [ + 315, + 185 + ], + [ + 343, + 196 + ] + ], + [ + [ + 316, + 236 + ], + [ + 318, + 221 + ], + [ + 307, + 222 + ], + [ + 327, + 251 + ], + [ + 306, + 237 + ], + [ + 325, + 266 + ] + ], + [ + [ + 319, + 244 + ], + [ + 325, + 235 + ], + [ + 331, + 261 + ], + [ + 306, + 226 + ], + [ + 325, + 271 + ], + [ + 299, + 234 + ] + ], + [ + [ + 306, + 264 + ], + [ + 293, + 275 + ], + [ + 298, + 244 + ], + [ + 315, + 283 + ], + [ + 310, + 235 + ], + [ + 327, + 272 + ] + ] + ] + }, + { + "task_id": "0001123", + "id": "RH20T_cfg5_task_0013_user_0007_scene_0001_cfg_0005_105422061350#260", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001123/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 143, + 174 + ], + [ + 136, + 157 + ], + [ + 159, + 149 + ], + [ + 127, + 200 + ], + [ + 164, + 164 + ], + [ + 135, + 214 + ] + ], + [ + [ + 164, + 182 + ], + [ + 169, + 190 + ], + [ + 179, + 156 + ], + [ + 148, + 209 + ], + [ + 174, + 145 + ], + [ + 141, + 201 + ] + ], + [ + [ + 143, + 157 + ], + [ + 173, + 147 + ], + [ + 162, + 170 + ], + [ + 122, + 143 + ], + [ + 135, + 179 + ], + [ + 91, + 154 + ] + ], + [ + [ + 155, + 190 + ], + [ + 126, + 171 + ], + [ + 140, + 211 + ], + [ + 170, + 167 + ], + [ + 168, + 228 + ], + [ + 199, + 187 + ] + ] + ] + }, + { + "task_id": "0001124", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122060811#219", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001124/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 396, + 150 + ], + [ + 409, + 141 + ], + [ + 403, + 154 + ], + [ + 389, + 146 + ], + [ + 391, + 162 + ], + [ + 376, + 155 + ] + ], + [ + [ + 397, + 166 + ], + [ + 400, + 181 + ], + [ + 405, + 167 + ], + [ + 389, + 165 + ], + [ + 402, + 153 + ], + [ + 386, + 150 + ] + ], + [ + [ + 411, + 152 + ], + [ + 409, + 138 + ], + [ + 403, + 158 + ], + [ + 419, + 145 + ], + [ + 404, + 172 + ], + [ + 421, + 159 + ] + ], + [ + [ + 386, + 158 + ], + [ + 381, + 144 + ], + [ + 376, + 158 + ], + [ + 396, + 159 + ], + [ + 382, + 173 + ], + [ + 401, + 173 + ] + ] + ] + }, + { + "task_id": "0001125", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122060811#339", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001125/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 403, + 134 + ], + [ + 397, + 120 + ], + [ + 413, + 128 + ], + [ + 394, + 140 + ], + [ + 420, + 142 + ], + [ + 401, + 153 + ] + ], + [ + [ + 412, + 152 + ], + [ + 412, + 140 + ], + [ + 397, + 152 + ], + [ + 428, + 152 + ], + [ + 398, + 165 + ], + [ + 429, + 164 + ] + ], + [ + [ + 397, + 137 + ], + [ + 405, + 125 + ], + [ + 392, + 140 + ], + [ + 403, + 135 + ], + [ + 384, + 152 + ], + [ + 395, + 146 + ] + ], + [ + [ + 395, + 148 + ], + [ + 409, + 140 + ], + [ + 398, + 153 + ], + [ + 392, + 142 + ], + [ + 385, + 160 + ], + [ + 378, + 150 + ] + ] + ] + }, + { + "task_id": "0001126", + "id": "RH20T_cfg4_task_0068_user_0007_scene_0005_cfg_0004_104122062823#172", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button horizontally with obstacles'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001126/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 493, + 125 + ], + [ + 492, + 123 + ], + [ + 498, + 158 + ], + [ + 489, + 91 + ], + [ + 499, + 157 + ], + [ + 491, + 95 + ] + ], + [ + [ + 527, + 86 + ], + [ + 544, + 82 + ], + [ + 490, + 99 + ], + [ + 564, + 74 + ], + [ + 479, + 102 + ], + [ + 547, + 79 + ] + ], + [ + [ + 522, + 127 + ], + [ + 530, + 97 + ], + [ + 563, + 135 + ], + [ + 483, + 121 + ], + [ + 553, + 162 + ], + [ + 478, + 148 + ] + ], + [ + [ + 531, + 125 + ], + [ + 522, + 151 + ], + [ + 564, + 140 + ], + [ + 498, + 110 + ], + [ + 571, + 115 + ], + [ + 507, + 86 + ] + ] + ] + }, + { + "task_id": "0001128", + "id": "RH20T_cfg2_task_0071_user_0014_scene_0003_cfg_0002_105422061350#178", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001128/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 277, + 65 + ], + [ + 270, + 39 + ], + [ + 263, + 70 + ], + [ + 289, + 59 + ], + [ + 270, + 96 + ], + [ + 295, + 83 + ] + ], + [ + [ + 262, + 83 + ], + [ + 264, + 66 + ], + [ + 277, + 71 + ], + [ + 246, + 96 + ], + [ + 275, + 87 + ], + [ + 244, + 112 + ] + ], + [ + [ + 285, + 70 + ], + [ + 272, + 56 + ], + [ + 273, + 90 + ], + [ + 298, + 49 + ], + [ + 286, + 101 + ], + [ + 310, + 62 + ] + ], + [ + [ + 276, + 95 + ], + [ + 290, + 80 + ], + [ + 267, + 99 + ], + [ + 285, + 90 + ], + [ + 252, + 114 + ], + [ + 269, + 105 + ] + ] + ] + }, + { + "task_id": "0001133", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0008_cfg_0004_038522062288#91", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001133/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 219, + 164 + ], + [ + 193, + 177 + ], + [ + 212, + 160 + ], + [ + 226, + 170 + ], + [ + 236, + 148 + ], + [ + 252, + 157 + ] + ], + [ + [ + 205, + 173 + ], + [ + 179, + 173 + ], + [ + 215, + 188 + ], + [ + 193, + 156 + ], + [ + 240, + 188 + ], + [ + 221, + 156 + ] + ], + [ + [ + 189, + 162 + ], + [ + 213, + 167 + ], + [ + 188, + 138 + ], + [ + 190, + 187 + ], + [ + 162, + 131 + ], + [ + 163, + 182 + ] + ], + [ + [ + 207, + 163 + ], + [ + 204, + 143 + ], + [ + 179, + 173 + ], + [ + 234, + 154 + ], + [ + 184, + 191 + ], + [ + 236, + 172 + ] + ] + ] + }, + { + "task_id": "0001138", + "id": "RH20T_cfg5_task_0004_user_0014_scene_0007_cfg_0005_105422061350#229", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001138/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 219, + 63 + ], + [ + 234, + 46 + ], + [ + 205, + 66 + ], + [ + 231, + 61 + ], + [ + 189, + 83 + ], + [ + 216, + 77 + ] + ], + [ + [ + 199, + 54 + ], + [ + 207, + 72 + ], + [ + 178, + 45 + ], + [ + 221, + 63 + ], + [ + 168, + 25 + ], + [ + 213, + 43 + ] + ], + [ + [ + 220, + 86 + ], + [ + 209, + 81 + ], + [ + 222, + 113 + ], + [ + 218, + 58 + ], + [ + 233, + 120 + ], + [ + 230, + 63 + ] + ], + [ + [ + 200, + 61 + ], + [ + 200, + 43 + ], + [ + 207, + 42 + ], + [ + 193, + 82 + ], + [ + 207, + 59 + ], + [ + 193, + 98 + ] + ] + ] + }, + { + "task_id": "0001140", + "id": "RH20T_cfg6_task_0130_user_0014_scene_0004_cfg_0006_104122061018#343", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001140/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 297, + 116 + ], + [ + 311, + 92 + ], + [ + 320, + 134 + ], + [ + 272, + 96 + ], + [ + 306, + 159 + ], + [ + 256, + 121 + ] + ], + [ + [ + 291, + 83 + ], + [ + 258, + 80 + ], + [ + 296, + 50 + ], + [ + 288, + 113 + ], + [ + 329, + 54 + ], + [ + 319, + 116 + ] + ], + [ + [ + 305, + 81 + ], + [ + 292, + 51 + ], + [ + 296, + 85 + ], + [ + 314, + 76 + ], + [ + 308, + 112 + ], + [ + 327, + 107 + ] + ], + [ + [ + 292, + 113 + ], + [ + 273, + 87 + ], + [ + 318, + 93 + ], + [ + 267, + 132 + ], + [ + 337, + 118 + ], + [ + 287, + 158 + ] + ] + ] + }, + { + "task_id": "0001142", + "id": "RH20T_cfg4_task_0065_user_0015_scene_0012_cfg_0004_104422070011#519", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001142/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 382, + 159 + ], + [ + 362, + 172 + ], + [ + 384, + 142 + ], + [ + 380, + 174 + ], + [ + 404, + 129 + ], + [ + 399, + 162 + ] + ], + [ + [ + 385, + 143 + ], + [ + 395, + 129 + ], + [ + 365, + 148 + ], + [ + 404, + 137 + ], + [ + 356, + 160 + ], + [ + 394, + 149 + ] + ], + [ + [ + 361, + 141 + ], + [ + 378, + 140 + ], + [ + 376, + 128 + ], + [ + 347, + 152 + ], + [ + 358, + 128 + ], + [ + 329, + 153 + ] + ], + [ + [ + 404, + 135 + ], + [ + 416, + 153 + ], + [ + 419, + 140 + ], + [ + 389, + 129 + ], + [ + 408, + 123 + ], + [ + 378, + 111 + ] + ] + ] + }, + { + "task_id": "0001143", + "id": "RH20T_cfg4_task_0002_user_0010_scene_0008_cfg_0004_104122062823#101", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001143/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 295, + 40 + ], + [ + 283, + 27 + ], + [ + 274, + 15 + ], + [ + 317, + 66 + ], + [ + 285, + 28 + ], + [ + 326, + 77 + ] + ], + [ + [ + 264, + 67 + ], + [ + 266, + 34 + ], + [ + 297, + 68 + ], + [ + 230, + 66 + ], + [ + 294, + 98 + ], + [ + 229, + 97 + ] + ], + [ + [ + 261, + 63 + ], + [ + 283, + 60 + ], + [ + 255, + 30 + ], + [ + 268, + 96 + ], + [ + 230, + 32 + ], + [ + 245, + 102 + ] + ], + [ + [ + 279, + 57 + ], + [ + 278, + 25 + ], + [ + 308, + 66 + ], + [ + 253, + 48 + ], + [ + 307, + 95 + ], + [ + 255, + 76 + ] + ] + ] + }, + { + "task_id": "0001144", + "id": "RH20T_cfg4_task_0003_user_0010_scene_0010_cfg_0004_104122062823#107", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001144/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 66 + ], + [ + 409, + 33 + ], + [ + 444, + 63 + ], + [ + 369, + 69 + ], + [ + 438, + 92 + ], + [ + 368, + 98 + ] + ], + [ + [ + 372, + 82 + ], + [ + 373, + 48 + ], + [ + 337, + 83 + ], + [ + 408, + 81 + ], + [ + 335, + 117 + ], + [ + 408, + 116 + ] + ], + [ + [ + 435, + 79 + ], + [ + 407, + 56 + ], + [ + 414, + 107 + ], + [ + 457, + 50 + ], + [ + 442, + 129 + ], + [ + 485, + 72 + ] + ], + [ + [ + 387, + 71 + ], + [ + 360, + 76 + ], + [ + 365, + 71 + ], + [ + 406, + 71 + ], + [ + 396, + 64 + ], + [ + 435, + 65 + ] + ] + ] + }, + { + "task_id": "0001149", + "id": "RH20T_cfg6_task_0106_user_0020_scene_0004_cfg_0006_104122061330#193", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the object down from the shelf'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001149/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 192, + 223 + ], + [ + 147, + 223 + ], + [ + 205, + 187 + ], + [ + 181, + 256 + ], + [ + 247, + 189 + ], + [ + 221, + 255 + ] + ], + [ + [ + 208, + 194 + ], + [ + 189, + 176 + ], + [ + 186, + 228 + ], + [ + 229, + 162 + ], + [ + 204, + 242 + ], + [ + 243, + 179 + ] + ], + [ + [ + 194, + 159 + ], + [ + 210, + 195 + ], + [ + 179, + 173 + ], + [ + 211, + 143 + ], + [ + 164, + 140 + ], + [ + 193, + 105 + ] + ], + [ + [ + 238, + 203 + ], + [ + 271, + 217 + ], + [ + 225, + 185 + ], + [ + 250, + 219 + ], + [ + 192, + 172 + ], + [ + 221, + 206 + ] + ] + ] + }, + { + "task_id": "0001152", + "id": "RH20T_cfg1_task_0081_user_0010_scene_0009_cfg_0001_038522062547#84", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001152/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 388, + 78 + ], + [ + 370, + 74 + ], + [ + 377, + 104 + ], + [ + 400, + 51 + ], + [ + 393, + 106 + ], + [ + 416, + 55 + ] + ], + [ + [ + 410, + 52 + ], + [ + 422, + 57 + ], + [ + 387, + 50 + ], + [ + 436, + 54 + ], + [ + 374, + 44 + ], + [ + 424, + 48 + ] + ], + [ + [ + 379, + 55 + ], + [ + 383, + 38 + ], + [ + 405, + 65 + ], + [ + 355, + 45 + ], + [ + 400, + 80 + ], + [ + 352, + 62 + ] + ], + [ + [ + 367, + 61 + ], + [ + 364, + 39 + ], + [ + 393, + 57 + ], + [ + 342, + 65 + ], + [ + 394, + 77 + ], + [ + 345, + 84 + ] + ] + ] + }, + { + "task_id": "0001153", + "id": "RH20T_cfg3_task_0007_user_0016_scene_0007_cfg_0003_104122062295#163", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001153/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 100, + 205 + ], + [ + 114, + 213 + ], + [ + 123, + 167 + ], + [ + 77, + 245 + ], + [ + 110, + 163 + ], + [ + 67, + 236 + ] + ], + [ + [ + 94, + 263 + ], + [ + 71, + 307 + ], + [ + 125, + 256 + ], + [ + 66, + 268 + ], + [ + 144, + 216 + ], + [ + 87, + 231 + ] + ], + [ + [ + 74, + 272 + ], + [ + 108, + 291 + ], + [ + 37, + 304 + ], + [ + 110, + 241 + ], + [ + 0, + 285 + ], + [ + 74, + 217 + ] + ], + [ + [ + 100, + 246 + ], + [ + 100, + 213 + ], + [ + 71, + 223 + ], + [ + 130, + 270 + ], + [ + 72, + 253 + ], + [ + 128, + 300 + ] + ] + ] + }, + { + "task_id": "0001157", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122061602#302", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001157/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 505, + 225 + ], + [ + 479, + 240 + ], + [ + 487, + 201 + ], + [ + 524, + 249 + ], + [ + 513, + 184 + ], + [ + 551, + 233 + ] + ], + [ + [ + 503, + 220 + ], + [ + 514, + 203 + ], + [ + 523, + 241 + ], + [ + 485, + 200 + ], + [ + 511, + 256 + ], + [ + 476, + 215 + ] + ], + [ + [ + 525, + 238 + ], + [ + 551, + 246 + ], + [ + 526, + 262 + ], + [ + 525, + 215 + ], + [ + 500, + 252 + ], + [ + 500, + 207 + ] + ], + [ + [ + 505, + 195 + ], + [ + 518, + 214 + ], + [ + 516, + 207 + ], + [ + 494, + 183 + ], + [ + 505, + 189 + ], + [ + 482, + 165 + ] + ] + ] + }, + { + "task_id": "0001161", + "id": "RH20T_cfg4_task_0105_user_0007_scene_0014_cfg_0004_104122062295#182", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001161/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 140 + ], + [ + 305, + 145 + ], + [ + 338, + 129 + ], + [ + 261, + 150 + ], + [ + 335, + 121 + ], + [ + 249, + 145 + ] + ], + [ + [ + 265, + 119 + ], + [ + 290, + 93 + ], + [ + 297, + 139 + ], + [ + 231, + 97 + ], + [ + 274, + 162 + ], + [ + 209, + 123 + ] + ], + [ + [ + 266, + 146 + ], + [ + 275, + 110 + ], + [ + 228, + 133 + ], + [ + 304, + 158 + ], + [ + 221, + 165 + ], + [ + 294, + 190 + ] + ], + [ + [ + 266, + 153 + ], + [ + 243, + 117 + ], + [ + 283, + 142 + ], + [ + 252, + 162 + ], + [ + 308, + 181 + ], + [ + 273, + 195 + ] + ] + ] + }, + { + "task_id": "0001164", + "id": "RH20T_cfg6_task_0091_user_0020_scene_0003_cfg_0006_104122061018#361", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001164/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 288, + 139 + ], + [ + 270, + 149 + ], + [ + 295, + 166 + ], + [ + 280, + 110 + ], + [ + 310, + 155 + ], + [ + 297, + 102 + ] + ], + [ + [ + 274, + 106 + ], + [ + 281, + 85 + ], + [ + 262, + 83 + ], + [ + 287, + 129 + ], + [ + 257, + 102 + ], + [ + 281, + 146 + ] + ], + [ + [ + 270, + 98 + ], + [ + 286, + 99 + ], + [ + 246, + 93 + ], + [ + 296, + 102 + ], + [ + 227, + 92 + ], + [ + 280, + 101 + ] + ], + [ + [ + 248, + 97 + ], + [ + 275, + 89 + ], + [ + 255, + 94 + ], + [ + 240, + 101 + ], + [ + 225, + 103 + ], + [ + 213, + 109 + ] + ] + ] + }, + { + "task_id": "0001166", + "id": "RH20T_cfg4_task_0009_user_0010_scene_0006_cfg_0004_104122062295#152", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001166/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 263, + 142 + ], + [ + 236, + 174 + ], + [ + 304, + 161 + ], + [ + 224, + 125 + ], + [ + 326, + 131 + ], + [ + 249, + 99 + ] + ], + [ + [ + 204, + 166 + ], + [ + 208, + 129 + ], + [ + 174, + 144 + ], + [ + 237, + 190 + ], + [ + 172, + 178 + ], + [ + 232, + 223 + ] + ], + [ + [ + 215, + 187 + ], + [ + 192, + 151 + ], + [ + 179, + 208 + ], + [ + 251, + 166 + ], + [ + 202, + 242 + ], + [ + 272, + 201 + ] + ], + [ + [ + 210, + 149 + ], + [ + 190, + 185 + ], + [ + 227, + 158 + ], + [ + 195, + 141 + ], + [ + 248, + 120 + ], + [ + 213, + 107 + ] + ] + ] + }, + { + "task_id": "0001171", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0002_cfg_0006_104122064161#903", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001171/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 435, + 160 + ], + [ + 451, + 139 + ], + [ + 436, + 150 + ], + [ + 434, + 168 + ], + [ + 419, + 172 + ], + [ + 418, + 188 + ] + ], + [ + [ + 415, + 165 + ], + [ + 436, + 151 + ], + [ + 428, + 189 + ], + [ + 403, + 143 + ], + [ + 407, + 202 + ], + [ + 384, + 157 + ] + ], + [ + [ + 412, + 173 + ], + [ + 414, + 192 + ], + [ + 405, + 192 + ], + [ + 419, + 153 + ], + [ + 403, + 174 + ], + [ + 416, + 136 + ] + ], + [ + [ + 399, + 145 + ], + [ + 373, + 143 + ], + [ + 401, + 119 + ], + [ + 397, + 172 + ], + [ + 427, + 122 + ], + [ + 424, + 175 + ] + ] + ] + }, + { + "task_id": "0001172", + "id": "RH20T_cfg7_task_0015_user_0020_scene_0007_cfg_0007_104122061018#220", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001172/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 170 + ], + [ + 320, + 171 + ], + [ + 355, + 201 + ], + [ + 339, + 140 + ], + [ + 380, + 200 + ], + [ + 363, + 141 + ] + ], + [ + [ + 350, + 188 + ], + [ + 334, + 189 + ], + [ + 354, + 222 + ], + [ + 346, + 154 + ], + [ + 373, + 224 + ], + [ + 364, + 151 + ] + ], + [ + [ + 391, + 218 + ], + [ + 365, + 232 + ], + [ + 389, + 244 + ], + [ + 394, + 189 + ], + [ + 412, + 231 + ], + [ + 419, + 177 + ] + ], + [ + [ + 342, + 159 + ], + [ + 364, + 152 + ], + [ + 359, + 186 + ], + [ + 326, + 132 + ], + [ + 335, + 195 + ], + [ + 302, + 138 + ] + ] + ] + }, + { + "task_id": "0001173", + "id": "RH20T_cfg1_task_0007_user_0015_scene_0002_cfg_0001_750612070853#447", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001173/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 275, + 222 + ], + [ + 267, + 186 + ], + [ + 261, + 201 + ], + [ + 292, + 247 + ], + [ + 269, + 233 + ], + [ + 299, + 281 + ] + ], + [ + [ + 273, + 236 + ], + [ + 303, + 210 + ], + [ + 288, + 253 + ], + [ + 256, + 216 + ], + [ + 260, + 279 + ], + [ + 222, + 244 + ] + ], + [ + [ + 316, + 240 + ], + [ + 313, + 279 + ], + [ + 271, + 239 + ], + [ + 362, + 241 + ], + [ + 277, + 206 + ], + [ + 362, + 208 + ] + ], + [ + [ + 253, + 244 + ], + [ + 267, + 249 + ], + [ + 292, + 215 + ], + [ + 212, + 275 + ], + [ + 278, + 214 + ], + [ + 204, + 268 + ] + ] + ] + }, + { + "task_id": "0001174", + "id": "RH20T_cfg1_task_0007_user_0015_scene_0002_cfg_0001_750612070853#635", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001174/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 232 + ], + [ + 309, + 192 + ], + [ + 293, + 212 + ], + [ + 341, + 254 + ], + [ + 299, + 247 + ], + [ + 346, + 292 + ] + ], + [ + [ + 279, + 218 + ], + [ + 245, + 185 + ], + [ + 247, + 256 + ], + [ + 311, + 178 + ], + [ + 281, + 292 + ], + [ + 349, + 213 + ] + ], + [ + [ + 336, + 206 + ], + [ + 357, + 229 + ], + [ + 294, + 224 + ], + [ + 377, + 189 + ], + [ + 279, + 202 + ], + [ + 356, + 172 + ] + ], + [ + [ + 304, + 248 + ], + [ + 284, + 272 + ], + [ + 293, + 211 + ], + [ + 314, + 283 + ], + [ + 312, + 192 + ], + [ + 330, + 260 + ] + ] + ] + }, + { + "task_id": "0001175", + "id": "RH20T_cfg1_task_0007_user_0015_scene_0002_cfg_0001_750612070853#792", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001175/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 431, + 342 + ], + [ + 432, + 299 + ], + [ + 391, + 306 + ], + [ + 479, + 360 + ], + [ + 392, + 343 + ], + [ + 474, + 360 + ] + ], + [ + [ + 363, + 359 + ], + [ + 321, + 359 + ], + [ + 342, + 360 + ], + [ + 385, + 312 + ], + [ + 379, + 360 + ], + [ + 422, + 314 + ] + ], + [ + [ + 466, + 360 + ], + [ + 486, + 360 + ], + [ + 405, + 360 + ], + [ + 537, + 360 + ], + [ + 378, + 352 + ], + [ + 518, + 360 + ] + ], + [ + [ + 439, + 360 + ], + [ + 397, + 360 + ], + [ + 408, + 303 + ], + [ + 469, + 360 + ], + [ + 447, + 267 + ], + [ + 503, + 360 + ] + ] + ] + }, + { + "task_id": "0001177", + "id": "RH20T_cfg4_task_0011_user_0016_scene_0002_cfg_0004_104122062295#150", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001177/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 397, + 272 + ], + [ + 434, + 317 + ], + [ + 355, + 301 + ], + [ + 435, + 246 + ], + [ + 322, + 257 + ], + [ + 401, + 208 + ] + ], + [ + [ + 376, + 240 + ], + [ + 424, + 223 + ], + [ + 396, + 288 + ], + [ + 359, + 196 + ], + [ + 344, + 306 + ], + [ + 313, + 211 + ] + ], + [ + [ + 370, + 192 + ], + [ + 377, + 186 + ], + [ + 314, + 191 + ], + [ + 431, + 194 + ], + [ + 297, + 197 + ], + [ + 430, + 202 + ] + ], + [ + [ + 381, + 246 + ], + [ + 411, + 206 + ], + [ + 343, + 266 + ], + [ + 414, + 229 + ], + [ + 318, + 302 + ], + [ + 388, + 263 + ] + ] + ] + }, + { + "task_id": "0001178", + "id": "RH20T_cfg6_task_0008_user_0014_scene_0007_cfg_0006_104122064161#106", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001178/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 357, + 197 + ], + [ + 359, + 180 + ], + [ + 356, + 217 + ], + [ + 357, + 179 + ], + [ + 354, + 233 + ], + [ + 354, + 196 + ] + ], + [ + [ + 352, + 204 + ], + [ + 353, + 181 + ], + [ + 378, + 207 + ], + [ + 328, + 200 + ], + [ + 376, + 229 + ], + [ + 328, + 222 + ] + ], + [ + [ + 374, + 197 + ], + [ + 391, + 179 + ], + [ + 390, + 216 + ], + [ + 358, + 177 + ], + [ + 373, + 235 + ], + [ + 340, + 196 + ] + ], + [ + [ + 375, + 195 + ], + [ + 384, + 203 + ], + [ + 359, + 180 + ], + [ + 390, + 209 + ], + [ + 351, + 173 + ], + [ + 381, + 201 + ] + ] + ] + }, + { + "task_id": "0001179", + "id": "RH20T_cfg7_task_0029_user_0020_scene_0002_cfg_0007_104122063633#214", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001179/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 169, + 194 + ], + [ + 174, + 201 + ], + [ + 168, + 223 + ], + [ + 169, + 166 + ], + [ + 161, + 217 + ], + [ + 163, + 156 + ] + ], + [ + [ + 141, + 182 + ], + [ + 163, + 204 + ], + [ + 121, + 201 + ], + [ + 162, + 162 + ], + [ + 99, + 178 + ], + [ + 140, + 139 + ] + ], + [ + [ + 175, + 163 + ], + [ + 176, + 158 + ], + [ + 167, + 191 + ], + [ + 183, + 134 + ], + [ + 164, + 198 + ], + [ + 182, + 137 + ] + ], + [ + [ + 154, + 201 + ], + [ + 145, + 214 + ], + [ + 181, + 196 + ], + [ + 127, + 206 + ], + [ + 188, + 185 + ], + [ + 137, + 194 + ] + ] + ] + }, + { + "task_id": "0001181", + "id": "RH20T_cfg5_task_0020_user_0010_scene_0007_cfg_0005_036422060215#230", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001181/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 222, + 179 + ], + [ + 239, + 187 + ], + [ + 206, + 178 + ], + [ + 236, + 180 + ], + [ + 190, + 170 + ], + [ + 221, + 172 + ] + ], + [ + [ + 224, + 188 + ], + [ + 224, + 166 + ], + [ + 245, + 184 + ], + [ + 202, + 192 + ], + [ + 245, + 205 + ], + [ + 202, + 213 + ] + ], + [ + [ + 239, + 179 + ], + [ + 237, + 199 + ], + [ + 216, + 175 + ], + [ + 262, + 183 + ], + [ + 219, + 155 + ], + [ + 263, + 164 + ] + ], + [ + [ + 206, + 189 + ], + [ + 220, + 201 + ], + [ + 202, + 209 + ], + [ + 210, + 168 + ], + [ + 189, + 197 + ], + [ + 197, + 157 + ] + ] + ] + }, + { + "task_id": "0001182", + "id": "RH20T_cfg5_task_0004_user_0010_scene_0001_cfg_0005_104422070044#155", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001182/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 490, + 252 + ], + [ + 511, + 259 + ], + [ + 498, + 236 + ], + [ + 484, + 266 + ], + [ + 476, + 229 + ], + [ + 464, + 259 + ] + ], + [ + [ + 484, + 205 + ], + [ + 493, + 191 + ], + [ + 499, + 218 + ], + [ + 469, + 192 + ], + [ + 490, + 233 + ], + [ + 460, + 206 + ] + ], + [ + [ + 486, + 223 + ], + [ + 488, + 210 + ], + [ + 507, + 232 + ], + [ + 467, + 215 + ], + [ + 504, + 244 + ], + [ + 466, + 228 + ] + ], + [ + [ + 467, + 225 + ], + [ + 453, + 237 + ], + [ + 480, + 226 + ], + [ + 455, + 224 + ], + [ + 493, + 214 + ], + [ + 468, + 213 + ] + ] + ] + }, + { + "task_id": "0001185", + "id": "RH20T_cfg1_task_0004_user_0005_scene_0003_cfg_0001_038522062547#134", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001185/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 364, + 229 + ], + [ + 361, + 230 + ], + [ + 383, + 249 + ], + [ + 346, + 209 + ], + [ + 389, + 249 + ], + [ + 349, + 206 + ] + ], + [ + [ + 382, + 236 + ], + [ + 389, + 228 + ], + [ + 383, + 264 + ], + [ + 381, + 209 + ], + [ + 378, + 270 + ], + [ + 375, + 218 + ] + ], + [ + [ + 393, + 266 + ], + [ + 369, + 268 + ], + [ + 403, + 245 + ], + [ + 383, + 286 + ], + [ + 426, + 244 + ], + [ + 405, + 284 + ] + ], + [ + [ + 364, + 243 + ], + [ + 388, + 254 + ], + [ + 369, + 266 + ], + [ + 360, + 223 + ], + [ + 343, + 254 + ], + [ + 337, + 212 + ] + ] + ] + }, + { + "task_id": "0001186", + "id": "RH20T_cfg5_task_0072_user_0007_scene_0005_cfg_0005_105422061350#406", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drop coins into a piggy bank'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001186/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 156, + 59 + ], + [ + 169, + 78 + ], + [ + 154, + 82 + ], + [ + 159, + 37 + ], + [ + 139, + 62 + ], + [ + 145, + 15 + ] + ], + [ + [ + 173, + 55 + ], + [ + 202, + 65 + ], + [ + 177, + 77 + ], + [ + 169, + 29 + ], + [ + 150, + 68 + ], + [ + 139, + 19 + ] + ], + [ + [ + 186, + 61 + ], + [ + 186, + 38 + ], + [ + 200, + 43 + ], + [ + 172, + 80 + ], + [ + 199, + 63 + ], + [ + 173, + 100 + ] + ], + [ + [ + 195, + 93 + ], + [ + 219, + 93 + ], + [ + 195, + 121 + ], + [ + 194, + 63 + ], + [ + 169, + 121 + ], + [ + 167, + 61 + ] + ] + ] + }, + { + "task_id": "0001194", + "id": "RH20T_cfg5_task_0041_user_0007_scene_0006_cfg_0005_105422061350#298", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the tabletop with a sponge'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001194/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 196, + 212 + ], + [ + 173, + 232 + ], + [ + 175, + 189 + ], + [ + 216, + 234 + ], + [ + 198, + 169 + ], + [ + 238, + 214 + ] + ], + [ + [ + 190, + 167 + ], + [ + 185, + 150 + ], + [ + 208, + 147 + ], + [ + 171, + 189 + ], + [ + 211, + 163 + ], + [ + 176, + 204 + ] + ], + [ + [ + 195, + 197 + ], + [ + 203, + 192 + ], + [ + 179, + 221 + ], + [ + 210, + 174 + ], + [ + 172, + 224 + ], + [ + 203, + 179 + ] + ], + [ + [ + 216, + 159 + ], + [ + 193, + 148 + ], + [ + 226, + 175 + ], + [ + 204, + 142 + ], + [ + 249, + 187 + ], + [ + 229, + 154 + ] + ] + ] + }, + { + "task_id": "0001197", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_750612070851#107", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001197/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 120, + 223 + ], + [ + 148, + 205 + ], + [ + 140, + 243 + ], + [ + 97, + 201 + ], + [ + 112, + 260 + ], + [ + 66, + 220 + ] + ], + [ + [ + 83, + 184 + ], + [ + 86, + 149 + ], + [ + 42, + 183 + ], + [ + 120, + 185 + ], + [ + 40, + 219 + ], + [ + 117, + 218 + ] + ], + [ + [ + 105, + 158 + ], + [ + 114, + 165 + ], + [ + 140, + 151 + ], + [ + 68, + 165 + ], + [ + 132, + 143 + ], + [ + 54, + 158 + ] + ], + [ + [ + 125, + 184 + ], + [ + 113, + 161 + ], + [ + 99, + 207 + ], + [ + 149, + 163 + ], + [ + 112, + 227 + ], + [ + 159, + 184 + ] + ] + ] + }, + { + "task_id": "0001199", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_750612070851#196", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001199/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 119, + 208 + ], + [ + 122, + 224 + ], + [ + 95, + 237 + ], + [ + 141, + 181 + ], + [ + 90, + 219 + ], + [ + 140, + 161 + ] + ], + [ + [ + 133, + 208 + ], + [ + 122, + 185 + ], + [ + 105, + 231 + ], + [ + 158, + 186 + ], + [ + 116, + 251 + ], + [ + 166, + 207 + ] + ], + [ + [ + 102, + 209 + ], + [ + 129, + 232 + ], + [ + 89, + 206 + ], + [ + 114, + 210 + ], + [ + 59, + 181 + ], + [ + 89, + 188 + ] + ], + [ + [ + 146, + 192 + ], + [ + 164, + 213 + ], + [ + 133, + 216 + ], + [ + 160, + 167 + ], + [ + 116, + 196 + ], + [ + 142, + 147 + ] + ] + ] + }, + { + "task_id": "0001201", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_750612070851#291", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001201/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 174, + 237 + ], + [ + 168, + 215 + ], + [ + 148, + 263 + ], + [ + 199, + 213 + ], + [ + 155, + 282 + ], + [ + 203, + 233 + ] + ], + [ + [ + 182, + 230 + ], + [ + 172, + 208 + ], + [ + 159, + 256 + ], + [ + 204, + 206 + ], + [ + 169, + 275 + ], + [ + 211, + 226 + ] + ], + [ + [ + 168, + 269 + ], + [ + 151, + 250 + ], + [ + 133, + 281 + ], + [ + 202, + 257 + ], + [ + 150, + 297 + ], + [ + 215, + 275 + ] + ], + [ + [ + 142, + 239 + ], + [ + 172, + 232 + ], + [ + 124, + 255 + ], + [ + 162, + 223 + ], + [ + 90, + 264 + ], + [ + 128, + 230 + ] + ] + ] + }, + { + "task_id": "0001202", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104122063550#74", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001202/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 276, + 219 + ], + [ + 246, + 218 + ], + [ + 289, + 198 + ], + [ + 264, + 239 + ], + [ + 318, + 201 + ], + [ + 291, + 240 + ] + ], + [ + [ + 294, + 193 + ], + [ + 292, + 179 + ], + [ + 306, + 222 + ], + [ + 282, + 164 + ], + [ + 307, + 232 + ], + [ + 284, + 177 + ] + ], + [ + [ + 283, + 182 + ], + [ + 256, + 181 + ], + [ + 302, + 174 + ], + [ + 266, + 190 + ], + [ + 327, + 176 + ], + [ + 290, + 190 + ] + ], + [ + [ + 310, + 146 + ], + [ + 341, + 135 + ], + [ + 321, + 154 + ], + [ + 299, + 139 + ], + [ + 286, + 167 + ], + [ + 269, + 150 + ] + ] + ] + }, + { + "task_id": "0001203", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104122063550#84", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001203/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 287, + 196 + ], + [ + 315, + 183 + ], + [ + 282, + 210 + ], + [ + 292, + 179 + ], + [ + 254, + 223 + ], + [ + 261, + 193 + ] + ], + [ + [ + 292, + 178 + ], + [ + 290, + 165 + ], + [ + 305, + 207 + ], + [ + 280, + 149 + ], + [ + 306, + 216 + ], + [ + 283, + 162 + ] + ], + [ + [ + 319, + 201 + ], + [ + 311, + 233 + ], + [ + 323, + 198 + ], + [ + 313, + 204 + ], + [ + 331, + 168 + ], + [ + 322, + 169 + ] + ], + [ + [ + 289, + 145 + ], + [ + 318, + 146 + ], + [ + 269, + 144 + ], + [ + 306, + 145 + ], + [ + 240, + 143 + ], + [ + 280, + 144 + ] + ] + ] + }, + { + "task_id": "0001204", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104122063550#99", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001204/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 352, + 164 + ], + [ + 382, + 165 + ], + [ + 350, + 187 + ], + [ + 354, + 139 + ], + [ + 322, + 186 + ], + [ + 323, + 138 + ] + ], + [ + [ + 331, + 153 + ], + [ + 334, + 135 + ], + [ + 346, + 178 + ], + [ + 317, + 130 + ], + [ + 343, + 193 + ], + [ + 316, + 147 + ] + ], + [ + [ + 317, + 169 + ], + [ + 342, + 152 + ], + [ + 331, + 194 + ], + [ + 303, + 146 + ], + [ + 305, + 211 + ], + [ + 278, + 163 + ] + ], + [ + [ + 353, + 134 + ], + [ + 322, + 137 + ], + [ + 352, + 102 + ], + [ + 353, + 164 + ], + [ + 383, + 100 + ], + [ + 382, + 161 + ] + ] + ] + }, + { + "task_id": "0001205", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104122063550#130", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001205/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 263, + 219 + ], + [ + 291, + 207 + ], + [ + 275, + 226 + ], + [ + 249, + 212 + ], + [ + 249, + 238 + ], + [ + 220, + 225 + ] + ], + [ + [ + 235, + 177 + ], + [ + 246, + 180 + ], + [ + 265, + 179 + ], + [ + 204, + 174 + ], + [ + 254, + 176 + ], + [ + 196, + 171 + ] + ], + [ + [ + 246, + 202 + ], + [ + 261, + 183 + ], + [ + 244, + 224 + ], + [ + 248, + 178 + ], + [ + 227, + 246 + ], + [ + 231, + 198 + ] + ], + [ + [ + 251, + 203 + ], + [ + 259, + 188 + ], + [ + 264, + 231 + ], + [ + 238, + 176 + ], + [ + 255, + 242 + ], + [ + 231, + 190 + ] + ] + ] + }, + { + "task_id": "0001206", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104122063550#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001206/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 327, + 211 + ], + [ + 318, + 243 + ], + [ + 338, + 204 + ], + [ + 317, + 217 + ], + [ + 347, + 172 + ], + [ + 326, + 189 + ] + ], + [ + [ + 341, + 225 + ], + [ + 329, + 240 + ], + [ + 322, + 239 + ], + [ + 362, + 210 + ], + [ + 333, + 226 + ], + [ + 372, + 198 + ] + ], + [ + [ + 292, + 213 + ], + [ + 307, + 241 + ], + [ + 264, + 228 + ], + [ + 321, + 199 + ], + [ + 249, + 200 + ], + [ + 306, + 171 + ] + ], + [ + [ + 320, + 223 + ], + [ + 322, + 207 + ], + [ + 341, + 246 + ], + [ + 299, + 201 + ], + [ + 337, + 259 + ], + [ + 298, + 216 + ] + ] + ] + }, + { + "task_id": "0001208", + "id": "RH20T_cfg5_task_0010_user_0010_scene_0009_cfg_0005_104422070044#103", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001208/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 369, + 173 + ], + [ + 363, + 157 + ], + [ + 376, + 165 + ], + [ + 362, + 180 + ], + [ + 382, + 182 + ], + [ + 367, + 195 + ] + ], + [ + [ + 360, + 159 + ], + [ + 370, + 158 + ], + [ + 360, + 176 + ], + [ + 360, + 142 + ], + [ + 351, + 177 + ], + [ + 352, + 144 + ] + ], + [ + [ + 377, + 139 + ], + [ + 362, + 148 + ], + [ + 369, + 136 + ], + [ + 385, + 142 + ], + [ + 383, + 127 + ], + [ + 399, + 133 + ] + ], + [ + [ + 375, + 158 + ], + [ + 373, + 147 + ], + [ + 388, + 166 + ], + [ + 362, + 151 + ], + [ + 390, + 177 + ], + [ + 364, + 162 + ] + ] + ] + }, + { + "task_id": "0001209", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0003_cfg_0005_037522062165#315", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001209/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 327, + 252 + ], + [ + 306, + 234 + ], + [ + 307, + 271 + ], + [ + 345, + 233 + ], + [ + 329, + 290 + ], + [ + 367, + 251 + ] + ], + [ + [ + 300, + 225 + ], + [ + 300, + 242 + ], + [ + 292, + 204 + ], + [ + 307, + 245 + ], + [ + 293, + 190 + ], + [ + 307, + 230 + ] + ], + [ + [ + 309, + 244 + ], + [ + 311, + 228 + ], + [ + 327, + 227 + ], + [ + 290, + 261 + ], + [ + 325, + 242 + ], + [ + 290, + 275 + ] + ], + [ + [ + 329, + 241 + ], + [ + 302, + 231 + ], + [ + 320, + 267 + ], + [ + 337, + 214 + ], + [ + 347, + 277 + ], + [ + 365, + 223 + ] + ] + ] + }, + { + "task_id": "0001214", + "id": "RH20T_cfg1_task_0004_user_0005_scene_0003_cfg_0001_750612070853#120", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001214/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 234, + 154 + ], + [ + 241, + 136 + ], + [ + 264, + 137 + ], + [ + 206, + 170 + ], + [ + 257, + 157 + ], + [ + 195, + 191 + ] + ], + [ + [ + 273, + 170 + ], + [ + 261, + 139 + ], + [ + 250, + 159 + ], + [ + 297, + 181 + ], + [ + 262, + 185 + ], + [ + 307, + 209 + ] + ], + [ + [ + 308, + 171 + ], + [ + 286, + 162 + ], + [ + 295, + 206 + ], + [ + 322, + 136 + ], + [ + 319, + 218 + ], + [ + 349, + 144 + ] + ], + [ + [ + 261, + 161 + ], + [ + 266, + 137 + ], + [ + 291, + 148 + ], + [ + 229, + 175 + ], + [ + 285, + 169 + ], + [ + 226, + 196 + ] + ] + ] + }, + { + "task_id": "0001216", + "id": "RH20T_cfg3_task_0043_user_0010_scene_0006_cfg_0003_038522062288#272", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001216/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 132, + 229 + ], + [ + 117, + 216 + ], + [ + 96, + 239 + ], + [ + 167, + 220 + ], + [ + 111, + 249 + ], + [ + 177, + 232 + ] + ], + [ + [ + 144, + 228 + ], + [ + 134, + 195 + ], + [ + 133, + 226 + ], + [ + 153, + 230 + ], + [ + 144, + 263 + ], + [ + 163, + 262 + ] + ], + [ + [ + 141, + 206 + ], + [ + 133, + 187 + ], + [ + 105, + 209 + ], + [ + 175, + 204 + ], + [ + 113, + 231 + ], + [ + 185, + 224 + ] + ], + [ + [ + 102, + 234 + ], + [ + 83, + 236 + ], + [ + 85, + 266 + ], + [ + 117, + 205 + ], + [ + 105, + 265 + ], + [ + 137, + 202 + ] + ] + ] + }, + { + "task_id": "0001217", + "id": "RH20T_cfg3_task_0043_user_0010_scene_0006_cfg_0003_038522062288#349", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001217/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 130, + 202 + ], + [ + 162, + 207 + ], + [ + 150, + 179 + ], + [ + 111, + 223 + ], + [ + 114, + 171 + ], + [ + 75, + 218 + ] + ], + [ + [ + 135, + 240 + ], + [ + 120, + 224 + ], + [ + 99, + 251 + ], + [ + 170, + 228 + ], + [ + 114, + 265 + ], + [ + 181, + 243 + ] + ], + [ + [ + 112, + 235 + ], + [ + 108, + 203 + ], + [ + 73, + 238 + ], + [ + 147, + 231 + ], + [ + 76, + 273 + ], + [ + 152, + 263 + ] + ], + [ + [ + 110, + 244 + ], + [ + 93, + 262 + ], + [ + 96, + 215 + ], + [ + 124, + 272 + ], + [ + 113, + 201 + ], + [ + 138, + 255 + ] + ] + ] + }, + { + "task_id": "0001219", + "id": "RH20T_cfg7_task_0006_user_0020_scene_0006_cfg_0007_104122061602#200", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001219/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 534, + 241 + ], + [ + 508, + 251 + ], + [ + 547, + 266 + ], + [ + 521, + 217 + ], + [ + 573, + 256 + ], + [ + 547, + 207 + ] + ], + [ + [ + 542, + 257 + ], + [ + 566, + 279 + ], + [ + 532, + 275 + ], + [ + 554, + 238 + ], + [ + 512, + 256 + ], + [ + 531, + 218 + ] + ], + [ + [ + 537, + 281 + ], + [ + 559, + 304 + ], + [ + 562, + 266 + ], + [ + 514, + 296 + ], + [ + 539, + 243 + ], + [ + 493, + 274 + ] + ], + [ + [ + 555, + 277 + ], + [ + 582, + 267 + ], + [ + 534, + 257 + ], + [ + 579, + 299 + ], + [ + 509, + 267 + ], + [ + 551, + 310 + ] + ] + ] + }, + { + "task_id": "0001220", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112070044#108", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001220/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 388, + 131 + ], + [ + 387, + 128 + ], + [ + 369, + 137 + ], + [ + 407, + 125 + ], + [ + 369, + 141 + ], + [ + 409, + 128 + ] + ], + [ + [ + 399, + 133 + ], + [ + 401, + 116 + ], + [ + 414, + 127 + ], + [ + 384, + 138 + ], + [ + 411, + 143 + ], + [ + 382, + 154 + ] + ], + [ + [ + 405, + 108 + ], + [ + 417, + 113 + ], + [ + 390, + 104 + ], + [ + 418, + 112 + ], + [ + 379, + 99 + ], + [ + 407, + 107 + ] + ], + [ + [ + 400, + 126 + ], + [ + 413, + 134 + ], + [ + 411, + 129 + ], + [ + 390, + 122 + ], + [ + 398, + 121 + ], + [ + 377, + 114 + ] + ] + ] + }, + { + "task_id": "0001221", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112070044#173", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001221/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 383, + 149 + ], + [ + 373, + 142 + ], + [ + 369, + 143 + ], + [ + 395, + 155 + ], + [ + 379, + 151 + ], + [ + 406, + 163 + ] + ], + [ + [ + 390, + 129 + ], + [ + 403, + 134 + ], + [ + 376, + 135 + ], + [ + 404, + 124 + ], + [ + 364, + 130 + ], + [ + 392, + 119 + ] + ], + [ + [ + 405, + 138 + ], + [ + 404, + 124 + ], + [ + 401, + 152 + ], + [ + 409, + 125 + ], + [ + 402, + 165 + ], + [ + 410, + 138 + ] + ], + [ + [ + 397, + 134 + ], + [ + 399, + 116 + ], + [ + 412, + 130 + ], + [ + 381, + 138 + ], + [ + 410, + 146 + ], + [ + 379, + 155 + ] + ] + ] + }, + { + "task_id": "0001222", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112070044#261", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001222/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 398, + 134 + ], + [ + 401, + 117 + ], + [ + 412, + 128 + ], + [ + 383, + 139 + ], + [ + 409, + 144 + ], + [ + 381, + 155 + ] + ], + [ + [ + 400, + 119 + ], + [ + 394, + 108 + ], + [ + 403, + 134 + ], + [ + 398, + 106 + ], + [ + 409, + 145 + ], + [ + 404, + 117 + ] + ], + [ + [ + 393, + 152 + ], + [ + 409, + 150 + ], + [ + 400, + 171 + ], + [ + 387, + 134 + ], + [ + 385, + 173 + ], + [ + 372, + 137 + ] + ], + [ + [ + 418, + 122 + ], + [ + 399, + 116 + ], + [ + 422, + 108 + ], + [ + 414, + 137 + ], + [ + 440, + 115 + ], + [ + 433, + 143 + ] + ] + ] + }, + { + "task_id": "0001223", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112070044#350", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001223/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 135 + ], + [ + 379, + 128 + ], + [ + 380, + 144 + ], + [ + 406, + 127 + ], + [ + 395, + 152 + ], + [ + 421, + 135 + ] + ], + [ + [ + 407, + 152 + ], + [ + 422, + 164 + ], + [ + 411, + 143 + ], + [ + 402, + 162 + ], + [ + 396, + 132 + ], + [ + 386, + 149 + ] + ], + [ + [ + 407, + 126 + ], + [ + 402, + 111 + ], + [ + 398, + 138 + ], + [ + 416, + 112 + ], + [ + 402, + 153 + ], + [ + 422, + 128 + ] + ], + [ + [ + 398, + 133 + ], + [ + 400, + 116 + ], + [ + 411, + 128 + ], + [ + 384, + 139 + ], + [ + 408, + 144 + ], + [ + 381, + 155 + ] + ] + ] + }, + { + "task_id": "0001227", + "id": "RH20T_cfg7_task_0065_user_0014_scene_0003_cfg_0007_104122061330#332", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001227/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 359, + 240 + ], + [ + 378, + 239 + ], + [ + 342, + 253 + ], + [ + 374, + 228 + ], + [ + 325, + 254 + ], + [ + 356, + 230 + ] + ], + [ + [ + 381, + 231 + ], + [ + 379, + 218 + ], + [ + 372, + 256 + ], + [ + 390, + 208 + ], + [ + 373, + 266 + ], + [ + 391, + 220 + ] + ], + [ + [ + 391, + 230 + ], + [ + 416, + 236 + ], + [ + 387, + 216 + ], + [ + 395, + 246 + ], + [ + 363, + 210 + ], + [ + 368, + 240 + ] + ], + [ + [ + 367, + 211 + ], + [ + 366, + 194 + ], + [ + 353, + 191 + ], + [ + 380, + 230 + ], + [ + 354, + 209 + ], + [ + 381, + 249 + ] + ] + ] + }, + { + "task_id": "0001228", + "id": "RH20T_cfg7_task_0065_user_0014_scene_0003_cfg_0007_104122061330#499", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001228/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 396, + 222 + ], + [ + 416, + 225 + ], + [ + 390, + 199 + ], + [ + 402, + 247 + ], + [ + 369, + 195 + ], + [ + 380, + 245 + ] + ], + [ + [ + 377, + 189 + ], + [ + 363, + 169 + ], + [ + 372, + 181 + ], + [ + 381, + 197 + ], + [ + 387, + 203 + ], + [ + 395, + 217 + ] + ], + [ + [ + 385, + 208 + ], + [ + 358, + 204 + ], + [ + 385, + 207 + ], + [ + 386, + 208 + ], + [ + 414, + 211 + ], + [ + 412, + 212 + ] + ], + [ + [ + 392, + 217 + ], + [ + 393, + 201 + ], + [ + 383, + 239 + ], + [ + 400, + 195 + ], + [ + 382, + 253 + ], + [ + 399, + 210 + ] + ] + ] + }, + { + "task_id": "0001229", + "id": "RH20T_cfg4_task_0009_user_0010_scene_0006_cfg_0004_036422060909#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001229/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 321, + 99 + ], + [ + 335, + 121 + ], + [ + 337, + 82 + ], + [ + 304, + 117 + ], + [ + 322, + 59 + ], + [ + 288, + 94 + ] + ], + [ + [ + 323, + 114 + ], + [ + 324, + 94 + ], + [ + 340, + 101 + ], + [ + 304, + 128 + ], + [ + 338, + 119 + ], + [ + 304, + 146 + ] + ], + [ + [ + 312, + 99 + ], + [ + 297, + 77 + ], + [ + 329, + 86 + ], + [ + 296, + 112 + ], + [ + 344, + 110 + ], + [ + 310, + 133 + ] + ], + [ + [ + 348, + 134 + ], + [ + 362, + 125 + ], + [ + 345, + 109 + ], + [ + 352, + 159 + ], + [ + 333, + 118 + ], + [ + 339, + 166 + ] + ] + ] + }, + { + "task_id": "0001232", + "id": "RH20T_cfg2_task_0022_user_0003_scene_0004_cfg_0002_104122063678#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Wave the flag'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001232/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 321, + 167 + ], + [ + 302, + 156 + ], + [ + 332, + 148 + ], + [ + 309, + 185 + ], + [ + 351, + 159 + ], + [ + 328, + 196 + ] + ], + [ + [ + 336, + 176 + ], + [ + 350, + 169 + ], + [ + 325, + 186 + ], + [ + 347, + 165 + ], + [ + 310, + 192 + ], + [ + 331, + 172 + ] + ], + [ + [ + 331, + 136 + ], + [ + 351, + 145 + ], + [ + 326, + 151 + ], + [ + 337, + 121 + ], + [ + 307, + 142 + ], + [ + 316, + 111 + ] + ], + [ + [ + 322, + 156 + ], + [ + 323, + 139 + ], + [ + 304, + 164 + ], + [ + 339, + 149 + ], + [ + 303, + 180 + ], + [ + 337, + 165 + ] + ] + ] + }, + { + "task_id": "0001234", + "id": "RH20T_cfg4_task_0071_user_0007_scene_0010_cfg_0004_038522062288#122", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001234/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 310, + 205 + ], + [ + 293, + 187 + ], + [ + 314, + 215 + ], + [ + 305, + 194 + ], + [ + 331, + 234 + ], + [ + 323, + 214 + ] + ], + [ + [ + 338, + 174 + ], + [ + 359, + 186 + ], + [ + 336, + 149 + ], + [ + 340, + 197 + ], + [ + 316, + 139 + ], + [ + 321, + 186 + ] + ], + [ + [ + 302, + 186 + ], + [ + 316, + 199 + ], + [ + 320, + 163 + ], + [ + 285, + 209 + ], + [ + 306, + 147 + ], + [ + 269, + 195 + ] + ], + [ + [ + 309, + 184 + ], + [ + 314, + 166 + ], + [ + 284, + 194 + ], + [ + 333, + 175 + ], + [ + 281, + 210 + ], + [ + 328, + 192 + ] + ] + ] + }, + { + "task_id": "0001235", + "id": "RH20T_cfg4_task_0071_user_0007_scene_0010_cfg_0004_038522062288#208", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001235/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 318, + 177 + ], + [ + 323, + 159 + ], + [ + 293, + 186 + ], + [ + 343, + 169 + ], + [ + 290, + 202 + ], + [ + 338, + 185 + ] + ], + [ + [ + 299, + 170 + ], + [ + 310, + 191 + ], + [ + 273, + 177 + ], + [ + 323, + 162 + ], + [ + 262, + 156 + ], + [ + 312, + 143 + ] + ], + [ + [ + 319, + 145 + ], + [ + 332, + 165 + ], + [ + 328, + 124 + ], + [ + 311, + 165 + ], + [ + 315, + 105 + ], + [ + 299, + 146 + ] + ], + [ + [ + 344, + 191 + ], + [ + 337, + 211 + ], + [ + 369, + 191 + ], + [ + 318, + 191 + ], + [ + 377, + 170 + ], + [ + 325, + 169 + ] + ] + ] + }, + { + "task_id": "0001236", + "id": "RH20T_cfg1_task_0064_user_0001_scene_0008_cfg_0001_038522063145#178", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001236/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 485, + 165 + ], + [ + 472, + 150 + ], + [ + 486, + 178 + ], + [ + 484, + 151 + ], + [ + 499, + 195 + ], + [ + 499, + 168 + ] + ], + [ + [ + 493, + 163 + ], + [ + 502, + 144 + ], + [ + 513, + 159 + ], + [ + 471, + 168 + ], + [ + 504, + 177 + ], + [ + 463, + 186 + ] + ], + [ + [ + 504, + 185 + ], + [ + 478, + 186 + ], + [ + 510, + 182 + ], + [ + 496, + 189 + ], + [ + 537, + 181 + ], + [ + 525, + 187 + ] + ], + [ + [ + 474, + 152 + ], + [ + 495, + 167 + ], + [ + 473, + 154 + ], + [ + 476, + 149 + ], + [ + 454, + 141 + ], + [ + 454, + 134 + ] + ] + ] + }, + { + "task_id": "0001240", + "id": "RH20T_cfg5_task_0010_user_0010_scene_0009_cfg_0005_104122061850#302", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001240/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 397, + 127 + ], + [ + 405, + 145 + ], + [ + 412, + 138 + ], + [ + 384, + 117 + ], + [ + 404, + 118 + ], + [ + 376, + 98 + ] + ], + [ + [ + 421, + 96 + ], + [ + 408, + 77 + ], + [ + 431, + 99 + ], + [ + 409, + 93 + ], + [ + 444, + 118 + ], + [ + 422, + 113 + ] + ], + [ + [ + 419, + 100 + ], + [ + 434, + 79 + ], + [ + 421, + 97 + ], + [ + 416, + 103 + ], + [ + 407, + 116 + ], + [ + 401, + 124 + ] + ], + [ + [ + 392, + 111 + ], + [ + 390, + 95 + ], + [ + 380, + 130 + ], + [ + 403, + 94 + ], + [ + 383, + 144 + ], + [ + 404, + 109 + ] + ] + ] + }, + { + "task_id": "0001242", + "id": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_819112071037#229", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001242/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 429, + 198 + ], + [ + 428, + 132 + ], + [ + 363, + 209 + ], + [ + 493, + 188 + ], + [ + 368, + 269 + ], + [ + 492, + 247 + ] + ], + [ + [ + 410, + 238 + ], + [ + 356, + 269 + ], + [ + 381, + 190 + ], + [ + 436, + 280 + ], + [ + 439, + 156 + ], + [ + 489, + 250 + ] + ], + [ + [ + 491, + 208 + ], + [ + 513, + 191 + ], + [ + 517, + 273 + ], + [ + 465, + 144 + ], + [ + 492, + 302 + ], + [ + 433, + 156 + ] + ], + [ + [ + 413, + 269 + ], + [ + 393, + 226 + ], + [ + 388, + 215 + ], + [ + 434, + 315 + ], + [ + 412, + 268 + ], + [ + 461, + 360 + ] + ] + ] + }, + { + "task_id": "0001243", + "id": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_819112071037#256", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001243/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 355, + 154 + ], + [ + 335, + 97 + ], + [ + 315, + 162 + ], + [ + 387, + 148 + ], + [ + 337, + 228 + ], + [ + 406, + 202 + ] + ], + [ + [ + 376, + 159 + ], + [ + 362, + 95 + ], + [ + 314, + 176 + ], + [ + 438, + 141 + ], + [ + 329, + 234 + ], + [ + 448, + 200 + ] + ], + [ + [ + 403, + 140 + ], + [ + 417, + 177 + ], + [ + 371, + 88 + ], + [ + 431, + 184 + ], + [ + 364, + 62 + ], + [ + 419, + 152 + ] + ], + [ + [ + 397, + 119 + ], + [ + 425, + 138 + ], + [ + 356, + 173 + ], + [ + 439, + 61 + ], + [ + 340, + 152 + ], + [ + 411, + 53 + ] + ] + ] + }, + { + "task_id": "0001247", + "id": "RH20T_cfg5_task_0039_user_0007_scene_0003_cfg_0005_104422070044#143", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001247/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 375, + 149 + ], + [ + 388, + 138 + ], + [ + 384, + 162 + ], + [ + 366, + 135 + ], + [ + 372, + 173 + ], + [ + 353, + 146 + ] + ], + [ + [ + 368, + 153 + ], + [ + 361, + 169 + ], + [ + 353, + 150 + ], + [ + 384, + 156 + ], + [ + 359, + 135 + ], + [ + 390, + 140 + ] + ], + [ + [ + 391, + 146 + ], + [ + 410, + 147 + ], + [ + 390, + 144 + ], + [ + 392, + 147 + ], + [ + 372, + 143 + ], + [ + 372, + 146 + ] + ], + [ + [ + 374, + 156 + ], + [ + 371, + 144 + ], + [ + 387, + 165 + ], + [ + 360, + 148 + ], + [ + 389, + 176 + ], + [ + 363, + 159 + ] + ] + ] + }, + { + "task_id": "0001253", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122064161#496", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001253/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 380, + 152 + ], + [ + 374, + 136 + ], + [ + 362, + 149 + ], + [ + 399, + 156 + ], + [ + 368, + 163 + ], + [ + 404, + 171 + ] + ], + [ + [ + 368, + 165 + ], + [ + 349, + 154 + ], + [ + 372, + 159 + ], + [ + 364, + 171 + ], + [ + 392, + 171 + ], + [ + 381, + 182 + ] + ], + [ + [ + 382, + 164 + ], + [ + 378, + 147 + ], + [ + 392, + 151 + ], + [ + 373, + 177 + ], + [ + 397, + 169 + ], + [ + 377, + 195 + ] + ], + [ + [ + 363, + 152 + ], + [ + 362, + 133 + ], + [ + 363, + 166 + ], + [ + 362, + 140 + ], + [ + 364, + 184 + ], + [ + 363, + 157 + ] + ] + ] + }, + { + "task_id": "0001261", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_750612070853#272", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001261/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 286, + 177 + ], + [ + 274, + 137 + ], + [ + 239, + 186 + ], + [ + 330, + 168 + ], + [ + 249, + 230 + ], + [ + 344, + 209 + ] + ], + [ + [ + 245, + 219 + ], + [ + 187, + 213 + ], + [ + 247, + 166 + ], + [ + 243, + 266 + ], + [ + 308, + 172 + ], + [ + 298, + 271 + ] + ], + [ + [ + 279, + 206 + ], + [ + 322, + 190 + ], + [ + 295, + 232 + ], + [ + 265, + 184 + ], + [ + 245, + 252 + ], + [ + 222, + 199 + ] + ], + [ + [ + 278, + 203 + ], + [ + 270, + 159 + ], + [ + 249, + 186 + ], + [ + 312, + 224 + ], + [ + 258, + 223 + ], + [ + 318, + 265 + ] + ] + ] + }, + { + "task_id": "0001262", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_750612070853#462", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001262/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 438, + 295 + ], + [ + 401, + 340 + ], + [ + 487, + 349 + ], + [ + 390, + 243 + ], + [ + 514, + 306 + ], + [ + 424, + 211 + ] + ], + [ + [ + 493, + 328 + ], + [ + 504, + 273 + ], + [ + 427, + 297 + ], + [ + 572, + 360 + ], + [ + 423, + 342 + ], + [ + 556, + 360 + ] + ], + [ + [ + 467, + 309 + ], + [ + 477, + 241 + ], + [ + 401, + 303 + ], + [ + 532, + 315 + ], + [ + 392, + 360 + ], + [ + 523, + 360 + ] + ], + [ + [ + 396, + 352 + ], + [ + 427, + 289 + ], + [ + 339, + 315 + ], + [ + 450, + 360 + ], + [ + 300, + 360 + ], + [ + 419, + 360 + ] + ] + ] + }, + { + "task_id": "0001263", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_037522061512#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001263/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 230, + 174 + ], + [ + 212, + 186 + ], + [ + 241, + 176 + ], + [ + 217, + 171 + ], + [ + 259, + 163 + ], + [ + 236, + 158 + ] + ], + [ + [ + 222, + 155 + ], + [ + 213, + 170 + ], + [ + 199, + 154 + ], + [ + 245, + 156 + ], + [ + 209, + 139 + ], + [ + 253, + 141 + ] + ], + [ + [ + 224, + 143 + ], + [ + 218, + 128 + ], + [ + 238, + 126 + ], + [ + 210, + 161 + ], + [ + 243, + 139 + ], + [ + 216, + 173 + ] + ], + [ + [ + 248, + 138 + ], + [ + 245, + 133 + ], + [ + 225, + 134 + ], + [ + 271, + 142 + ], + [ + 228, + 140 + ], + [ + 276, + 149 + ] + ] + ] + }, + { + "task_id": "0001265", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_037522061512#616", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001265/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 200, + 142 + ], + [ + 180, + 134 + ], + [ + 211, + 125 + ], + [ + 189, + 160 + ], + [ + 231, + 132 + ], + [ + 209, + 168 + ] + ], + [ + [ + 205, + 138 + ], + [ + 216, + 144 + ], + [ + 183, + 137 + ], + [ + 226, + 140 + ], + [ + 173, + 131 + ], + [ + 215, + 134 + ] + ], + [ + [ + 222, + 144 + ], + [ + 214, + 135 + ], + [ + 231, + 124 + ], + [ + 212, + 164 + ], + [ + 237, + 132 + ], + [ + 219, + 171 + ] + ], + [ + [ + 210, + 131 + ], + [ + 232, + 139 + ], + [ + 208, + 127 + ], + [ + 211, + 136 + ], + [ + 184, + 118 + ], + [ + 190, + 128 + ] + ] + ] + }, + { + "task_id": "0001266", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_037522061512#751", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001266/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 228, + 157 + ], + [ + 216, + 162 + ], + [ + 231, + 181 + ], + [ + 225, + 133 + ], + [ + 244, + 176 + ], + [ + 238, + 127 + ] + ], + [ + [ + 260, + 157 + ], + [ + 256, + 154 + ], + [ + 243, + 141 + ], + [ + 277, + 172 + ], + [ + 247, + 143 + ], + [ + 282, + 176 + ] + ], + [ + [ + 244, + 152 + ], + [ + 249, + 130 + ], + [ + 257, + 147 + ], + [ + 231, + 158 + ], + [ + 252, + 166 + ], + [ + 227, + 179 + ] + ], + [ + [ + 224, + 163 + ], + [ + 226, + 157 + ], + [ + 247, + 165 + ], + [ + 200, + 161 + ], + [ + 244, + 171 + ], + [ + 200, + 168 + ] + ] + ] + }, + { + "task_id": "0001267", + "id": "RH20T_cfg6_task_0038_user_0014_scene_0010_cfg_0006_104122063633#227", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001267/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 235, + 193 + ], + [ + 237, + 174 + ], + [ + 240, + 175 + ], + [ + 229, + 213 + ], + [ + 238, + 192 + ], + [ + 227, + 231 + ] + ], + [ + [ + 261, + 178 + ], + [ + 244, + 183 + ], + [ + 281, + 181 + ], + [ + 238, + 174 + ], + [ + 299, + 176 + ], + [ + 256, + 168 + ] + ], + [ + [ + 208, + 201 + ], + [ + 213, + 214 + ], + [ + 180, + 210 + ], + [ + 235, + 193 + ], + [ + 173, + 196 + ], + [ + 232, + 179 + ] + ], + [ + [ + 234, + 201 + ], + [ + 224, + 221 + ], + [ + 257, + 216 + ], + [ + 211, + 186 + ], + [ + 268, + 195 + ], + [ + 220, + 164 + ] + ] + ] + }, + { + "task_id": "0001269", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_104422070044#217", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001269/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 446, + 170 + ], + [ + 447, + 158 + ], + [ + 462, + 178 + ], + [ + 431, + 161 + ], + [ + 461, + 190 + ], + [ + 431, + 173 + ] + ], + [ + [ + 459, + 165 + ], + [ + 451, + 177 + ], + [ + 470, + 179 + ], + [ + 448, + 151 + ], + [ + 479, + 167 + ], + [ + 456, + 139 + ] + ], + [ + [ + 435, + 161 + ], + [ + 434, + 153 + ], + [ + 440, + 145 + ], + [ + 429, + 176 + ], + [ + 442, + 153 + ], + [ + 430, + 185 + ] + ], + [ + [ + 446, + 169 + ], + [ + 460, + 160 + ], + [ + 459, + 170 + ], + [ + 434, + 167 + ], + [ + 445, + 179 + ], + [ + 421, + 176 + ] + ] + ] + }, + { + "task_id": "0001270", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_104422070044#368", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001270/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 403, + 137 + ], + [ + 416, + 128 + ], + [ + 415, + 147 + ], + [ + 391, + 128 + ], + [ + 401, + 157 + ], + [ + 378, + 137 + ] + ], + [ + [ + 421, + 170 + ], + [ + 404, + 168 + ], + [ + 423, + 153 + ], + [ + 419, + 188 + ], + [ + 441, + 155 + ], + [ + 437, + 190 + ] + ], + [ + [ + 406, + 161 + ], + [ + 395, + 164 + ], + [ + 403, + 144 + ], + [ + 409, + 178 + ], + [ + 413, + 141 + ], + [ + 420, + 174 + ] + ], + [ + [ + 412, + 158 + ], + [ + 411, + 145 + ], + [ + 427, + 165 + ], + [ + 397, + 151 + ], + [ + 427, + 177 + ], + [ + 398, + 163 + ] + ] + ] + }, + { + "task_id": "0001278", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0009_cfg_0003_104122063550#292", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001278/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 330, + 141 + ], + [ + 333, + 128 + ], + [ + 332, + 165 + ], + [ + 329, + 117 + ], + [ + 330, + 175 + ], + [ + 327, + 129 + ] + ], + [ + [ + 336, + 143 + ], + [ + 335, + 137 + ], + [ + 338, + 169 + ], + [ + 335, + 116 + ], + [ + 339, + 173 + ], + [ + 336, + 123 + ] + ], + [ + [ + 352, + 132 + ], + [ + 357, + 108 + ], + [ + 359, + 120 + ], + [ + 345, + 145 + ], + [ + 354, + 141 + ], + [ + 340, + 167 + ] + ], + [ + [ + 326, + 167 + ], + [ + 301, + 167 + ], + [ + 319, + 153 + ], + [ + 332, + 180 + ], + [ + 345, + 153 + ], + [ + 356, + 181 + ] + ] + ] + }, + { + "task_id": "0001280", + "id": "RH20T_cfg5_task_0052_user_0007_scene_0010_cfg_0005_037522062165#199", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab a basketball, release it and shoot it into the basket'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001280/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 256, + 312 + ], + [ + 238, + 300 + ], + [ + 242, + 298 + ], + [ + 271, + 328 + ], + [ + 258, + 310 + ], + [ + 287, + 339 + ] + ], + [ + [ + 278, + 287 + ], + [ + 278, + 275 + ], + [ + 262, + 265 + ], + [ + 295, + 311 + ], + [ + 262, + 278 + ], + [ + 294, + 322 + ] + ], + [ + [ + 248, + 286 + ], + [ + 240, + 259 + ], + [ + 220, + 292 + ], + [ + 276, + 279 + ], + [ + 228, + 318 + ], + [ + 283, + 305 + ] + ], + [ + [ + 265, + 314 + ], + [ + 286, + 291 + ], + [ + 283, + 327 + ], + [ + 247, + 303 + ], + [ + 259, + 353 + ], + [ + 225, + 326 + ] + ] + ] + }, + { + "task_id": "0001281", + "id": "RH20T_cfg1_task_0005_user_0004_scene_0007_cfg_0001_038522062547#225", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001281/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 458, + 170 + ], + [ + 436, + 172 + ], + [ + 451, + 190 + ], + [ + 465, + 148 + ], + [ + 472, + 188 + ], + [ + 487, + 147 + ] + ], + [ + [ + 455, + 124 + ], + [ + 470, + 134 + ], + [ + 443, + 146 + ], + [ + 467, + 100 + ], + [ + 430, + 136 + ], + [ + 453, + 92 + ] + ], + [ + [ + 478, + 164 + ], + [ + 499, + 166 + ], + [ + 461, + 176 + ], + [ + 495, + 153 + ], + [ + 442, + 173 + ], + [ + 475, + 152 + ] + ], + [ + [ + 457, + 148 + ], + [ + 464, + 138 + ], + [ + 463, + 172 + ], + [ + 452, + 125 + ], + [ + 457, + 180 + ], + [ + 446, + 135 + ] + ] + ] + }, + { + "task_id": "0001282", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0001_cfg_0007_104122063633#188", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001282/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 200, + 140 + ], + [ + 182, + 154 + ], + [ + 216, + 153 + ], + [ + 182, + 126 + ], + [ + 234, + 139 + ], + [ + 202, + 110 + ] + ], + [ + [ + 182, + 156 + ], + [ + 188, + 158 + ], + [ + 181, + 179 + ], + [ + 184, + 133 + ], + [ + 174, + 178 + ], + [ + 178, + 130 + ] + ], + [ + [ + 182, + 154 + ], + [ + 188, + 169 + ], + [ + 186, + 171 + ], + [ + 178, + 136 + ], + [ + 181, + 157 + ], + [ + 173, + 122 + ] + ], + [ + [ + 186, + 140 + ], + [ + 168, + 141 + ], + [ + 195, + 158 + ], + [ + 177, + 120 + ], + [ + 214, + 157 + ], + [ + 196, + 119 + ] + ] + ] + }, + { + "task_id": "0001285", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_037522062165#653", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001285/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 233, + 323 + ], + [ + 210, + 306 + ], + [ + 247, + 297 + ], + [ + 219, + 348 + ], + [ + 270, + 315 + ], + [ + 241, + 360 + ] + ], + [ + [ + 236, + 290 + ], + [ + 234, + 277 + ], + [ + 212, + 274 + ], + [ + 260, + 306 + ], + [ + 215, + 286 + ], + [ + 260, + 317 + ] + ], + [ + [ + 254, + 306 + ], + [ + 233, + 328 + ], + [ + 235, + 285 + ], + [ + 273, + 326 + ], + [ + 257, + 264 + ], + [ + 293, + 305 + ] + ], + [ + [ + 217, + 280 + ], + [ + 218, + 299 + ], + [ + 246, + 279 + ], + [ + 186, + 281 + ], + [ + 246, + 258 + ], + [ + 184, + 261 + ] + ] + ] + }, + { + "task_id": "0001287", + "id": "RH20T_cfg2_task_0018_user_0003_scene_0010_cfg_0002_036422060215#217", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001287/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 387, + 261 + ], + [ + 396, + 248 + ], + [ + 403, + 274 + ], + [ + 372, + 248 + ], + [ + 394, + 286 + ], + [ + 364, + 260 + ] + ], + [ + [ + 395, + 252 + ], + [ + 383, + 247 + ], + [ + 391, + 234 + ], + [ + 400, + 270 + ], + [ + 405, + 239 + ], + [ + 414, + 276 + ] + ], + [ + [ + 388, + 278 + ], + [ + 397, + 297 + ], + [ + 373, + 284 + ], + [ + 403, + 272 + ], + [ + 364, + 265 + ], + [ + 394, + 254 + ] + ], + [ + [ + 382, + 275 + ], + [ + 390, + 259 + ], + [ + 399, + 286 + ], + [ + 366, + 264 + ], + [ + 391, + 302 + ], + [ + 359, + 279 + ] + ] + ] + }, + { + "task_id": "0001288", + "id": "RH20T_cfg6_task_0007_user_0014_scene_0007_cfg_0006_104122060811#208", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001288/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 434, + 138 + ], + [ + 434, + 152 + ], + [ + 449, + 137 + ], + [ + 420, + 139 + ], + [ + 449, + 122 + ], + [ + 420, + 125 + ] + ], + [ + [ + 419, + 141 + ], + [ + 429, + 149 + ], + [ + 424, + 147 + ], + [ + 415, + 135 + ], + [ + 414, + 138 + ], + [ + 406, + 127 + ] + ], + [ + [ + 426, + 131 + ], + [ + 432, + 142 + ], + [ + 439, + 129 + ], + [ + 413, + 133 + ], + [ + 434, + 118 + ], + [ + 408, + 121 + ] + ], + [ + [ + 423, + 129 + ], + [ + 433, + 122 + ], + [ + 433, + 130 + ], + [ + 413, + 129 + ], + [ + 423, + 137 + ], + [ + 404, + 137 + ] + ] + ] + }, + { + "task_id": "0001291", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0003_cfg_0004_104122062823#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001291/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 307, + 8 + ], + [ + 307, + 0 + ], + [ + 342, + 15 + ], + [ + 273, + 1 + ], + [ + 339, + 49 + ], + [ + 274, + 34 + ] + ], + [ + [ + 317, + 22 + ], + [ + 323, + 58 + ], + [ + 347, + 1 + ], + [ + 290, + 41 + ], + [ + 340, + 0 + ], + [ + 285, + 8 + ] + ], + [ + [ + 285, + 0 + ], + [ + 316, + 0 + ], + [ + 310, + 0 + ], + [ + 263, + 0 + ], + [ + 274, + 0 + ], + [ + 230, + 0 + ] + ], + [ + [ + 317, + 15 + ], + [ + 298, + 0 + ], + [ + 289, + 32 + ], + [ + 343, + 0 + ], + [ + 309, + 67 + ], + [ + 360, + 32 + ] + ] + ] + }, + { + "task_id": "0001293", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0003_cfg_0004_104122062823#251", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001293/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 355, + 0 + ], + [ + 349, + 0 + ], + [ + 396, + 0 + ], + [ + 315, + 0 + ], + [ + 398, + 0 + ], + [ + 324, + 0 + ] + ], + [ + [ + 331, + 0 + ], + [ + 328, + 0 + ], + [ + 368, + 0 + ], + [ + 295, + 0 + ], + [ + 368, + 20 + ], + [ + 299, + 15 + ] + ], + [ + [ + 307, + 0 + ], + [ + 268, + 0 + ], + [ + 310, + 0 + ], + [ + 304, + 0 + ], + [ + 346, + 0 + ], + [ + 340, + 6 + ] + ], + [ + [ + 299, + 15 + ], + [ + 309, + 0 + ], + [ + 292, + 40 + ], + [ + 307, + 0 + ], + [ + 282, + 64 + ], + [ + 295, + 13 + ] + ] + ] + }, + { + "task_id": "0001296", + "id": "RH20T_cfg4_task_0057_user_0014_scene_0008_cfg_0004_036422060909#120", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001296/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 322, + 140 + ], + [ + 297, + 136 + ], + [ + 331, + 156 + ], + [ + 313, + 125 + ], + [ + 355, + 159 + ], + [ + 336, + 129 + ] + ], + [ + [ + 343, + 166 + ], + [ + 343, + 149 + ], + [ + 345, + 143 + ], + [ + 341, + 189 + ], + [ + 345, + 159 + ], + [ + 341, + 203 + ] + ], + [ + [ + 370, + 176 + ], + [ + 397, + 172 + ], + [ + 366, + 148 + ], + [ + 373, + 202 + ], + [ + 339, + 152 + ], + [ + 346, + 206 + ] + ], + [ + [ + 322, + 169 + ], + [ + 341, + 159 + ], + [ + 323, + 147 + ], + [ + 320, + 193 + ], + [ + 305, + 156 + ], + [ + 301, + 201 + ] + ] + ] + }, + { + "task_id": "0001300", + "id": "RH20T_cfg1_task_0108_user_0014_scene_0004_cfg_0001_035622060973#134", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Sharpen the pencil with a pencil sharpener'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001300/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 259, + 192 + ], + [ + 263, + 198 + ], + [ + 278, + 202 + ], + [ + 240, + 183 + ], + [ + 274, + 196 + ], + [ + 234, + 177 + ] + ], + [ + [ + 275, + 189 + ], + [ + 275, + 169 + ], + [ + 254, + 190 + ], + [ + 296, + 187 + ], + [ + 254, + 211 + ], + [ + 297, + 208 + ] + ], + [ + [ + 278, + 175 + ], + [ + 263, + 191 + ], + [ + 262, + 160 + ], + [ + 294, + 190 + ], + [ + 278, + 143 + ], + [ + 309, + 174 + ] + ], + [ + [ + 272, + 194 + ], + [ + 270, + 183 + ], + [ + 250, + 190 + ], + [ + 294, + 198 + ], + [ + 252, + 201 + ], + [ + 295, + 208 + ] + ] + ] + }, + { + "task_id": "0001302", + "id": "RH20T_cfg7_task_0046_user_0014_scene_0001_cfg_0007_104122061330#103", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001302/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 123 + ], + [ + 333, + 103 + ], + [ + 321, + 142 + ], + [ + 345, + 105 + ], + [ + 322, + 160 + ], + [ + 345, + 123 + ] + ], + [ + [ + 319, + 114 + ], + [ + 343, + 101 + ], + [ + 324, + 136 + ], + [ + 314, + 89 + ], + [ + 301, + 150 + ], + [ + 288, + 103 + ] + ], + [ + [ + 337, + 133 + ], + [ + 313, + 143 + ], + [ + 337, + 115 + ], + [ + 337, + 152 + ], + [ + 361, + 103 + ], + [ + 363, + 141 + ] + ], + [ + [ + 319, + 96 + ], + [ + 344, + 92 + ], + [ + 329, + 120 + ], + [ + 310, + 70 + ], + [ + 306, + 122 + ], + [ + 286, + 74 + ] + ] + ] + }, + { + "task_id": "0001303", + "id": "RH20T_cfg2_task_0077_user_0014_scene_0006_cfg_0002_105422061350#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001303/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 181, + 68 + ], + [ + 205, + 72 + ], + [ + 183, + 46 + ], + [ + 178, + 89 + ], + [ + 157, + 41 + ], + [ + 154, + 86 + ] + ], + [ + [ + 173, + 73 + ], + [ + 172, + 57 + ], + [ + 179, + 54 + ], + [ + 167, + 93 + ], + [ + 180, + 69 + ], + [ + 168, + 107 + ] + ], + [ + [ + 198, + 54 + ], + [ + 177, + 64 + ], + [ + 209, + 78 + ], + [ + 187, + 29 + ], + [ + 229, + 67 + ], + [ + 208, + 20 + ] + ], + [ + [ + 198, + 85 + ], + [ + 219, + 97 + ], + [ + 208, + 66 + ], + [ + 188, + 105 + ], + [ + 186, + 53 + ], + [ + 166, + 93 + ] + ] + ] + }, + { + "task_id": "0001304", + "id": "RH20T_cfg2_task_0077_user_0014_scene_0006_cfg_0002_105422061350#409", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001304/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 179, + 32 + ], + [ + 198, + 49 + ], + [ + 194, + 17 + ], + [ + 164, + 47 + ], + [ + 175, + 0 + ], + [ + 144, + 30 + ] + ], + [ + [ + 178, + 27 + ], + [ + 200, + 24 + ], + [ + 189, + 32 + ], + [ + 165, + 22 + ], + [ + 168, + 36 + ], + [ + 141, + 26 + ] + ], + [ + [ + 189, + 56 + ], + [ + 188, + 39 + ], + [ + 195, + 39 + ], + [ + 183, + 74 + ], + [ + 196, + 55 + ], + [ + 185, + 90 + ] + ], + [ + [ + 209, + 63 + ], + [ + 208, + 87 + ], + [ + 227, + 60 + ], + [ + 192, + 66 + ], + [ + 228, + 35 + ], + [ + 192, + 42 + ] + ] + ] + }, + { + "task_id": "0001305", + "id": "RH20T_cfg2_task_0077_user_0014_scene_0006_cfg_0002_105422061350#655", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001305/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 162, + 88 + ], + [ + 158, + 71 + ], + [ + 172, + 69 + ], + [ + 152, + 107 + ], + [ + 174, + 85 + ], + [ + 155, + 122 + ] + ], + [ + [ + 158, + 70 + ], + [ + 164, + 79 + ], + [ + 165, + 47 + ], + [ + 152, + 94 + ], + [ + 159, + 36 + ], + [ + 145, + 86 + ] + ], + [ + [ + 192, + 82 + ], + [ + 207, + 87 + ], + [ + 197, + 58 + ], + [ + 187, + 107 + ], + [ + 180, + 51 + ], + [ + 170, + 102 + ] + ], + [ + [ + 175, + 67 + ], + [ + 188, + 49 + ], + [ + 197, + 82 + ], + [ + 153, + 52 + ], + [ + 184, + 100 + ], + [ + 138, + 71 + ] + ] + ] + }, + { + "task_id": "0001307", + "id": "RH20T_cfg2_task_0077_user_0014_scene_0006_cfg_0002_105422061350#1070", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001307/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 143, + 120 + ], + [ + 123, + 122 + ], + [ + 139, + 91 + ], + [ + 146, + 148 + ], + [ + 160, + 88 + ], + [ + 167, + 146 + ] + ], + [ + [ + 168, + 120 + ], + [ + 148, + 113 + ], + [ + 151, + 127 + ], + [ + 184, + 113 + ], + [ + 172, + 135 + ], + [ + 205, + 119 + ] + ], + [ + [ + 145, + 117 + ], + [ + 138, + 100 + ], + [ + 158, + 98 + ], + [ + 131, + 137 + ], + [ + 163, + 114 + ], + [ + 138, + 152 + ] + ], + [ + [ + 148, + 98 + ], + [ + 131, + 83 + ], + [ + 134, + 119 + ], + [ + 162, + 76 + ], + [ + 152, + 135 + ], + [ + 180, + 91 + ] + ] + ] + }, + { + "task_id": "0001309", + "id": "RH20T_cfg7_task_0016_user_0020_scene_0004_cfg_0007_104122060811#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001309/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 137 + ], + [ + 363, + 135 + ], + [ + 346, + 141 + ], + [ + 350, + 133 + ], + [ + 332, + 144 + ], + [ + 335, + 136 + ] + ], + [ + [ + 345, + 131 + ], + [ + 341, + 144 + ], + [ + 337, + 136 + ], + [ + 352, + 127 + ], + [ + 341, + 122 + ], + [ + 356, + 114 + ] + ], + [ + [ + 341, + 129 + ], + [ + 343, + 143 + ], + [ + 344, + 133 + ], + [ + 338, + 124 + ], + [ + 343, + 119 + ], + [ + 336, + 109 + ] + ], + [ + [ + 359, + 138 + ], + [ + 363, + 152 + ], + [ + 371, + 138 + ], + [ + 346, + 139 + ], + [ + 366, + 125 + ], + [ + 342, + 125 + ] + ] + ] + }, + { + "task_id": "0001310", + "id": "RH20T_cfg3_task_0071_user_0010_scene_0008_cfg_0003_104122062823#167", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001310/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 175, + 152 + ], + [ + 196, + 161 + ], + [ + 168, + 187 + ], + [ + 181, + 117 + ], + [ + 143, + 179 + ], + [ + 158, + 104 + ] + ], + [ + [ + 150, + 111 + ], + [ + 149, + 137 + ], + [ + 111, + 101 + ], + [ + 187, + 121 + ], + [ + 115, + 77 + ], + [ + 186, + 97 + ] + ], + [ + [ + 178, + 147 + ], + [ + 169, + 126 + ], + [ + 205, + 165 + ], + [ + 152, + 130 + ], + [ + 211, + 182 + ], + [ + 161, + 149 + ] + ], + [ + [ + 199, + 124 + ], + [ + 226, + 133 + ], + [ + 184, + 93 + ], + [ + 212, + 151 + ], + [ + 157, + 86 + ], + [ + 187, + 143 + ] + ] + ] + }, + { + "task_id": "0001311", + "id": "RH20T_cfg3_task_0071_user_0010_scene_0008_cfg_0003_104122062823#270", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001311/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 213, + 132 + ], + [ + 204, + 109 + ], + [ + 247, + 118 + ], + [ + 179, + 147 + ], + [ + 253, + 139 + ], + [ + 189, + 166 + ] + ], + [ + [ + 209, + 139 + ], + [ + 229, + 117 + ], + [ + 206, + 119 + ], + [ + 211, + 156 + ], + [ + 183, + 144 + ], + [ + 190, + 178 + ] + ], + [ + [ + 178, + 149 + ], + [ + 190, + 156 + ], + [ + 190, + 178 + ], + [ + 167, + 120 + ], + [ + 176, + 173 + ], + [ + 152, + 110 + ] + ], + [ + [ + 196, + 144 + ], + [ + 190, + 122 + ], + [ + 225, + 160 + ], + [ + 168, + 129 + ], + [ + 228, + 179 + ], + [ + 174, + 150 + ] + ] + ] + }, + { + "task_id": "0001312", + "id": "RH20T_cfg1_task_0016_user_0001_scene_0004_cfg_0001_750612070853#94", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001312/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 273, + 141 + ], + [ + 276, + 173 + ], + [ + 240, + 129 + ], + [ + 302, + 151 + ], + [ + 239, + 98 + ], + [ + 299, + 122 + ] + ], + [ + [ + 213, + 122 + ], + [ + 180, + 122 + ], + [ + 202, + 82 + ], + [ + 223, + 159 + ], + [ + 239, + 81 + ], + [ + 257, + 161 + ] + ], + [ + [ + 246, + 143 + ], + [ + 242, + 108 + ], + [ + 218, + 158 + ], + [ + 270, + 130 + ], + [ + 224, + 192 + ], + [ + 273, + 161 + ] + ], + [ + [ + 202, + 155 + ], + [ + 240, + 157 + ], + [ + 201, + 127 + ], + [ + 203, + 186 + ], + [ + 162, + 124 + ], + [ + 161, + 185 + ] + ] + ] + }, + { + "task_id": "0001314", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0010_cfg_0001_750612070851#212", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001314/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 362, + 219 + ], + [ + 355, + 240 + ], + [ + 366, + 194 + ], + [ + 357, + 248 + ], + [ + 372, + 170 + ], + [ + 364, + 225 + ] + ], + [ + [ + 379, + 139 + ], + [ + 382, + 146 + ], + [ + 344, + 148 + ], + [ + 413, + 129 + ], + [ + 344, + 141 + ], + [ + 408, + 124 + ] + ], + [ + [ + 362, + 188 + ], + [ + 365, + 161 + ], + [ + 327, + 192 + ], + [ + 395, + 183 + ], + [ + 326, + 217 + ], + [ + 391, + 208 + ] + ], + [ + [ + 372, + 165 + ], + [ + 407, + 178 + ], + [ + 358, + 196 + ], + [ + 384, + 137 + ], + [ + 323, + 183 + ], + [ + 351, + 125 + ] + ] + ] + }, + { + "task_id": "0001316", + "id": "RH20T_cfg6_task_0104_user_0020_scene_0006_cfg_0006_104122061330#202", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate back after holding it down'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001316/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 378, + 101 + ], + [ + 370, + 115 + ], + [ + 368, + 117 + ], + [ + 388, + 84 + ], + [ + 375, + 104 + ], + [ + 396, + 72 + ] + ], + [ + [ + 360, + 107 + ], + [ + 339, + 105 + ], + [ + 360, + 130 + ], + [ + 361, + 83 + ], + [ + 382, + 132 + ], + [ + 383, + 84 + ] + ], + [ + [ + 377, + 111 + ], + [ + 386, + 93 + ], + [ + 387, + 124 + ], + [ + 366, + 97 + ], + [ + 379, + 143 + ], + [ + 357, + 116 + ] + ], + [ + [ + 389, + 122 + ], + [ + 369, + 108 + ], + [ + 376, + 140 + ], + [ + 404, + 104 + ], + [ + 395, + 153 + ], + [ + 424, + 118 + ] + ] + ] + }, + { + "task_id": "0001320", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_105422061350#523", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001320/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 209, + 69 + ], + [ + 189, + 76 + ], + [ + 198, + 85 + ], + [ + 221, + 52 + ], + [ + 216, + 78 + ], + [ + 240, + 46 + ] + ], + [ + [ + 172, + 74 + ], + [ + 190, + 96 + ], + [ + 150, + 90 + ], + [ + 193, + 58 + ], + [ + 131, + 68 + ], + [ + 175, + 36 + ] + ], + [ + [ + 209, + 100 + ], + [ + 184, + 101 + ], + [ + 208, + 74 + ], + [ + 210, + 126 + ], + [ + 234, + 73 + ], + [ + 236, + 126 + ] + ], + [ + [ + 194, + 74 + ], + [ + 195, + 55 + ], + [ + 207, + 58 + ], + [ + 180, + 92 + ], + [ + 205, + 75 + ], + [ + 179, + 109 + ] + ] + ] + }, + { + "task_id": "0001323", + "id": "RH20T_cfg5_task_0081_user_0007_scene_0003_cfg_0005_037522062165#235", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001323/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 278, + 273 + ], + [ + 257, + 260 + ], + [ + 284, + 249 + ], + [ + 272, + 295 + ], + [ + 306, + 262 + ], + [ + 293, + 309 + ] + ], + [ + [ + 263, + 283 + ], + [ + 258, + 256 + ], + [ + 279, + 275 + ], + [ + 247, + 291 + ], + [ + 285, + 304 + ], + [ + 252, + 319 + ] + ], + [ + [ + 277, + 281 + ], + [ + 275, + 270 + ], + [ + 253, + 267 + ], + [ + 302, + 295 + ], + [ + 256, + 277 + ], + [ + 303, + 303 + ] + ], + [ + [ + 271, + 245 + ], + [ + 284, + 226 + ], + [ + 256, + 222 + ], + [ + 285, + 266 + ], + [ + 240, + 242 + ], + [ + 272, + 286 + ] + ] + ] + }, + { + "task_id": "0001324", + "id": "RH20T_cfg5_task_0002_user_0010_scene_0001_cfg_0005_037522062165#80", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001324/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 291, + 246 + ], + [ + 309, + 249 + ], + [ + 276, + 225 + ], + [ + 305, + 267 + ], + [ + 260, + 224 + ], + [ + 289, + 264 + ] + ], + [ + [ + 278, + 240 + ], + [ + 290, + 213 + ], + [ + 276, + 237 + ], + [ + 280, + 243 + ], + [ + 265, + 263 + ], + [ + 267, + 272 + ] + ], + [ + [ + 280, + 234 + ], + [ + 294, + 249 + ], + [ + 285, + 253 + ], + [ + 275, + 216 + ], + [ + 268, + 237 + ], + [ + 260, + 200 + ] + ], + [ + [ + 290, + 264 + ], + [ + 277, + 283 + ], + [ + 289, + 283 + ], + [ + 291, + 244 + ], + [ + 301, + 266 + ], + [ + 303, + 227 + ] + ] + ] + }, + { + "task_id": "0001325", + "id": "RH20T_cfg1_task_0120_user_0014_scene_0002_cfg_0001_039422060546#99", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001325/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 288, + 137 + ], + [ + 256, + 133 + ], + [ + 292, + 144 + ], + [ + 285, + 132 + ], + [ + 326, + 148 + ], + [ + 314, + 136 + ] + ], + [ + [ + 304, + 133 + ], + [ + 299, + 102 + ], + [ + 317, + 144 + ], + [ + 291, + 123 + ], + [ + 321, + 174 + ], + [ + 296, + 151 + ] + ], + [ + [ + 275, + 115 + ], + [ + 268, + 119 + ], + [ + 242, + 116 + ], + [ + 309, + 113 + ], + [ + 250, + 112 + ], + [ + 313, + 110 + ] + ], + [ + [ + 328, + 131 + ], + [ + 357, + 127 + ], + [ + 338, + 116 + ], + [ + 320, + 145 + ], + [ + 305, + 120 + ], + [ + 291, + 149 + ] + ] + ] + }, + { + "task_id": "0001326", + "id": "RH20T_cfg1_task_0120_user_0014_scene_0002_cfg_0001_039422060546#275", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001326/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 310, + 112 + ], + [ + 334, + 116 + ], + [ + 297, + 144 + ], + [ + 322, + 82 + ], + [ + 276, + 139 + ], + [ + 300, + 80 + ] + ], + [ + [ + 352, + 152 + ], + [ + 324, + 172 + ], + [ + 338, + 130 + ], + [ + 367, + 176 + ], + [ + 366, + 110 + ], + [ + 397, + 154 + ] + ], + [ + [ + 348, + 159 + ], + [ + 366, + 162 + ], + [ + 319, + 164 + ], + [ + 376, + 155 + ], + [ + 305, + 161 + ], + [ + 359, + 153 + ] + ], + [ + [ + 317, + 149 + ], + [ + 316, + 127 + ], + [ + 343, + 168 + ], + [ + 293, + 132 + ], + [ + 343, + 187 + ], + [ + 296, + 153 + ] + ] + ] + }, + { + "task_id": "0001327", + "id": "RH20T_cfg1_task_0120_user_0014_scene_0002_cfg_0001_039422060546#331", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001327/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 327, + 118 + ], + [ + 331, + 99 + ], + [ + 296, + 104 + ], + [ + 357, + 132 + ], + [ + 290, + 124 + ], + [ + 354, + 153 + ] + ], + [ + [ + 339, + 152 + ], + [ + 332, + 139 + ], + [ + 363, + 178 + ], + [ + 317, + 127 + ], + [ + 367, + 187 + ], + [ + 324, + 139 + ] + ], + [ + [ + 317, + 149 + ], + [ + 316, + 126 + ], + [ + 351, + 154 + ], + [ + 284, + 144 + ], + [ + 350, + 174 + ], + [ + 287, + 163 + ] + ], + [ + [ + 319, + 133 + ], + [ + 314, + 103 + ], + [ + 308, + 121 + ], + [ + 328, + 144 + ], + [ + 314, + 155 + ], + [ + 333, + 174 + ] + ] + ] + }, + { + "task_id": "0001328", + "id": "RH20T_cfg1_task_0120_user_0014_scene_0002_cfg_0001_039422060546#405", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001328/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 149 + ], + [ + 314, + 126 + ], + [ + 345, + 163 + ], + [ + 287, + 136 + ], + [ + 345, + 183 + ], + [ + 289, + 156 + ] + ], + [ + [ + 349, + 139 + ], + [ + 378, + 127 + ], + [ + 337, + 107 + ], + [ + 361, + 170 + ], + [ + 307, + 119 + ], + [ + 332, + 183 + ] + ], + [ + [ + 336, + 141 + ], + [ + 326, + 171 + ], + [ + 311, + 149 + ], + [ + 364, + 133 + ], + [ + 321, + 122 + ], + [ + 372, + 104 + ] + ], + [ + [ + 302, + 145 + ], + [ + 286, + 168 + ], + [ + 275, + 135 + ], + [ + 330, + 156 + ], + [ + 290, + 115 + ], + [ + 344, + 135 + ] + ] + ] + }, + { + "task_id": "0001329", + "id": "RH20T_cfg1_task_0120_user_0014_scene_0002_cfg_0001_039422060546#450", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001329/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 302, + 135 + ], + [ + 318, + 164 + ], + [ + 310, + 128 + ], + [ + 294, + 142 + ], + [ + 292, + 97 + ], + [ + 278, + 115 + ] + ], + [ + [ + 316, + 123 + ], + [ + 319, + 158 + ], + [ + 284, + 131 + ], + [ + 347, + 116 + ], + [ + 279, + 93 + ], + [ + 345, + 81 + ] + ], + [ + [ + 290, + 151 + ], + [ + 291, + 115 + ], + [ + 252, + 149 + ], + [ + 327, + 152 + ], + [ + 250, + 186 + ], + [ + 326, + 189 + ] + ], + [ + [ + 314, + 150 + ], + [ + 315, + 128 + ], + [ + 324, + 178 + ], + [ + 305, + 124 + ], + [ + 322, + 197 + ], + [ + 305, + 144 + ] + ] + ] + }, + { + "task_id": "0001336", + "id": "RH20T_cfg6_task_0089_user_0020_scene_0009_cfg_0006_104122060811#156", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push an object with obstacles'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001336/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 331, + 173 + ], + [ + 320, + 179 + ], + [ + 337, + 188 + ], + [ + 325, + 159 + ], + [ + 348, + 182 + ], + [ + 336, + 154 + ] + ], + [ + [ + 343, + 172 + ], + [ + 358, + 172 + ], + [ + 342, + 187 + ], + [ + 344, + 157 + ], + [ + 327, + 186 + ], + [ + 330, + 156 + ] + ], + [ + [ + 336, + 169 + ], + [ + 335, + 163 + ], + [ + 350, + 163 + ], + [ + 322, + 175 + ], + [ + 352, + 169 + ], + [ + 324, + 180 + ] + ], + [ + [ + 327, + 183 + ], + [ + 313, + 178 + ], + [ + 327, + 172 + ], + [ + 327, + 194 + ], + [ + 341, + 177 + ], + [ + 341, + 199 + ] + ] + ] + }, + { + "task_id": "0001337", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0008_cfg_0005_105422061350#428", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001337/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 185, + 102 + ], + [ + 204, + 124 + ], + [ + 205, + 86 + ], + [ + 167, + 118 + ], + [ + 185, + 62 + ], + [ + 148, + 96 + ] + ], + [ + [ + 156, + 96 + ], + [ + 179, + 109 + ], + [ + 173, + 73 + ], + [ + 139, + 119 + ], + [ + 151, + 61 + ], + [ + 117, + 106 + ] + ], + [ + [ + 148, + 129 + ], + [ + 123, + 144 + ], + [ + 133, + 105 + ], + [ + 162, + 151 + ], + [ + 159, + 91 + ], + [ + 186, + 137 + ] + ], + [ + [ + 147, + 106 + ], + [ + 140, + 87 + ], + [ + 167, + 88 + ], + [ + 127, + 126 + ], + [ + 172, + 105 + ], + [ + 135, + 142 + ] + ] + ] + }, + { + "task_id": "0001339", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0008_cfg_0005_105422061350#996", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001339/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 170, + 45 + ], + [ + 197, + 47 + ], + [ + 161, + 32 + ], + [ + 179, + 58 + ], + [ + 132, + 29 + ], + [ + 149, + 56 + ] + ], + [ + [ + 172, + 57 + ], + [ + 168, + 37 + ], + [ + 197, + 48 + ], + [ + 146, + 66 + ], + [ + 199, + 66 + ], + [ + 151, + 84 + ] + ], + [ + [ + 169, + 40 + ], + [ + 181, + 58 + ], + [ + 182, + 19 + ], + [ + 157, + 60 + ], + [ + 170, + 2 + ], + [ + 146, + 43 + ] + ], + [ + [ + 193, + 54 + ], + [ + 164, + 49 + ], + [ + 189, + 78 + ], + [ + 196, + 31 + ], + [ + 217, + 83 + ], + [ + 223, + 36 + ] + ] + ] + }, + { + "task_id": "0001340", + "id": "RH20T_cfg1_task_0002_user_0014_scene_0004_cfg_0001_035622060973#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001340/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 212, + 178 + ], + [ + 216, + 199 + ], + [ + 211, + 179 + ], + [ + 212, + 178 + ], + [ + 207, + 158 + ], + [ + 208, + 158 + ] + ], + [ + [ + 207, + 150 + ], + [ + 225, + 143 + ], + [ + 214, + 143 + ], + [ + 200, + 157 + ], + [ + 195, + 151 + ], + [ + 181, + 165 + ] + ], + [ + [ + 226, + 156 + ], + [ + 224, + 143 + ], + [ + 215, + 143 + ], + [ + 238, + 171 + ], + [ + 217, + 155 + ], + [ + 239, + 182 + ] + ], + [ + [ + 229, + 143 + ], + [ + 219, + 151 + ], + [ + 238, + 162 + ], + [ + 219, + 124 + ], + [ + 249, + 154 + ], + [ + 229, + 115 + ] + ] + ] + }, + { + "task_id": "0001346", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0001_cfg_0007_104122061018#190", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001346/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 270, + 102 + ], + [ + 263, + 101 + ], + [ + 278, + 131 + ], + [ + 262, + 72 + ], + [ + 286, + 133 + ], + [ + 269, + 70 + ] + ], + [ + [ + 279, + 81 + ], + [ + 305, + 73 + ], + [ + 277, + 56 + ], + [ + 280, + 103 + ], + [ + 248, + 64 + ], + [ + 255, + 111 + ] + ], + [ + [ + 252, + 109 + ], + [ + 243, + 91 + ], + [ + 278, + 92 + ], + [ + 227, + 126 + ], + [ + 289, + 112 + ], + [ + 236, + 147 + ] + ], + [ + [ + 281, + 75 + ], + [ + 306, + 64 + ], + [ + 270, + 78 + ], + [ + 293, + 72 + ], + [ + 245, + 89 + ], + [ + 265, + 84 + ] + ] + ] + }, + { + "task_id": "0001347", + "id": "RH20T_cfg1_task_0083_user_0014_scene_0003_cfg_0001_750612070853#156", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Swat with a flyswatter'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001347/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 392, + 358 + ], + [ + 387, + 326 + ], + [ + 330, + 353 + ], + [ + 458, + 360 + ], + [ + 338, + 360 + ], + [ + 456, + 360 + ] + ], + [ + [ + 372, + 360 + ], + [ + 378, + 320 + ], + [ + 414, + 360 + ], + [ + 339, + 360 + ], + [ + 411, + 360 + ], + [ + 328, + 360 + ] + ], + [ + [ + 427, + 360 + ], + [ + 416, + 360 + ], + [ + 476, + 360 + ], + [ + 387, + 360 + ], + [ + 487, + 352 + ], + [ + 397, + 315 + ] + ], + [ + [ + 431, + 360 + ], + [ + 453, + 360 + ], + [ + 381, + 360 + ], + [ + 493, + 360 + ], + [ + 351, + 360 + ], + [ + 472, + 360 + ] + ] + ] + }, + { + "task_id": "0001348", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_104122061850#822", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001348/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 226, + 156 + ], + [ + 230, + 142 + ], + [ + 210, + 182 + ], + [ + 240, + 132 + ], + [ + 208, + 194 + ], + [ + 236, + 145 + ] + ], + [ + [ + 235, + 147 + ], + [ + 203, + 157 + ], + [ + 251, + 174 + ], + [ + 220, + 121 + ], + [ + 280, + 163 + ], + [ + 249, + 113 + ] + ], + [ + [ + 234, + 166 + ], + [ + 237, + 152 + ], + [ + 232, + 135 + ], + [ + 236, + 194 + ], + [ + 228, + 149 + ], + [ + 233, + 211 + ] + ], + [ + [ + 237, + 126 + ], + [ + 233, + 155 + ], + [ + 266, + 138 + ], + [ + 206, + 113 + ], + [ + 269, + 112 + ], + [ + 211, + 86 + ] + ] + ] + }, + { + "task_id": "0001352", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0002_cfg_0001_035622060973#723", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001352/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 275, + 269 + ], + [ + 275, + 259 + ], + [ + 262, + 251 + ], + [ + 289, + 287 + ], + [ + 263, + 260 + ], + [ + 289, + 295 + ] + ], + [ + [ + 268, + 279 + ], + [ + 262, + 256 + ], + [ + 245, + 287 + ], + [ + 291, + 271 + ], + [ + 251, + 311 + ], + [ + 298, + 294 + ] + ], + [ + [ + 285, + 270 + ], + [ + 284, + 271 + ], + [ + 304, + 284 + ], + [ + 267, + 257 + ], + [ + 306, + 284 + ], + [ + 268, + 256 + ] + ], + [ + [ + 256, + 252 + ], + [ + 271, + 251 + ], + [ + 237, + 263 + ], + [ + 273, + 242 + ], + [ + 223, + 264 + ], + [ + 258, + 244 + ] + ] + ] + }, + { + "task_id": "0001353", + "id": "RH20T_cfg4_task_0120_user_0007_scene_0004_cfg_0004_104422070011#114", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001353/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 151 + ], + [ + 298, + 130 + ], + [ + 337, + 137 + ], + [ + 291, + 165 + ], + [ + 353, + 158 + ], + [ + 307, + 188 + ] + ], + [ + [ + 327, + 142 + ], + [ + 329, + 130 + ], + [ + 310, + 164 + ], + [ + 343, + 121 + ], + [ + 308, + 173 + ], + [ + 339, + 132 + ] + ], + [ + [ + 303, + 154 + ], + [ + 283, + 152 + ], + [ + 317, + 168 + ], + [ + 287, + 139 + ], + [ + 338, + 170 + ], + [ + 309, + 141 + ] + ], + [ + [ + 323, + 141 + ], + [ + 316, + 114 + ], + [ + 350, + 134 + ], + [ + 297, + 148 + ], + [ + 357, + 160 + ], + [ + 304, + 173 + ] + ] + ] + }, + { + "task_id": "0001354", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_039422060546#315", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001354/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 345, + 310 + ], + [ + 309, + 305 + ], + [ + 341, + 328 + ], + [ + 348, + 295 + ], + [ + 381, + 335 + ], + [ + 382, + 300 + ] + ], + [ + [ + 331, + 267 + ], + [ + 331, + 250 + ], + [ + 318, + 298 + ], + [ + 343, + 238 + ], + [ + 319, + 311 + ], + [ + 343, + 254 + ] + ], + [ + [ + 326, + 256 + ], + [ + 300, + 258 + ], + [ + 341, + 239 + ], + [ + 310, + 274 + ], + [ + 368, + 236 + ], + [ + 339, + 272 + ] + ], + [ + [ + 305, + 295 + ], + [ + 325, + 288 + ], + [ + 287, + 264 + ], + [ + 323, + 326 + ], + [ + 270, + 272 + ], + [ + 304, + 331 + ] + ] + ] + }, + { + "task_id": "0001356", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_039422060546#1076", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001356/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 359, + 176 + ], + [ + 379, + 191 + ], + [ + 365, + 147 + ], + [ + 353, + 204 + ], + [ + 345, + 134 + ], + [ + 336, + 189 + ] + ], + [ + [ + 344, + 200 + ], + [ + 339, + 180 + ], + [ + 333, + 226 + ], + [ + 354, + 176 + ], + [ + 339, + 244 + ], + [ + 358, + 194 + ] + ], + [ + [ + 371, + 179 + ], + [ + 342, + 170 + ], + [ + 359, + 210 + ], + [ + 382, + 149 + ], + [ + 389, + 221 + ], + [ + 413, + 158 + ] + ], + [ + [ + 361, + 206 + ], + [ + 368, + 179 + ], + [ + 362, + 217 + ], + [ + 360, + 193 + ], + [ + 355, + 243 + ], + [ + 352, + 223 + ] + ] + ] + }, + { + "task_id": "0001359", + "id": "RH20T_cfg5_task_0033_user_0010_scene_0010_cfg_0005_104422070044#152", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001359/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 416, + 209 + ], + [ + 431, + 200 + ], + [ + 427, + 227 + ], + [ + 405, + 192 + ], + [ + 412, + 236 + ], + [ + 389, + 201 + ] + ], + [ + [ + 394, + 200 + ], + [ + 404, + 187 + ], + [ + 391, + 211 + ], + [ + 396, + 189 + ], + [ + 381, + 224 + ], + [ + 387, + 201 + ] + ], + [ + [ + 408, + 207 + ], + [ + 407, + 193 + ], + [ + 422, + 217 + ], + [ + 394, + 196 + ], + [ + 421, + 230 + ], + [ + 395, + 209 + ] + ], + [ + [ + 391, + 203 + ], + [ + 375, + 203 + ], + [ + 385, + 186 + ], + [ + 397, + 221 + ], + [ + 400, + 187 + ], + [ + 412, + 220 + ] + ] + ] + }, + { + "task_id": "0001362", + "id": "RH20T_cfg1_task_0108_user_0014_scene_0004_cfg_0001_819112071037#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Sharpen the pencil with a pencil sharpener'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001362/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 551, + 102 + ], + [ + 563, + 31 + ], + [ + 529, + 116 + ], + [ + 569, + 90 + ], + [ + 518, + 185 + ], + [ + 558, + 148 + ] + ], + [ + [ + 540, + 47 + ], + [ + 472, + 39 + ], + [ + 535, + 64 + ], + [ + 544, + 33 + ], + [ + 620, + 76 + ], + [ + 611, + 40 + ] + ], + [ + [ + 501, + 36 + ], + [ + 486, + 58 + ], + [ + 458, + 80 + ], + [ + 553, + 0 + ], + [ + 472, + 60 + ], + [ + 561, + 0 + ] + ], + [ + [ + 555, + 71 + ], + [ + 553, + 48 + ], + [ + 489, + 85 + ], + [ + 631, + 54 + ], + [ + 483, + 112 + ], + [ + 640, + 83 + ] + ] + ] + }, + { + "task_id": "0001363", + "id": "RH20T_cfg3_task_0077_user_0010_scene_0009_cfg_0003_104122063550#75", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001363/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 313, + 181 + ], + [ + 323, + 165 + ], + [ + 312, + 165 + ], + [ + 314, + 196 + ], + [ + 300, + 182 + ], + [ + 303, + 212 + ] + ], + [ + [ + 341, + 176 + ], + [ + 335, + 182 + ], + [ + 327, + 195 + ], + [ + 355, + 156 + ], + [ + 333, + 188 + ], + [ + 359, + 151 + ] + ], + [ + [ + 339, + 175 + ], + [ + 336, + 156 + ], + [ + 343, + 189 + ], + [ + 334, + 159 + ], + [ + 346, + 209 + ], + [ + 337, + 180 + ] + ], + [ + [ + 321, + 169 + ], + [ + 323, + 158 + ], + [ + 346, + 169 + ], + [ + 297, + 169 + ], + [ + 342, + 179 + ], + [ + 296, + 179 + ] + ] + ] + }, + { + "task_id": "0001364", + "id": "RH20T_cfg2_task_0010_user_0003_scene_0002_cfg_0002_104422070044#178", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001364/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 242, + 148 + ], + [ + 247, + 164 + ], + [ + 226, + 153 + ], + [ + 258, + 144 + ], + [ + 220, + 137 + ], + [ + 253, + 128 + ] + ], + [ + [ + 221, + 152 + ], + [ + 227, + 136 + ], + [ + 210, + 148 + ], + [ + 232, + 156 + ], + [ + 204, + 163 + ], + [ + 225, + 172 + ] + ], + [ + [ + 235, + 137 + ], + [ + 229, + 128 + ], + [ + 251, + 139 + ], + [ + 219, + 135 + ], + [ + 256, + 147 + ], + [ + 225, + 144 + ] + ], + [ + [ + 231, + 149 + ], + [ + 225, + 165 + ], + [ + 215, + 144 + ], + [ + 247, + 155 + ], + [ + 221, + 128 + ], + [ + 253, + 139 + ] + ] + ] + }, + { + "task_id": "0001365", + "id": "RH20T_cfg6_task_0082_user_0020_scene_0001_cfg_0006_104122060811#276", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001365/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 381, + 115 + ], + [ + 392, + 116 + ], + [ + 371, + 122 + ], + [ + 392, + 107 + ], + [ + 360, + 120 + ], + [ + 380, + 105 + ] + ], + [ + [ + 372, + 127 + ], + [ + 380, + 139 + ], + [ + 381, + 126 + ], + [ + 363, + 127 + ], + [ + 374, + 114 + ], + [ + 356, + 114 + ] + ], + [ + [ + 395, + 119 + ], + [ + 386, + 118 + ], + [ + 405, + 129 + ], + [ + 386, + 110 + ], + [ + 414, + 130 + ], + [ + 395, + 112 + ] + ], + [ + [ + 373, + 126 + ], + [ + 380, + 139 + ], + [ + 370, + 132 + ], + [ + 376, + 120 + ], + [ + 363, + 120 + ], + [ + 368, + 107 + ] + ] + ] + }, + { + "task_id": "0001368", + "id": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122061330#271", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001368/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 297, + 219 + ], + [ + 298, + 205 + ], + [ + 285, + 242 + ], + [ + 309, + 198 + ], + [ + 285, + 255 + ], + [ + 308, + 212 + ] + ], + [ + [ + 307, + 207 + ], + [ + 325, + 203 + ], + [ + 317, + 234 + ], + [ + 297, + 181 + ], + [ + 297, + 240 + ], + [ + 276, + 183 + ] + ], + [ + [ + 304, + 189 + ], + [ + 305, + 185 + ], + [ + 295, + 161 + ], + [ + 313, + 216 + ], + [ + 294, + 163 + ], + [ + 313, + 222 + ] + ], + [ + [ + 292, + 233 + ], + [ + 267, + 218 + ], + [ + 306, + 207 + ], + [ + 277, + 260 + ], + [ + 330, + 222 + ], + [ + 301, + 274 + ] + ] + ] + }, + { + "task_id": "0001370", + "id": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122061330#587", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001370/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 262, + 217 + ], + [ + 262, + 203 + ], + [ + 250, + 240 + ], + [ + 274, + 196 + ], + [ + 251, + 252 + ], + [ + 274, + 209 + ] + ], + [ + [ + 280, + 203 + ], + [ + 299, + 197 + ], + [ + 260, + 186 + ], + [ + 298, + 219 + ], + [ + 242, + 192 + ], + [ + 280, + 224 + ] + ], + [ + [ + 230, + 208 + ], + [ + 229, + 180 + ], + [ + 218, + 201 + ], + [ + 242, + 216 + ], + [ + 219, + 227 + ], + [ + 243, + 245 + ] + ], + [ + [ + 255, + 198 + ], + [ + 264, + 175 + ], + [ + 236, + 196 + ], + [ + 271, + 200 + ], + [ + 226, + 221 + ], + [ + 262, + 222 + ] + ] + ] + }, + { + "task_id": "0001371", + "id": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122061330#720", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001371/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 292, + 208 + ], + [ + 320, + 211 + ], + [ + 293, + 182 + ], + [ + 291, + 236 + ], + [ + 265, + 178 + ], + [ + 262, + 233 + ] + ], + [ + [ + 285, + 198 + ], + [ + 285, + 180 + ], + [ + 274, + 219 + ], + [ + 294, + 179 + ], + [ + 274, + 234 + ], + [ + 293, + 196 + ] + ], + [ + [ + 264, + 212 + ], + [ + 248, + 226 + ], + [ + 281, + 203 + ], + [ + 249, + 221 + ], + [ + 296, + 191 + ], + [ + 264, + 209 + ] + ], + [ + [ + 277, + 191 + ], + [ + 288, + 167 + ], + [ + 288, + 193 + ], + [ + 267, + 189 + ], + [ + 277, + 218 + ], + [ + 257, + 212 + ] + ] + ] + }, + { + "task_id": "0001372", + "id": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122061330#849", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001372/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 155 + ], + [ + 290, + 174 + ], + [ + 305, + 151 + ], + [ + 312, + 159 + ], + [ + 326, + 130 + ], + [ + 330, + 141 + ] + ], + [ + [ + 270, + 176 + ], + [ + 281, + 165 + ], + [ + 285, + 161 + ], + [ + 254, + 192 + ], + [ + 275, + 171 + ], + [ + 244, + 202 + ] + ], + [ + [ + 296, + 156 + ], + [ + 320, + 166 + ], + [ + 306, + 132 + ], + [ + 286, + 181 + ], + [ + 281, + 121 + ], + [ + 261, + 170 + ] + ], + [ + [ + 289, + 160 + ], + [ + 291, + 143 + ], + [ + 278, + 179 + ], + [ + 300, + 143 + ], + [ + 276, + 194 + ], + [ + 297, + 158 + ] + ] + ] + }, + { + "task_id": "0001373", + "id": "RH20T_cfg2_task_0015_user_0006_scene_0003_cfg_0002_036422060215#374", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001373/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 446, + 324 + ], + [ + 438, + 305 + ], + [ + 427, + 321 + ], + [ + 465, + 326 + ], + [ + 435, + 339 + ], + [ + 473, + 345 + ] + ], + [ + [ + 423, + 343 + ], + [ + 432, + 330 + ], + [ + 442, + 357 + ], + [ + 404, + 329 + ], + [ + 433, + 360 + ], + [ + 395, + 341 + ] + ], + [ + [ + 426, + 351 + ], + [ + 425, + 343 + ], + [ + 439, + 333 + ], + [ + 412, + 360 + ], + [ + 440, + 340 + ], + [ + 414, + 360 + ] + ], + [ + [ + 425, + 354 + ], + [ + 436, + 358 + ], + [ + 446, + 358 + ], + [ + 406, + 351 + ], + [ + 435, + 354 + ], + [ + 394, + 348 + ] + ] + ] + }, + { + "task_id": "0001375", + "id": "RH20T_cfg1_task_0013_user_0001_scene_0005_cfg_0001_819112071037#362", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001375/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 552, + 149 + ], + [ + 606, + 198 + ], + [ + 489, + 161 + ], + [ + 611, + 138 + ], + [ + 452, + 119 + ], + [ + 564, + 102 + ] + ], + [ + [ + 640, + 82 + ], + [ + 640, + 95 + ], + [ + 615, + 38 + ], + [ + 640, + 143 + ], + [ + 547, + 23 + ], + [ + 640, + 132 + ] + ], + [ + [ + 587, + 97 + ], + [ + 626, + 10 + ], + [ + 526, + 105 + ], + [ + 639, + 89 + ], + [ + 498, + 180 + ], + [ + 605, + 157 + ] + ], + [ + [ + 547, + 100 + ], + [ + 499, + 40 + ], + [ + 622, + 48 + ], + [ + 482, + 145 + ], + [ + 640, + 112 + ], + [ + 527, + 202 + ] + ] + ] + }, + { + "task_id": "0001379", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0007_cfg_0004_104122062823#635", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001379/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 200 + ], + [ + 272, + 168 + ], + [ + 311, + 183 + ], + [ + 234, + 218 + ], + [ + 310, + 211 + ], + [ + 238, + 246 + ] + ], + [ + [ + 323, + 210 + ], + [ + 287, + 207 + ], + [ + 328, + 165 + ], + [ + 317, + 255 + ], + [ + 359, + 170 + ], + [ + 349, + 255 + ] + ], + [ + [ + 273, + 243 + ], + [ + 232, + 227 + ], + [ + 291, + 213 + ], + [ + 252, + 278 + ], + [ + 331, + 227 + ], + [ + 298, + 296 + ] + ], + [ + [ + 275, + 190 + ], + [ + 300, + 224 + ], + [ + 312, + 179 + ], + [ + 234, + 202 + ], + [ + 290, + 150 + ], + [ + 213, + 168 + ] + ] + ] + }, + { + "task_id": "0001381", + "id": "RH20T_cfg5_task_0123_user_0007_scene_0007_cfg_0005_104422070044#130", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the knife on the knife rack'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001381/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 395, + 151 + ], + [ + 393, + 139 + ], + [ + 406, + 160 + ], + [ + 383, + 143 + ], + [ + 407, + 172 + ], + [ + 385, + 154 + ] + ], + [ + [ + 397, + 133 + ], + [ + 382, + 135 + ], + [ + 407, + 141 + ], + [ + 389, + 125 + ], + [ + 422, + 138 + ], + [ + 403, + 123 + ] + ], + [ + [ + 412, + 143 + ], + [ + 400, + 138 + ], + [ + 422, + 150 + ], + [ + 402, + 136 + ], + [ + 433, + 156 + ], + [ + 413, + 141 + ] + ], + [ + [ + 398, + 139 + ], + [ + 395, + 155 + ], + [ + 414, + 142 + ], + [ + 382, + 136 + ], + [ + 416, + 126 + ], + [ + 384, + 119 + ] + ] + ] + }, + { + "task_id": "0001382", + "id": "RH20T_cfg4_task_0120_user_0007_scene_0004_cfg_0004_036422060909#128", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001382/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 101 + ], + [ + 323, + 105 + ], + [ + 309, + 78 + ], + [ + 288, + 122 + ], + [ + 283, + 73 + ], + [ + 263, + 119 + ] + ], + [ + [ + 306, + 107 + ], + [ + 297, + 132 + ], + [ + 308, + 91 + ], + [ + 303, + 121 + ], + [ + 317, + 67 + ], + [ + 311, + 98 + ] + ], + [ + [ + 301, + 110 + ], + [ + 325, + 122 + ], + [ + 297, + 89 + ], + [ + 304, + 129 + ], + [ + 272, + 77 + ], + [ + 281, + 117 + ] + ], + [ + [ + 320, + 117 + ], + [ + 329, + 94 + ], + [ + 329, + 104 + ], + [ + 310, + 131 + ], + [ + 322, + 125 + ], + [ + 302, + 152 + ] + ] + ] + }, + { + "task_id": "0001383", + "id": "RH20T_cfg4_task_0120_user_0007_scene_0004_cfg_0004_036422060909#286", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001383/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 122 + ], + [ + 343, + 101 + ], + [ + 329, + 103 + ], + [ + 348, + 144 + ], + [ + 326, + 122 + ], + [ + 343, + 162 + ] + ], + [ + [ + 338, + 121 + ], + [ + 309, + 127 + ], + [ + 348, + 135 + ], + [ + 328, + 108 + ], + [ + 377, + 128 + ], + [ + 355, + 103 + ] + ], + [ + [ + 366, + 118 + ], + [ + 378, + 137 + ], + [ + 341, + 136 + ], + [ + 392, + 100 + ], + [ + 331, + 118 + ], + [ + 379, + 83 + ] + ], + [ + [ + 350, + 116 + ], + [ + 324, + 114 + ], + [ + 350, + 86 + ], + [ + 351, + 146 + ], + [ + 374, + 89 + ], + [ + 376, + 148 + ] + ] + ] + }, + { + "task_id": "0001385", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0009_cfg_0004_104422070011#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001385/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 409, + 162 + ], + [ + 412, + 161 + ], + [ + 402, + 186 + ], + [ + 415, + 137 + ], + [ + 399, + 189 + ], + [ + 412, + 137 + ] + ], + [ + [ + 413, + 177 + ], + [ + 425, + 165 + ], + [ + 394, + 183 + ], + [ + 431, + 171 + ], + [ + 384, + 195 + ], + [ + 419, + 183 + ] + ], + [ + [ + 411, + 169 + ], + [ + 428, + 165 + ], + [ + 397, + 178 + ], + [ + 427, + 160 + ], + [ + 379, + 183 + ], + [ + 409, + 164 + ] + ], + [ + [ + 413, + 192 + ], + [ + 426, + 174 + ], + [ + 418, + 203 + ], + [ + 407, + 180 + ], + [ + 405, + 221 + ], + [ + 392, + 199 + ] + ] + ] + }, + { + "task_id": "0001386", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0009_cfg_0004_104422070011#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001386/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 403, + 195 + ], + [ + 389, + 216 + ], + [ + 382, + 180 + ], + [ + 425, + 210 + ], + [ + 396, + 158 + ], + [ + 440, + 188 + ] + ], + [ + [ + 407, + 203 + ], + [ + 415, + 207 + ], + [ + 390, + 220 + ], + [ + 424, + 186 + ], + [ + 380, + 217 + ], + [ + 415, + 181 + ] + ], + [ + [ + 418, + 194 + ], + [ + 437, + 182 + ], + [ + 409, + 173 + ], + [ + 428, + 215 + ], + [ + 389, + 185 + ], + [ + 409, + 228 + ] + ], + [ + [ + 401, + 182 + ], + [ + 412, + 169 + ], + [ + 381, + 188 + ], + [ + 420, + 176 + ], + [ + 371, + 200 + ], + [ + 409, + 188 + ] + ] + ] + }, + { + "task_id": "0001387", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0009_cfg_0004_104422070011#321", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001387/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 388, + 181 + ], + [ + 400, + 168 + ], + [ + 367, + 184 + ], + [ + 409, + 178 + ], + [ + 357, + 196 + ], + [ + 398, + 189 + ] + ], + [ + [ + 386, + 170 + ], + [ + 371, + 178 + ], + [ + 401, + 157 + ], + [ + 372, + 183 + ], + [ + 414, + 150 + ], + [ + 386, + 175 + ] + ], + [ + [ + 381, + 192 + ], + [ + 372, + 214 + ], + [ + 361, + 181 + ], + [ + 402, + 204 + ], + [ + 369, + 159 + ], + [ + 412, + 181 + ] + ], + [ + [ + 378, + 159 + ], + [ + 370, + 139 + ], + [ + 359, + 154 + ], + [ + 395, + 163 + ], + [ + 367, + 175 + ], + [ + 403, + 184 + ] + ] + ] + }, + { + "task_id": "0001389", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0009_cfg_0004_104422070011#552", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001389/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 366, + 222 + ], + [ + 360, + 201 + ], + [ + 354, + 209 + ], + [ + 379, + 235 + ], + [ + 359, + 229 + ], + [ + 385, + 255 + ] + ], + [ + [ + 355, + 185 + ], + [ + 344, + 198 + ], + [ + 378, + 197 + ], + [ + 334, + 174 + ], + [ + 388, + 185 + ], + [ + 345, + 164 + ] + ], + [ + [ + 338, + 199 + ], + [ + 333, + 225 + ], + [ + 357, + 204 + ], + [ + 319, + 195 + ], + [ + 362, + 177 + ], + [ + 324, + 170 + ] + ], + [ + [ + 362, + 196 + ], + [ + 373, + 185 + ], + [ + 339, + 201 + ], + [ + 384, + 192 + ], + [ + 330, + 212 + ], + [ + 373, + 202 + ] + ] + ] + }, + { + "task_id": "0001391", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0009_cfg_0004_104422070011#746", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001391/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 329, + 214 + ], + [ + 339, + 202 + ], + [ + 305, + 222 + ], + [ + 351, + 206 + ], + [ + 297, + 232 + ], + [ + 341, + 217 + ] + ], + [ + [ + 357, + 218 + ], + [ + 354, + 242 + ], + [ + 335, + 205 + ], + [ + 379, + 231 + ], + [ + 338, + 183 + ], + [ + 380, + 209 + ] + ], + [ + [ + 305, + 206 + ], + [ + 314, + 233 + ], + [ + 332, + 196 + ], + [ + 278, + 216 + ], + [ + 323, + 170 + ], + [ + 270, + 189 + ] + ], + [ + [ + 312, + 227 + ], + [ + 324, + 250 + ], + [ + 288, + 239 + ], + [ + 335, + 215 + ], + [ + 276, + 216 + ], + [ + 323, + 193 + ] + ] + ] + }, + { + "task_id": "0001392", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0009_cfg_0004_104422070011#929", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001392/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 237 + ], + [ + 283, + 214 + ], + [ + 317, + 231 + ], + [ + 279, + 243 + ], + [ + 334, + 254 + ], + [ + 296, + 268 + ] + ], + [ + [ + 313, + 209 + ], + [ + 284, + 206 + ], + [ + 312, + 232 + ], + [ + 314, + 185 + ], + [ + 340, + 236 + ], + [ + 344, + 188 + ] + ], + [ + [ + 322, + 228 + ], + [ + 341, + 234 + ], + [ + 341, + 210 + ], + [ + 304, + 245 + ], + [ + 321, + 203 + ], + [ + 283, + 240 + ] + ], + [ + [ + 306, + 219 + ], + [ + 314, + 207 + ], + [ + 281, + 227 + ], + [ + 330, + 211 + ], + [ + 274, + 238 + ], + [ + 321, + 222 + ] + ] + ] + }, + { + "task_id": "0001395", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_037522061512#118", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001395/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 244, + 112 + ], + [ + 258, + 108 + ], + [ + 255, + 102 + ], + [ + 232, + 122 + ], + [ + 242, + 106 + ], + [ + 219, + 126 + ] + ], + [ + [ + 236, + 86 + ], + [ + 256, + 83 + ], + [ + 235, + 71 + ], + [ + 237, + 102 + ], + [ + 215, + 74 + ], + [ + 216, + 105 + ] + ], + [ + [ + 246, + 98 + ], + [ + 244, + 81 + ], + [ + 255, + 86 + ], + [ + 236, + 111 + ], + [ + 257, + 102 + ], + [ + 238, + 126 + ] + ], + [ + [ + 248, + 99 + ], + [ + 240, + 116 + ], + [ + 244, + 108 + ], + [ + 253, + 90 + ], + [ + 250, + 92 + ], + [ + 260, + 73 + ] + ] + ] + }, + { + "task_id": "0001396", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_037522061512#249", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001396/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 221, + 116 + ], + [ + 215, + 102 + ], + [ + 231, + 100 + ], + [ + 211, + 132 + ], + [ + 237, + 114 + ], + [ + 217, + 144 + ] + ], + [ + [ + 224, + 114 + ], + [ + 232, + 100 + ], + [ + 242, + 118 + ], + [ + 205, + 111 + ], + [ + 234, + 131 + ], + [ + 198, + 124 + ] + ], + [ + [ + 233, + 125 + ], + [ + 245, + 108 + ], + [ + 241, + 129 + ], + [ + 226, + 120 + ], + [ + 229, + 145 + ], + [ + 213, + 138 + ] + ], + [ + [ + 221, + 124 + ], + [ + 227, + 140 + ], + [ + 241, + 117 + ], + [ + 200, + 131 + ], + [ + 236, + 100 + ], + [ + 192, + 115 + ] + ] + ] + }, + { + "task_id": "0001397", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_037522061512#370", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001397/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 213, + 136 + ], + [ + 211, + 156 + ], + [ + 190, + 133 + ], + [ + 237, + 140 + ], + [ + 192, + 111 + ], + [ + 240, + 119 + ] + ], + [ + [ + 206, + 135 + ], + [ + 224, + 143 + ], + [ + 203, + 155 + ], + [ + 209, + 114 + ], + [ + 186, + 147 + ], + [ + 192, + 106 + ] + ], + [ + [ + 212, + 132 + ], + [ + 207, + 116 + ], + [ + 224, + 119 + ], + [ + 200, + 146 + ], + [ + 228, + 134 + ], + [ + 205, + 161 + ] + ], + [ + [ + 208, + 128 + ], + [ + 193, + 113 + ], + [ + 220, + 119 + ], + [ + 196, + 138 + ], + [ + 234, + 132 + ], + [ + 212, + 153 + ] + ] + ] + }, + { + "task_id": "0001399", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_037522061512#653", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001399/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 184, + 154 + ], + [ + 179, + 136 + ], + [ + 198, + 141 + ], + [ + 169, + 167 + ], + [ + 202, + 158 + ], + [ + 175, + 184 + ] + ], + [ + [ + 206, + 160 + ], + [ + 188, + 147 + ], + [ + 218, + 152 + ], + [ + 195, + 167 + ], + [ + 235, + 164 + ], + [ + 212, + 178 + ] + ], + [ + [ + 191, + 177 + ], + [ + 173, + 169 + ], + [ + 176, + 184 + ], + [ + 205, + 170 + ], + [ + 194, + 193 + ], + [ + 223, + 177 + ] + ], + [ + [ + 164, + 165 + ], + [ + 155, + 168 + ], + [ + 167, + 144 + ], + [ + 160, + 187 + ], + [ + 176, + 141 + ], + [ + 170, + 185 + ] + ] + ] + }, + { + "task_id": "0001400", + "id": "RH20T_cfg7_task_0006_user_0020_scene_0006_cfg_0007_037522061512#168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001400/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 283, + 114 + ], + [ + 283, + 134 + ], + [ + 283, + 118 + ], + [ + 283, + 109 + ], + [ + 282, + 99 + ], + [ + 282, + 88 + ] + ], + [ + [ + 289, + 83 + ], + [ + 276, + 67 + ], + [ + 274, + 97 + ], + [ + 305, + 70 + ], + [ + 287, + 113 + ], + [ + 317, + 86 + ] + ], + [ + [ + 278, + 99 + ], + [ + 288, + 87 + ], + [ + 281, + 114 + ], + [ + 274, + 82 + ], + [ + 271, + 126 + ], + [ + 262, + 95 + ] + ], + [ + [ + 268, + 96 + ], + [ + 249, + 93 + ], + [ + 272, + 76 + ], + [ + 263, + 116 + ], + [ + 291, + 78 + ], + [ + 282, + 119 + ] + ] + ] + }, + { + "task_id": "0001404", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0008_cfg_0004_104422070011#94", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001404/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 252, + 150 + ], + [ + 267, + 167 + ], + [ + 269, + 128 + ], + [ + 234, + 173 + ], + [ + 253, + 109 + ], + [ + 215, + 154 + ] + ], + [ + [ + 272, + 175 + ], + [ + 271, + 183 + ], + [ + 257, + 196 + ], + [ + 288, + 153 + ], + [ + 259, + 187 + ], + [ + 289, + 147 + ] + ], + [ + [ + 247, + 183 + ], + [ + 273, + 193 + ], + [ + 248, + 202 + ], + [ + 246, + 162 + ], + [ + 224, + 193 + ], + [ + 220, + 153 + ] + ], + [ + [ + 261, + 167 + ], + [ + 266, + 152 + ], + [ + 261, + 193 + ], + [ + 261, + 144 + ], + [ + 257, + 205 + ], + [ + 257, + 158 + ] + ] + ] + }, + { + "task_id": "0001405", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0008_cfg_0004_104422070011#211", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001405/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 277, + 140 + ], + [ + 250, + 146 + ], + [ + 272, + 118 + ], + [ + 281, + 159 + ], + [ + 300, + 112 + ], + [ + 307, + 152 + ] + ], + [ + [ + 271, + 161 + ], + [ + 273, + 146 + ], + [ + 272, + 185 + ], + [ + 270, + 138 + ], + [ + 269, + 198 + ], + [ + 268, + 152 + ] + ], + [ + [ + 271, + 169 + ], + [ + 294, + 176 + ], + [ + 259, + 186 + ], + [ + 283, + 150 + ], + [ + 234, + 180 + ], + [ + 258, + 142 + ] + ], + [ + [ + 288, + 157 + ], + [ + 308, + 139 + ], + [ + 308, + 177 + ], + [ + 269, + 137 + ], + [ + 288, + 195 + ], + [ + 248, + 155 + ] + ] + ] + }, + { + "task_id": "0001408", + "id": "RH20T_cfg5_task_0085_user_0007_scene_0007_cfg_0005_104122061850#246", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the bubble ring from the assembled bubble ring and ball'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001408/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 352, + 187 + ], + [ + 366, + 176 + ], + [ + 364, + 209 + ], + [ + 340, + 165 + ], + [ + 350, + 221 + ], + [ + 325, + 176 + ] + ], + [ + [ + 374, + 156 + ], + [ + 388, + 136 + ], + [ + 379, + 148 + ], + [ + 370, + 164 + ], + [ + 364, + 170 + ], + [ + 357, + 183 + ] + ], + [ + [ + 357, + 168 + ], + [ + 355, + 155 + ], + [ + 332, + 163 + ], + [ + 381, + 174 + ], + [ + 335, + 176 + ], + [ + 382, + 187 + ] + ], + [ + [ + 352, + 172 + ], + [ + 357, + 147 + ], + [ + 347, + 171 + ], + [ + 357, + 174 + ], + [ + 343, + 195 + ], + [ + 353, + 201 + ] + ] + ] + }, + { + "task_id": "0001414", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0006_cfg_0003_104122063550#213", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001414/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 339, + 304 + ], + [ + 368, + 291 + ], + [ + 341, + 286 + ], + [ + 337, + 325 + ], + [ + 314, + 297 + ], + [ + 306, + 337 + ] + ], + [ + [ + 365, + 276 + ], + [ + 336, + 280 + ], + [ + 372, + 285 + ], + [ + 358, + 269 + ], + [ + 403, + 281 + ], + [ + 385, + 265 + ] + ], + [ + [ + 352, + 289 + ], + [ + 351, + 286 + ], + [ + 363, + 320 + ], + [ + 341, + 260 + ], + [ + 363, + 321 + ], + [ + 343, + 265 + ] + ], + [ + [ + 384, + 294 + ], + [ + 372, + 278 + ], + [ + 397, + 311 + ], + [ + 370, + 276 + ], + [ + 411, + 329 + ], + [ + 384, + 293 + ] + ] + ] + }, + { + "task_id": "0001415", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0006_cfg_0003_104122063550#273", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001415/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 301, + 179 + ], + [ + 284, + 172 + ], + [ + 327, + 175 + ], + [ + 275, + 182 + ], + [ + 341, + 182 + ], + [ + 291, + 188 + ] + ], + [ + [ + 300, + 189 + ], + [ + 273, + 189 + ], + [ + 306, + 209 + ], + [ + 294, + 171 + ], + [ + 333, + 209 + ], + [ + 318, + 172 + ] + ], + [ + [ + 311, + 160 + ], + [ + 331, + 155 + ], + [ + 303, + 180 + ], + [ + 319, + 138 + ], + [ + 281, + 186 + ], + [ + 296, + 143 + ] + ], + [ + [ + 315, + 180 + ], + [ + 313, + 167 + ], + [ + 313, + 207 + ], + [ + 317, + 154 + ], + [ + 315, + 218 + ], + [ + 319, + 167 + ] + ] + ] + }, + { + "task_id": "0001417", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_819112070044#98", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001417/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 420, + 183 + ], + [ + 421, + 162 + ], + [ + 421, + 176 + ], + [ + 419, + 190 + ], + [ + 420, + 199 + ], + [ + 419, + 211 + ] + ], + [ + [ + 419, + 188 + ], + [ + 426, + 173 + ], + [ + 435, + 182 + ], + [ + 402, + 195 + ], + [ + 428, + 196 + ], + [ + 396, + 209 + ] + ], + [ + [ + 425, + 185 + ], + [ + 416, + 176 + ], + [ + 411, + 176 + ], + [ + 440, + 195 + ], + [ + 420, + 184 + ], + [ + 449, + 203 + ] + ], + [ + [ + 443, + 193 + ], + [ + 430, + 176 + ], + [ + 430, + 197 + ], + [ + 456, + 189 + ], + [ + 443, + 214 + ], + [ + 469, + 206 + ] + ] + ] + }, + { + "task_id": "0001418", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_819112070044#195", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001418/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 370, + 214 + ], + [ + 376, + 199 + ], + [ + 389, + 206 + ], + [ + 351, + 222 + ], + [ + 383, + 220 + ], + [ + 347, + 235 + ] + ], + [ + [ + 365, + 208 + ], + [ + 371, + 193 + ], + [ + 357, + 220 + ], + [ + 374, + 195 + ], + [ + 351, + 237 + ], + [ + 368, + 212 + ] + ], + [ + [ + 394, + 209 + ], + [ + 403, + 226 + ], + [ + 372, + 210 + ], + [ + 415, + 208 + ], + [ + 364, + 194 + ], + [ + 406, + 193 + ] + ], + [ + [ + 390, + 210 + ], + [ + 407, + 214 + ], + [ + 388, + 188 + ], + [ + 391, + 234 + ], + [ + 370, + 183 + ], + [ + 372, + 231 + ] + ] + ] + }, + { + "task_id": "0001420", + "id": "RH20T_cfg6_task_0039_user_0014_scene_0002_cfg_0006_104122060811#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001420/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 383, + 143 + ], + [ + 390, + 130 + ], + [ + 374, + 141 + ], + [ + 394, + 146 + ], + [ + 367, + 154 + ], + [ + 387, + 160 + ] + ], + [ + [ + 390, + 130 + ], + [ + 392, + 120 + ], + [ + 398, + 120 + ], + [ + 383, + 140 + ], + [ + 396, + 130 + ], + [ + 381, + 150 + ] + ], + [ + [ + 387, + 140 + ], + [ + 390, + 132 + ], + [ + 401, + 134 + ], + [ + 374, + 145 + ], + [ + 399, + 141 + ], + [ + 372, + 152 + ] + ], + [ + [ + 393, + 153 + ], + [ + 384, + 140 + ], + [ + 380, + 162 + ], + [ + 405, + 144 + ], + [ + 388, + 174 + ], + [ + 413, + 156 + ] + ] + ] + }, + { + "task_id": "0001421", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122063633#325", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001421/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 221, + 139 + ], + [ + 208, + 151 + ], + [ + 228, + 158 + ], + [ + 212, + 118 + ], + [ + 240, + 146 + ], + [ + 225, + 107 + ] + ], + [ + [ + 186, + 151 + ], + [ + 178, + 165 + ], + [ + 164, + 142 + ], + [ + 207, + 159 + ], + [ + 172, + 126 + ], + [ + 216, + 145 + ] + ], + [ + [ + 205, + 132 + ], + [ + 204, + 112 + ], + [ + 209, + 119 + ], + [ + 200, + 145 + ], + [ + 209, + 137 + ], + [ + 201, + 163 + ] + ], + [ + [ + 214, + 126 + ], + [ + 209, + 130 + ], + [ + 201, + 108 + ], + [ + 227, + 144 + ], + [ + 206, + 103 + ], + [ + 233, + 141 + ] + ] + ] + }, + { + "task_id": "0001422", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122063633#756", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001422/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 226, + 214 + ], + [ + 227, + 192 + ], + [ + 230, + 202 + ], + [ + 221, + 227 + ], + [ + 228, + 222 + ], + [ + 220, + 248 + ] + ], + [ + [ + 234, + 241 + ], + [ + 212, + 254 + ], + [ + 230, + 241 + ], + [ + 238, + 242 + ], + [ + 254, + 227 + ], + [ + 259, + 230 + ] + ], + [ + [ + 218, + 237 + ], + [ + 202, + 218 + ], + [ + 207, + 252 + ], + [ + 229, + 221 + ], + [ + 222, + 270 + ], + [ + 246, + 240 + ] + ], + [ + [ + 199, + 203 + ], + [ + 224, + 205 + ], + [ + 199, + 177 + ], + [ + 198, + 227 + ], + [ + 173, + 174 + ], + [ + 172, + 225 + ] + ] + ] + }, + { + "task_id": "0001426", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_819112070044#95", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001426/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 434, + 157 + ], + [ + 437, + 142 + ], + [ + 448, + 151 + ], + [ + 419, + 163 + ], + [ + 444, + 165 + ], + [ + 416, + 177 + ] + ], + [ + [ + 428, + 157 + ], + [ + 433, + 140 + ], + [ + 435, + 160 + ], + [ + 420, + 155 + ], + [ + 430, + 178 + ], + [ + 416, + 171 + ] + ], + [ + [ + 424, + 173 + ], + [ + 435, + 183 + ], + [ + 420, + 189 + ], + [ + 429, + 156 + ], + [ + 410, + 180 + ], + [ + 419, + 147 + ] + ], + [ + [ + 429, + 156 + ], + [ + 411, + 158 + ], + [ + 427, + 176 + ], + [ + 430, + 138 + ], + [ + 447, + 175 + ], + [ + 449, + 136 + ] + ] + ] + }, + { + "task_id": "0001428", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_819112070044#366", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001428/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 376, + 163 + ], + [ + 362, + 174 + ], + [ + 386, + 162 + ], + [ + 366, + 164 + ], + [ + 400, + 152 + ], + [ + 380, + 153 + ] + ], + [ + [ + 395, + 168 + ], + [ + 398, + 154 + ], + [ + 411, + 162 + ], + [ + 379, + 174 + ], + [ + 408, + 176 + ], + [ + 377, + 188 + ] + ], + [ + [ + 385, + 175 + ], + [ + 375, + 168 + ], + [ + 376, + 192 + ], + [ + 394, + 159 + ], + [ + 385, + 198 + ], + [ + 403, + 166 + ] + ], + [ + [ + 404, + 173 + ], + [ + 403, + 165 + ], + [ + 415, + 187 + ], + [ + 393, + 158 + ], + [ + 417, + 195 + ], + [ + 394, + 166 + ] + ] + ] + }, + { + "task_id": "0001431", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122063633#161", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001431/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 188, + 172 + ], + [ + 212, + 173 + ], + [ + 189, + 149 + ], + [ + 188, + 196 + ], + [ + 165, + 149 + ], + [ + 163, + 196 + ] + ], + [ + [ + 226, + 151 + ], + [ + 248, + 149 + ], + [ + 225, + 146 + ], + [ + 228, + 157 + ], + [ + 204, + 148 + ], + [ + 205, + 160 + ] + ], + [ + [ + 177, + 156 + ], + [ + 153, + 154 + ], + [ + 175, + 169 + ], + [ + 179, + 145 + ], + [ + 200, + 171 + ], + [ + 202, + 147 + ] + ], + [ + [ + 201, + 165 + ], + [ + 202, + 144 + ], + [ + 205, + 157 + ], + [ + 198, + 174 + ], + [ + 204, + 177 + ], + [ + 197, + 195 + ] + ] + ] + }, + { + "task_id": "0001432", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122063633#366", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001432/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 220, + 148 + ], + [ + 239, + 155 + ], + [ + 219, + 168 + ], + [ + 222, + 127 + ], + [ + 201, + 162 + ], + [ + 203, + 121 + ] + ], + [ + [ + 228, + 153 + ], + [ + 244, + 159 + ], + [ + 244, + 149 + ], + [ + 214, + 156 + ], + [ + 227, + 142 + ], + [ + 197, + 149 + ] + ], + [ + [ + 224, + 151 + ], + [ + 226, + 130 + ], + [ + 227, + 146 + ], + [ + 221, + 156 + ], + [ + 224, + 166 + ], + [ + 219, + 178 + ] + ], + [ + [ + 224, + 147 + ], + [ + 203, + 146 + ], + [ + 222, + 127 + ], + [ + 226, + 166 + ], + [ + 244, + 127 + ], + [ + 246, + 167 + ] + ] + ] + }, + { + "task_id": "0001437", + "id": "RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_104122061018#141", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Cover the box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001437/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 243, + 64 + ], + [ + 241, + 59 + ], + [ + 214, + 58 + ], + [ + 272, + 70 + ], + [ + 215, + 64 + ], + [ + 276, + 76 + ] + ], + [ + [ + 278, + 73 + ], + [ + 303, + 76 + ], + [ + 267, + 55 + ], + [ + 289, + 90 + ], + [ + 242, + 53 + ], + [ + 266, + 88 + ] + ], + [ + [ + 244, + 82 + ], + [ + 216, + 88 + ], + [ + 249, + 66 + ], + [ + 239, + 96 + ], + [ + 276, + 60 + ], + [ + 264, + 90 + ] + ], + [ + [ + 275, + 90 + ], + [ + 281, + 72 + ], + [ + 265, + 67 + ], + [ + 285, + 113 + ], + [ + 259, + 84 + ], + [ + 279, + 128 + ] + ] + ] + }, + { + "task_id": "0001438", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122064161#111", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001438/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 173 + ], + [ + 385, + 185 + ], + [ + 355, + 190 + ], + [ + 372, + 155 + ], + [ + 335, + 179 + ], + [ + 350, + 144 + ] + ], + [ + [ + 364, + 155 + ], + [ + 346, + 165 + ], + [ + 379, + 170 + ], + [ + 349, + 140 + ], + [ + 397, + 159 + ], + [ + 367, + 130 + ] + ], + [ + [ + 375, + 158 + ], + [ + 377, + 138 + ], + [ + 375, + 173 + ], + [ + 376, + 145 + ], + [ + 374, + 192 + ], + [ + 374, + 164 + ] + ], + [ + [ + 380, + 155 + ], + [ + 383, + 172 + ], + [ + 390, + 140 + ], + [ + 370, + 171 + ], + [ + 388, + 122 + ], + [ + 368, + 152 + ] + ] + ] + }, + { + "task_id": "0001439", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122064161#284", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001439/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 374, + 158 + ], + [ + 374, + 173 + ], + [ + 372, + 176 + ], + [ + 377, + 140 + ], + [ + 372, + 162 + ], + [ + 377, + 127 + ] + ], + [ + [ + 364, + 146 + ], + [ + 384, + 138 + ], + [ + 369, + 160 + ], + [ + 359, + 132 + ], + [ + 349, + 168 + ], + [ + 338, + 140 + ] + ], + [ + [ + 380, + 168 + ], + [ + 378, + 175 + ], + [ + 399, + 156 + ], + [ + 361, + 181 + ], + [ + 402, + 149 + ], + [ + 362, + 175 + ] + ], + [ + [ + 380, + 134 + ], + [ + 366, + 121 + ], + [ + 394, + 116 + ], + [ + 366, + 152 + ], + [ + 407, + 129 + ], + [ + 379, + 163 + ] + ] + ] + }, + { + "task_id": "0001440", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122064161#412", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001440/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 164 + ], + [ + 416, + 164 + ], + [ + 392, + 186 + ], + [ + 393, + 142 + ], + [ + 369, + 187 + ], + [ + 369, + 142 + ] + ], + [ + [ + 411, + 163 + ], + [ + 420, + 179 + ], + [ + 433, + 160 + ], + [ + 388, + 166 + ], + [ + 424, + 144 + ], + [ + 381, + 150 + ] + ], + [ + [ + 386, + 161 + ], + [ + 385, + 179 + ], + [ + 409, + 167 + ], + [ + 363, + 154 + ], + [ + 410, + 150 + ], + [ + 365, + 137 + ] + ], + [ + [ + 389, + 177 + ], + [ + 388, + 155 + ], + [ + 392, + 187 + ], + [ + 386, + 168 + ], + [ + 393, + 209 + ], + [ + 386, + 188 + ] + ] + ] + }, + { + "task_id": "0001443", + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_819112070044#193", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001443/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 492, + 186 + ], + [ + 474, + 178 + ], + [ + 483, + 186 + ], + [ + 500, + 186 + ], + [ + 503, + 194 + ], + [ + 519, + 193 + ] + ], + [ + [ + 484, + 210 + ], + [ + 482, + 195 + ], + [ + 472, + 223 + ], + [ + 497, + 198 + ], + [ + 474, + 237 + ], + [ + 498, + 213 + ] + ], + [ + [ + 474, + 173 + ], + [ + 471, + 193 + ], + [ + 494, + 175 + ], + [ + 454, + 171 + ], + [ + 497, + 155 + ], + [ + 456, + 152 + ] + ], + [ + [ + 482, + 199 + ], + [ + 488, + 185 + ], + [ + 480, + 185 + ], + [ + 483, + 213 + ], + [ + 475, + 198 + ], + [ + 477, + 227 + ] + ] + ] + }, + { + "task_id": "0001444", + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_819112070044#252", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001444/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 463, + 187 + ], + [ + 473, + 191 + ], + [ + 465, + 168 + ], + [ + 461, + 207 + ], + [ + 454, + 164 + ], + [ + 451, + 203 + ] + ], + [ + [ + 467, + 190 + ], + [ + 473, + 176 + ], + [ + 467, + 177 + ], + [ + 468, + 205 + ], + [ + 462, + 191 + ], + [ + 462, + 219 + ] + ], + [ + [ + 455, + 185 + ], + [ + 445, + 176 + ], + [ + 468, + 197 + ], + [ + 442, + 174 + ], + [ + 477, + 206 + ], + [ + 451, + 184 + ] + ], + [ + [ + 457, + 189 + ], + [ + 476, + 186 + ], + [ + 468, + 192 + ], + [ + 446, + 186 + ], + [ + 451, + 195 + ], + [ + 429, + 189 + ] + ] + ] + }, + { + "task_id": "0001445", + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_819112070044#280", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001445/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 443, + 200 + ], + [ + 456, + 210 + ], + [ + 446, + 213 + ], + [ + 439, + 188 + ], + [ + 432, + 203 + ], + [ + 425, + 178 + ] + ], + [ + [ + 460, + 206 + ], + [ + 443, + 197 + ], + [ + 471, + 191 + ], + [ + 448, + 220 + ], + [ + 489, + 200 + ], + [ + 466, + 229 + ] + ], + [ + [ + 442, + 196 + ], + [ + 427, + 200 + ], + [ + 458, + 198 + ], + [ + 428, + 195 + ], + [ + 472, + 194 + ], + [ + 442, + 191 + ] + ], + [ + [ + 450, + 194 + ], + [ + 455, + 180 + ], + [ + 450, + 181 + ], + [ + 449, + 209 + ], + [ + 446, + 194 + ], + [ + 445, + 222 + ] + ] + ] + }, + { + "task_id": "0001446", + "id": "RH20T_cfg6_task_0125_user_0020_scene_0004_cfg_0006_104122061330#492", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull up the lever'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001446/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 474, + 124 + ], + [ + 477, + 100 + ], + [ + 458, + 134 + ], + [ + 490, + 114 + ], + [ + 456, + 157 + ], + [ + 487, + 136 + ] + ], + [ + [ + 445, + 129 + ], + [ + 442, + 114 + ], + [ + 438, + 152 + ], + [ + 452, + 107 + ], + [ + 441, + 164 + ], + [ + 454, + 121 + ] + ], + [ + [ + 438, + 127 + ], + [ + 414, + 124 + ], + [ + 444, + 146 + ], + [ + 431, + 107 + ], + [ + 470, + 150 + ], + [ + 459, + 110 + ] + ], + [ + [ + 421, + 135 + ], + [ + 449, + 139 + ], + [ + 427, + 112 + ], + [ + 414, + 159 + ], + [ + 401, + 108 + ], + [ + 388, + 155 + ] + ] + ] + }, + { + "task_id": "0001448", + "id": "RH20T_cfg6_task_0125_user_0020_scene_0004_cfg_0006_104122061330#690", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull up the lever'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001448/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 429, + 86 + ], + [ + 401, + 75 + ], + [ + 411, + 110 + ], + [ + 446, + 62 + ], + [ + 440, + 122 + ], + [ + 475, + 73 + ] + ], + [ + [ + 441, + 89 + ], + [ + 438, + 71 + ], + [ + 427, + 116 + ], + [ + 454, + 62 + ], + [ + 430, + 130 + ], + [ + 455, + 79 + ] + ], + [ + [ + 440, + 90 + ], + [ + 425, + 69 + ], + [ + 457, + 66 + ], + [ + 423, + 115 + ], + [ + 469, + 86 + ], + [ + 438, + 133 + ] + ], + [ + [ + 429, + 54 + ], + [ + 426, + 39 + ], + [ + 458, + 65 + ], + [ + 400, + 43 + ], + [ + 463, + 82 + ], + [ + 402, + 59 + ] + ] + ] + }, + { + "task_id": "0001449", + "id": "RH20T_cfg5_task_0104_user_0007_scene_0007_cfg_0005_036422060215#381", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate back after holding it down'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001449/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 279, + 157 + ], + [ + 273, + 174 + ], + [ + 276, + 146 + ], + [ + 283, + 168 + ], + [ + 282, + 130 + ], + [ + 289, + 152 + ] + ], + [ + [ + 295, + 123 + ], + [ + 313, + 116 + ], + [ + 288, + 104 + ], + [ + 302, + 143 + ], + [ + 269, + 112 + ], + [ + 284, + 150 + ] + ], + [ + [ + 283, + 150 + ], + [ + 283, + 137 + ], + [ + 302, + 146 + ], + [ + 263, + 155 + ], + [ + 301, + 158 + ], + [ + 264, + 168 + ] + ], + [ + [ + 294, + 152 + ], + [ + 302, + 171 + ], + [ + 303, + 153 + ], + [ + 284, + 151 + ], + [ + 295, + 135 + ], + [ + 276, + 132 + ] + ] + ] + }, + { + "task_id": "0001453", + "id": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122060811#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the pen into the pen holder'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001453/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 376, + 169 + ], + [ + 380, + 160 + ], + [ + 389, + 166 + ], + [ + 363, + 172 + ], + [ + 385, + 175 + ], + [ + 359, + 182 + ] + ], + [ + [ + 394, + 168 + ], + [ + 390, + 168 + ], + [ + 403, + 179 + ], + [ + 385, + 156 + ], + [ + 407, + 179 + ], + [ + 389, + 156 + ] + ], + [ + [ + 368, + 176 + ], + [ + 357, + 174 + ], + [ + 358, + 175 + ], + [ + 378, + 178 + ], + [ + 370, + 177 + ], + [ + 390, + 180 + ] + ], + [ + [ + 369, + 178 + ], + [ + 378, + 189 + ], + [ + 380, + 173 + ], + [ + 358, + 183 + ], + [ + 371, + 163 + ], + [ + 349, + 172 + ] + ] + ] + }, + { + "task_id": "0001454", + "id": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122060811#279", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the pen into the pen holder'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001454/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 365, + 184 + ], + [ + 378, + 176 + ], + [ + 370, + 187 + ], + [ + 359, + 179 + ], + [ + 358, + 194 + ], + [ + 346, + 187 + ] + ], + [ + [ + 356, + 188 + ], + [ + 360, + 174 + ], + [ + 361, + 185 + ], + [ + 351, + 190 + ], + [ + 358, + 200 + ], + [ + 348, + 203 + ] + ], + [ + [ + 372, + 197 + ], + [ + 371, + 211 + ], + [ + 385, + 202 + ], + [ + 360, + 193 + ], + [ + 385, + 188 + ], + [ + 360, + 180 + ] + ], + [ + [ + 360, + 196 + ], + [ + 350, + 195 + ], + [ + 368, + 207 + ], + [ + 352, + 185 + ], + [ + 377, + 207 + ], + [ + 362, + 186 + ] + ] + ] + }, + { + "task_id": "0001455", + "id": "RH20T_cfg1_task_0051_user_0015_scene_0002_cfg_0001_038522062547#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the dish off the dish rack'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001455/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 203 + ], + [ + 431, + 219 + ], + [ + 389, + 227 + ], + [ + 423, + 179 + ], + [ + 364, + 212 + ], + [ + 398, + 163 + ] + ], + [ + [ + 433, + 179 + ], + [ + 452, + 182 + ], + [ + 449, + 202 + ], + [ + 416, + 156 + ], + [ + 431, + 198 + ], + [ + 399, + 155 + ] + ], + [ + [ + 404, + 198 + ], + [ + 418, + 173 + ], + [ + 433, + 211 + ], + [ + 377, + 185 + ], + [ + 419, + 238 + ], + [ + 363, + 210 + ] + ], + [ + [ + 449, + 222 + ], + [ + 433, + 210 + ], + [ + 478, + 204 + ], + [ + 422, + 238 + ], + [ + 492, + 216 + ], + [ + 438, + 248 + ] + ] + ] + }, + { + "task_id": "0001456", + "id": "RH20T_cfg4_task_0092_user_0014_scene_0007_cfg_0004_104122062823#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the hands of a clock'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001456/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 86 + ], + [ + 302, + 95 + ], + [ + 308, + 53 + ], + [ + 319, + 117 + ], + [ + 320, + 48 + ], + [ + 329, + 108 + ] + ], + [ + [ + 291, + 100 + ], + [ + 292, + 70 + ], + [ + 324, + 104 + ], + [ + 258, + 97 + ], + [ + 322, + 132 + ], + [ + 259, + 125 + ] + ], + [ + [ + 289, + 120 + ], + [ + 269, + 99 + ], + [ + 264, + 143 + ], + [ + 315, + 97 + ], + [ + 286, + 166 + ], + [ + 337, + 118 + ] + ], + [ + [ + 284, + 119 + ], + [ + 249, + 119 + ], + [ + 286, + 150 + ], + [ + 282, + 90 + ], + [ + 322, + 150 + ], + [ + 316, + 91 + ] + ] + ] + }, + { + "task_id": "0001457", + "id": "RH20T_cfg2_task_0057_user_0016_scene_0001_cfg_0002_036422060215#323", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001457/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 345, + 160 + ], + [ + 350, + 146 + ], + [ + 359, + 155 + ], + [ + 330, + 165 + ], + [ + 353, + 167 + ], + [ + 325, + 178 + ] + ], + [ + [ + 340, + 168 + ], + [ + 353, + 156 + ], + [ + 335, + 163 + ], + [ + 346, + 173 + ], + [ + 322, + 175 + ], + [ + 332, + 186 + ] + ], + [ + [ + 341, + 158 + ], + [ + 324, + 160 + ], + [ + 336, + 146 + ], + [ + 347, + 169 + ], + [ + 354, + 143 + ], + [ + 364, + 166 + ] + ], + [ + [ + 344, + 166 + ], + [ + 347, + 154 + ], + [ + 350, + 155 + ], + [ + 339, + 177 + ], + [ + 346, + 169 + ], + [ + 335, + 190 + ] + ] + ] + }, + { + "task_id": "0001459", + "id": "RH20T_cfg2_task_0057_user_0016_scene_0001_cfg_0002_036422060215#486", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001459/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 156 + ], + [ + 326, + 158 + ], + [ + 308, + 137 + ], + [ + 309, + 174 + ], + [ + 292, + 135 + ], + [ + 293, + 171 + ] + ], + [ + [ + 327, + 123 + ], + [ + 341, + 130 + ], + [ + 320, + 114 + ], + [ + 334, + 132 + ], + [ + 306, + 107 + ], + [ + 322, + 126 + ] + ], + [ + [ + 328, + 148 + ], + [ + 334, + 157 + ], + [ + 311, + 146 + ], + [ + 346, + 149 + ], + [ + 304, + 137 + ], + [ + 341, + 139 + ] + ], + [ + [ + 320, + 149 + ], + [ + 326, + 136 + ], + [ + 335, + 145 + ], + [ + 305, + 153 + ], + [ + 330, + 156 + ], + [ + 300, + 165 + ] + ] + ] + }, + { + "task_id": "0001462", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_104122064161#195", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001462/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 382, + 120 + ], + [ + 400, + 116 + ], + [ + 371, + 130 + ], + [ + 393, + 110 + ], + [ + 354, + 134 + ], + [ + 376, + 114 + ] + ], + [ + [ + 369, + 120 + ], + [ + 350, + 130 + ], + [ + 368, + 105 + ], + [ + 370, + 136 + ], + [ + 386, + 94 + ], + [ + 390, + 126 + ] + ], + [ + [ + 375, + 117 + ], + [ + 378, + 100 + ], + [ + 374, + 133 + ], + [ + 375, + 103 + ], + [ + 371, + 148 + ], + [ + 372, + 118 + ] + ], + [ + [ + 372, + 108 + ], + [ + 380, + 89 + ], + [ + 369, + 98 + ], + [ + 376, + 119 + ], + [ + 362, + 116 + ], + [ + 368, + 138 + ] + ] + ] + }, + { + "task_id": "0001468", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_819112071037#677", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001468/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 240, + 168 + ], + [ + 287, + 201 + ], + [ + 219, + 102 + ], + [ + 258, + 221 + ], + [ + 179, + 79 + ], + [ + 222, + 192 + ] + ], + [ + [ + 239, + 213 + ], + [ + 237, + 134 + ], + [ + 169, + 241 + ], + [ + 292, + 192 + ], + [ + 176, + 321 + ], + [ + 291, + 256 + ] + ], + [ + [ + 200, + 195 + ], + [ + 131, + 198 + ], + [ + 153, + 130 + ], + [ + 237, + 246 + ], + [ + 241, + 120 + ], + [ + 309, + 246 + ] + ], + [ + [ + 218, + 200 + ], + [ + 166, + 190 + ], + [ + 218, + 269 + ], + [ + 218, + 127 + ], + [ + 277, + 285 + ], + [ + 280, + 132 + ] + ] + ] + }, + { + "task_id": "0001471", + "id": "RH20T_cfg4_task_0034_user_0016_scene_0001_cfg_0004_104122062823#167", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001471/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 293, + 60 + ], + [ + 281, + 89 + ], + [ + 316, + 74 + ], + [ + 269, + 44 + ], + [ + 328, + 46 + ], + [ + 282, + 14 + ] + ], + [ + [ + 320, + 61 + ], + [ + 342, + 41 + ], + [ + 306, + 65 + ], + [ + 335, + 57 + ], + [ + 284, + 85 + ], + [ + 310, + 80 + ] + ], + [ + [ + 314, + 66 + ], + [ + 334, + 49 + ], + [ + 290, + 57 + ], + [ + 339, + 76 + ], + [ + 269, + 75 + ], + [ + 317, + 95 + ] + ], + [ + [ + 296, + 58 + ], + [ + 295, + 32 + ], + [ + 328, + 59 + ], + [ + 265, + 56 + ], + [ + 327, + 82 + ], + [ + 268, + 79 + ] + ] + ] + }, + { + "task_id": "0001472", + "id": "RH20T_cfg4_task_0034_user_0016_scene_0001_cfg_0004_104122062823#229", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001472/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 238, + 127 + ], + [ + 259, + 153 + ], + [ + 209, + 146 + ], + [ + 266, + 109 + ], + [ + 190, + 121 + ], + [ + 247, + 86 + ] + ], + [ + [ + 252, + 117 + ], + [ + 249, + 99 + ], + [ + 286, + 105 + ], + [ + 219, + 130 + ], + [ + 291, + 125 + ], + [ + 220, + 151 + ] + ], + [ + [ + 224, + 135 + ], + [ + 189, + 132 + ], + [ + 209, + 134 + ], + [ + 242, + 135 + ], + [ + 240, + 137 + ], + [ + 275, + 138 + ] + ], + [ + [ + 229, + 109 + ], + [ + 224, + 79 + ], + [ + 264, + 103 + ], + [ + 191, + 115 + ], + [ + 266, + 130 + ], + [ + 198, + 142 + ] + ] + ] + }, + { + "task_id": "0001476", + "id": "RH20T_cfg4_task_0058_user_0014_scene_0003_cfg_0004_038522062288#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001476/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 227, + 156 + ], + [ + 230, + 142 + ], + [ + 197, + 149 + ], + [ + 256, + 162 + ], + [ + 195, + 162 + ], + [ + 251, + 174 + ] + ], + [ + [ + 200, + 138 + ], + [ + 191, + 152 + ], + [ + 191, + 163 + ], + [ + 208, + 114 + ], + [ + 201, + 148 + ], + [ + 217, + 97 + ] + ], + [ + [ + 228, + 157 + ], + [ + 250, + 164 + ], + [ + 210, + 166 + ], + [ + 247, + 147 + ], + [ + 185, + 159 + ], + [ + 223, + 138 + ] + ], + [ + [ + 237, + 166 + ], + [ + 236, + 192 + ], + [ + 260, + 157 + ], + [ + 215, + 175 + ], + [ + 260, + 132 + ], + [ + 216, + 151 + ] + ] + ] + }, + { + "task_id": "0001481", + "id": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_104422070011#256", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001481/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 270, + 157 + ], + [ + 254, + 131 + ], + [ + 293, + 146 + ], + [ + 249, + 168 + ], + [ + 308, + 173 + ], + [ + 264, + 192 + ] + ], + [ + [ + 284, + 138 + ], + [ + 307, + 145 + ], + [ + 280, + 164 + ], + [ + 289, + 113 + ], + [ + 255, + 157 + ], + [ + 265, + 105 + ] + ], + [ + [ + 270, + 163 + ], + [ + 271, + 192 + ], + [ + 299, + 162 + ], + [ + 240, + 164 + ], + [ + 297, + 133 + ], + [ + 238, + 135 + ] + ], + [ + [ + 281, + 155 + ], + [ + 286, + 140 + ], + [ + 260, + 170 + ], + [ + 300, + 141 + ], + [ + 256, + 183 + ], + [ + 295, + 155 + ] + ] + ] + }, + { + "task_id": "0001484", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061330#244", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001484/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 365, + 262 + ], + [ + 381, + 257 + ], + [ + 344, + 276 + ], + [ + 388, + 247 + ], + [ + 327, + 283 + ], + [ + 371, + 252 + ] + ], + [ + [ + 364, + 252 + ], + [ + 363, + 236 + ], + [ + 352, + 278 + ], + [ + 375, + 229 + ], + [ + 354, + 292 + ], + [ + 375, + 244 + ] + ], + [ + [ + 383, + 267 + ], + [ + 367, + 247 + ], + [ + 393, + 245 + ], + [ + 373, + 292 + ], + [ + 407, + 264 + ], + [ + 389, + 312 + ] + ], + [ + [ + 347, + 253 + ], + [ + 333, + 260 + ], + [ + 320, + 247 + ], + [ + 372, + 259 + ], + [ + 334, + 239 + ], + [ + 389, + 252 + ] + ] + ] + }, + { + "task_id": "0001487", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061330#747", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001487/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 425, + 181 + ], + [ + 395, + 186 + ], + [ + 427, + 202 + ], + [ + 422, + 158 + ], + [ + 455, + 197 + ], + [ + 453, + 151 + ] + ], + [ + [ + 400, + 183 + ], + [ + 402, + 168 + ], + [ + 389, + 208 + ], + [ + 411, + 160 + ], + [ + 388, + 220 + ], + [ + 410, + 174 + ] + ], + [ + [ + 373, + 168 + ], + [ + 391, + 189 + ], + [ + 398, + 160 + ], + [ + 347, + 177 + ], + [ + 381, + 141 + ], + [ + 331, + 157 + ] + ], + [ + [ + 422, + 193 + ], + [ + 390, + 196 + ], + [ + 418, + 173 + ], + [ + 426, + 211 + ], + [ + 454, + 169 + ], + [ + 458, + 208 + ] + ] + ] + }, + { + "task_id": "0001488", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061330#876", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001488/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 381, + 158 + ], + [ + 361, + 139 + ], + [ + 362, + 179 + ], + [ + 401, + 136 + ], + [ + 382, + 198 + ], + [ + 423, + 155 + ] + ], + [ + [ + 420, + 156 + ], + [ + 439, + 141 + ], + [ + 410, + 133 + ], + [ + 429, + 178 + ], + [ + 390, + 148 + ], + [ + 410, + 194 + ] + ], + [ + [ + 410, + 152 + ], + [ + 409, + 137 + ], + [ + 397, + 173 + ], + [ + 422, + 131 + ], + [ + 398, + 186 + ], + [ + 422, + 145 + ] + ], + [ + [ + 403, + 162 + ], + [ + 388, + 148 + ], + [ + 379, + 175 + ], + [ + 426, + 149 + ], + [ + 394, + 190 + ], + [ + 443, + 163 + ] + ] + ] + }, + { + "task_id": "0001489", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061330#1025", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001489/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 414, + 107 + ], + [ + 391, + 115 + ], + [ + 414, + 86 + ], + [ + 415, + 126 + ], + [ + 438, + 77 + ], + [ + 437, + 118 + ] + ], + [ + [ + 421, + 131 + ], + [ + 397, + 126 + ], + [ + 429, + 114 + ], + [ + 412, + 149 + ], + [ + 454, + 119 + ], + [ + 437, + 156 + ] + ], + [ + [ + 419, + 140 + ], + [ + 446, + 133 + ], + [ + 408, + 132 + ], + [ + 429, + 146 + ], + [ + 382, + 139 + ], + [ + 405, + 152 + ] + ], + [ + [ + 428, + 135 + ], + [ + 424, + 122 + ], + [ + 418, + 158 + ], + [ + 438, + 112 + ], + [ + 421, + 169 + ], + [ + 440, + 125 + ] + ] + ] + }, + { + "task_id": "0001490", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0008_cfg_0006_104122064161#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001490/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 320, + 192 + ], + [ + 346, + 195 + ], + [ + 323, + 175 + ], + [ + 316, + 208 + ], + [ + 295, + 172 + ], + [ + 291, + 205 + ] + ], + [ + [ + 332, + 177 + ], + [ + 332, + 185 + ], + [ + 329, + 201 + ], + [ + 334, + 152 + ], + [ + 329, + 192 + ], + [ + 334, + 145 + ] + ], + [ + [ + 319, + 182 + ], + [ + 339, + 172 + ], + [ + 327, + 180 + ], + [ + 311, + 185 + ], + [ + 305, + 191 + ], + [ + 290, + 195 + ] + ], + [ + [ + 351, + 157 + ], + [ + 326, + 153 + ], + [ + 357, + 133 + ], + [ + 346, + 179 + ], + [ + 383, + 137 + ], + [ + 370, + 182 + ] + ] + ] + }, + { + "task_id": "0001492", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0008_cfg_0006_104122064161#304", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001492/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 325, + 173 + ], + [ + 330, + 163 + ], + [ + 298, + 167 + ], + [ + 352, + 179 + ], + [ + 290, + 177 + ], + [ + 347, + 191 + ] + ], + [ + [ + 325, + 159 + ], + [ + 329, + 177 + ], + [ + 300, + 166 + ], + [ + 349, + 151 + ], + [ + 295, + 147 + ], + [ + 346, + 132 + ] + ], + [ + [ + 336, + 197 + ], + [ + 311, + 198 + ], + [ + 336, + 219 + ], + [ + 337, + 174 + ], + [ + 360, + 217 + ], + [ + 362, + 173 + ] + ], + [ + [ + 323, + 182 + ], + [ + 324, + 186 + ], + [ + 323, + 207 + ], + [ + 323, + 155 + ], + [ + 322, + 202 + ], + [ + 322, + 153 + ] + ] + ] + }, + { + "task_id": "0001495", + "id": "RH20T_cfg5_task_0029_user_0010_scene_0006_cfg_0005_037522062165#176", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001495/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 282, + 250 + ], + [ + 251, + 249 + ], + [ + 277, + 260 + ], + [ + 287, + 242 + ], + [ + 311, + 261 + ], + [ + 316, + 243 + ] + ], + [ + [ + 277, + 214 + ], + [ + 276, + 194 + ], + [ + 267, + 191 + ], + [ + 289, + 238 + ], + [ + 269, + 210 + ], + [ + 290, + 255 + ] + ], + [ + [ + 300, + 196 + ], + [ + 317, + 208 + ], + [ + 281, + 188 + ], + [ + 321, + 204 + ], + [ + 262, + 175 + ], + [ + 303, + 191 + ] + ], + [ + [ + 282, + 174 + ], + [ + 283, + 204 + ], + [ + 255, + 179 + ], + [ + 311, + 170 + ], + [ + 256, + 150 + ], + [ + 310, + 140 + ] + ] + ] + }, + { + "task_id": "0001496", + "id": "RH20T_cfg5_task_0001_user_0010_scene_0001_cfg_0005_104422070044#84", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001496/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 400, + 172 + ], + [ + 417, + 169 + ], + [ + 403, + 192 + ], + [ + 396, + 153 + ], + [ + 385, + 195 + ], + [ + 378, + 156 + ] + ], + [ + [ + 383, + 191 + ], + [ + 392, + 201 + ], + [ + 378, + 207 + ], + [ + 388, + 174 + ], + [ + 369, + 197 + ], + [ + 379, + 164 + ] + ], + [ + [ + 402, + 187 + ], + [ + 402, + 173 + ], + [ + 418, + 197 + ], + [ + 387, + 178 + ], + [ + 417, + 210 + ], + [ + 387, + 192 + ] + ], + [ + [ + 397, + 190 + ], + [ + 399, + 181 + ], + [ + 410, + 202 + ], + [ + 384, + 177 + ], + [ + 408, + 213 + ], + [ + 381, + 186 + ] + ] + ] + }, + { + "task_id": "0001497", + "id": "RH20T_cfg6_task_0032_user_0020_scene_0009_cfg_0006_104122061018#203", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001497/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 242, + 33 + ], + [ + 256, + 28 + ], + [ + 231, + 56 + ], + [ + 252, + 10 + ], + [ + 218, + 59 + ], + [ + 239, + 15 + ] + ], + [ + [ + 263, + 70 + ], + [ + 259, + 96 + ], + [ + 264, + 66 + ], + [ + 262, + 74 + ], + [ + 268, + 41 + ], + [ + 266, + 46 + ] + ], + [ + [ + 240, + 49 + ], + [ + 237, + 70 + ], + [ + 264, + 45 + ], + [ + 218, + 52 + ], + [ + 266, + 24 + ], + [ + 221, + 32 + ] + ], + [ + [ + 246, + 59 + ], + [ + 247, + 37 + ], + [ + 241, + 44 + ], + [ + 250, + 76 + ], + [ + 240, + 65 + ], + [ + 248, + 97 + ] + ] + ] + }, + { + "task_id": "0001500", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0001_cfg_0004_036422060909#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001500/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 88 + ], + [ + 319, + 63 + ], + [ + 321, + 72 + ], + [ + 307, + 105 + ], + [ + 317, + 94 + ], + [ + 303, + 128 + ] + ], + [ + [ + 299, + 95 + ], + [ + 275, + 105 + ], + [ + 309, + 101 + ], + [ + 288, + 88 + ], + [ + 333, + 91 + ], + [ + 315, + 76 + ] + ], + [ + [ + 325, + 67 + ], + [ + 353, + 69 + ], + [ + 337, + 68 + ], + [ + 315, + 67 + ], + [ + 306, + 66 + ], + [ + 287, + 65 + ] + ], + [ + [ + 328, + 102 + ], + [ + 357, + 110 + ], + [ + 322, + 128 + ], + [ + 334, + 74 + ], + [ + 295, + 120 + ], + [ + 304, + 66 + ] + ] + ] + }, + { + "task_id": "0001501", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0004_cfg_0006_104122061602#273", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001501/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 536, + 192 + ], + [ + 547, + 217 + ], + [ + 517, + 200 + ], + [ + 557, + 184 + ], + [ + 506, + 177 + ], + [ + 545, + 159 + ] + ], + [ + [ + 549, + 174 + ], + [ + 558, + 167 + ], + [ + 532, + 193 + ], + [ + 565, + 156 + ], + [ + 525, + 198 + ], + [ + 555, + 163 + ] + ], + [ + [ + 529, + 177 + ], + [ + 547, + 196 + ], + [ + 552, + 163 + ], + [ + 507, + 191 + ], + [ + 532, + 144 + ], + [ + 489, + 173 + ] + ], + [ + [ + 500, + 182 + ], + [ + 492, + 205 + ], + [ + 488, + 180 + ], + [ + 511, + 183 + ], + [ + 495, + 155 + ], + [ + 519, + 160 + ] + ] + ] + }, + { + "task_id": "0001502", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0004_cfg_0006_104122061602#617", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001502/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 576, + 220 + ], + [ + 584, + 212 + ], + [ + 606, + 224 + ], + [ + 547, + 217 + ], + [ + 599, + 234 + ], + [ + 538, + 226 + ] + ], + [ + [ + 539, + 205 + ], + [ + 553, + 196 + ], + [ + 567, + 209 + ], + [ + 514, + 200 + ], + [ + 552, + 219 + ], + [ + 499, + 209 + ] + ], + [ + [ + 563, + 222 + ], + [ + 585, + 243 + ], + [ + 550, + 240 + ], + [ + 577, + 203 + ], + [ + 530, + 221 + ], + [ + 555, + 184 + ] + ], + [ + [ + 593, + 228 + ], + [ + 594, + 227 + ], + [ + 561, + 228 + ], + [ + 628, + 228 + ], + [ + 558, + 229 + ], + [ + 629, + 229 + ] + ] + ] + }, + { + "task_id": "0001503", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0004_cfg_0006_104122061602#974", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001503/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 452, + 185 + ], + [ + 457, + 165 + ], + [ + 449, + 168 + ], + [ + 456, + 204 + ], + [ + 445, + 186 + ], + [ + 451, + 222 + ] + ], + [ + [ + 503, + 214 + ], + [ + 519, + 197 + ], + [ + 517, + 234 + ], + [ + 488, + 194 + ], + [ + 501, + 252 + ], + [ + 471, + 212 + ] + ], + [ + [ + 485, + 200 + ], + [ + 496, + 225 + ], + [ + 469, + 207 + ], + [ + 504, + 191 + ], + [ + 459, + 184 + ], + [ + 492, + 166 + ] + ], + [ + [ + 481, + 185 + ], + [ + 487, + 164 + ], + [ + 455, + 185 + ], + [ + 510, + 185 + ], + [ + 449, + 206 + ], + [ + 505, + 208 + ] + ] + ] + }, + { + "task_id": "0001504", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0002_cfg_0006_104122061018#107", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001504/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 302, + 100 + ], + [ + 310, + 104 + ], + [ + 305, + 131 + ], + [ + 300, + 70 + ], + [ + 296, + 128 + ], + [ + 290, + 63 + ] + ], + [ + [ + 291, + 135 + ], + [ + 319, + 129 + ], + [ + 303, + 128 + ], + [ + 281, + 140 + ], + [ + 272, + 135 + ], + [ + 254, + 147 + ] + ], + [ + [ + 299, + 122 + ], + [ + 272, + 132 + ], + [ + 287, + 98 + ], + [ + 312, + 147 + ], + [ + 313, + 90 + ], + [ + 339, + 137 + ] + ], + [ + [ + 296, + 127 + ], + [ + 302, + 111 + ], + [ + 283, + 103 + ], + [ + 310, + 152 + ], + [ + 279, + 119 + ], + [ + 304, + 166 + ] + ] + ] + }, + { + "task_id": "0001506", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0002_cfg_0006_104122061018#418", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001506/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 367, + 132 + ], + [ + 355, + 126 + ], + [ + 340, + 149 + ], + [ + 393, + 114 + ], + [ + 352, + 157 + ], + [ + 409, + 120 + ] + ], + [ + [ + 352, + 154 + ], + [ + 381, + 144 + ], + [ + 344, + 122 + ], + [ + 360, + 185 + ], + [ + 314, + 132 + ], + [ + 331, + 195 + ] + ], + [ + [ + 353, + 145 + ], + [ + 354, + 120 + ], + [ + 339, + 128 + ], + [ + 368, + 163 + ], + [ + 339, + 150 + ], + [ + 367, + 185 + ] + ], + [ + [ + 316, + 140 + ], + [ + 328, + 166 + ], + [ + 326, + 132 + ], + [ + 304, + 149 + ], + [ + 314, + 106 + ], + [ + 291, + 120 + ] + ] + ] + }, + { + "task_id": "0001507", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0002_cfg_0006_104122061018#581", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001507/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 356, + 261 + ], + [ + 322, + 265 + ], + [ + 372, + 282 + ], + [ + 342, + 243 + ], + [ + 406, + 278 + ], + [ + 372, + 240 + ] + ], + [ + [ + 399, + 221 + ], + [ + 377, + 205 + ], + [ + 376, + 249 + ], + [ + 422, + 193 + ], + [ + 399, + 268 + ], + [ + 448, + 208 + ] + ], + [ + [ + 378, + 234 + ], + [ + 370, + 199 + ], + [ + 362, + 233 + ], + [ + 396, + 235 + ], + [ + 370, + 265 + ], + [ + 405, + 273 + ] + ], + [ + [ + 356, + 243 + ], + [ + 327, + 253 + ], + [ + 381, + 268 + ], + [ + 333, + 221 + ], + [ + 407, + 258 + ], + [ + 358, + 214 + ] + ] + ] + }, + { + "task_id": "0001508", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0002_cfg_0006_104122061018#727", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001508/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 381, + 179 + ], + [ + 378, + 148 + ], + [ + 411, + 180 + ], + [ + 349, + 178 + ], + [ + 415, + 211 + ], + [ + 352, + 211 + ] + ], + [ + [ + 350, + 193 + ], + [ + 378, + 178 + ], + [ + 346, + 163 + ], + [ + 354, + 221 + ], + [ + 315, + 179 + ], + [ + 326, + 237 + ] + ], + [ + [ + 368, + 148 + ], + [ + 392, + 148 + ], + [ + 348, + 150 + ], + [ + 390, + 145 + ], + [ + 322, + 150 + ], + [ + 362, + 145 + ] + ], + [ + [ + 344, + 183 + ], + [ + 351, + 157 + ], + [ + 328, + 164 + ], + [ + 363, + 204 + ], + [ + 323, + 187 + ], + [ + 355, + 228 + ] + ] + ] + }, + { + "task_id": "0001510", + "id": "RH20T_cfg2_task_0065_user_0007_scene_0012_cfg_0002_105422061350#177", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001510/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 175, + 135 + ], + [ + 156, + 132 + ], + [ + 169, + 158 + ], + [ + 182, + 112 + ], + [ + 189, + 161 + ], + [ + 203, + 114 + ] + ], + [ + [ + 190, + 118 + ], + [ + 190, + 101 + ], + [ + 202, + 103 + ], + [ + 176, + 134 + ], + [ + 202, + 119 + ], + [ + 177, + 150 + ] + ], + [ + [ + 211, + 99 + ], + [ + 202, + 121 + ], + [ + 210, + 94 + ], + [ + 213, + 105 + ], + [ + 219, + 72 + ], + [ + 223, + 81 + ] + ], + [ + [ + 168, + 137 + ], + [ + 167, + 116 + ], + [ + 180, + 126 + ], + [ + 157, + 147 + ], + [ + 181, + 149 + ], + [ + 158, + 169 + ] + ] + ] + }, + { + "task_id": "0001512", + "id": "RH20T_cfg6_task_0039_user_0014_scene_0003_cfg_0006_104122061602#171", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001512/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 408, + 88 + ], + [ + 382, + 92 + ], + [ + 406, + 72 + ], + [ + 410, + 103 + ], + [ + 434, + 68 + ], + [ + 435, + 99 + ] + ], + [ + [ + 391, + 115 + ], + [ + 398, + 96 + ], + [ + 386, + 129 + ], + [ + 397, + 100 + ], + [ + 379, + 148 + ], + [ + 389, + 121 + ] + ], + [ + [ + 413, + 126 + ], + [ + 432, + 127 + ], + [ + 421, + 147 + ], + [ + 405, + 104 + ], + [ + 403, + 145 + ], + [ + 387, + 104 + ] + ], + [ + [ + 403, + 121 + ], + [ + 408, + 102 + ], + [ + 418, + 138 + ], + [ + 389, + 106 + ], + [ + 412, + 156 + ], + [ + 385, + 124 + ] + ] + ] + }, + { + "task_id": "0001514", + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122060811#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001514/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 383, + 134 + ], + [ + 386, + 144 + ], + [ + 397, + 130 + ], + [ + 369, + 137 + ], + [ + 394, + 120 + ], + [ + 366, + 127 + ] + ], + [ + [ + 383, + 143 + ], + [ + 393, + 135 + ], + [ + 392, + 145 + ], + [ + 373, + 141 + ], + [ + 383, + 152 + ], + [ + 363, + 148 + ] + ], + [ + [ + 384, + 144 + ], + [ + 382, + 130 + ], + [ + 399, + 142 + ], + [ + 370, + 145 + ], + [ + 401, + 156 + ], + [ + 372, + 158 + ] + ], + [ + [ + 375, + 140 + ], + [ + 385, + 149 + ], + [ + 371, + 138 + ], + [ + 380, + 142 + ], + [ + 361, + 129 + ], + [ + 370, + 133 + ] + ] + ] + }, + { + "task_id": "0001515", + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122060811#177", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001515/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 378, + 153 + ], + [ + 389, + 145 + ], + [ + 388, + 155 + ], + [ + 369, + 151 + ], + [ + 378, + 162 + ], + [ + 359, + 158 + ] + ], + [ + [ + 367, + 151 + ], + [ + 359, + 147 + ], + [ + 372, + 138 + ], + [ + 362, + 165 + ], + [ + 380, + 143 + ], + [ + 370, + 169 + ] + ], + [ + [ + 378, + 149 + ], + [ + 374, + 160 + ], + [ + 385, + 142 + ], + [ + 370, + 155 + ], + [ + 389, + 131 + ], + [ + 374, + 143 + ] + ], + [ + [ + 367, + 157 + ], + [ + 358, + 168 + ], + [ + 367, + 155 + ], + [ + 367, + 158 + ], + [ + 376, + 144 + ], + [ + 376, + 146 + ] + ] + ] + }, + { + "task_id": "0001516", + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122060811#218", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001516/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 379, + 147 + ], + [ + 380, + 160 + ], + [ + 394, + 145 + ], + [ + 365, + 149 + ], + [ + 393, + 132 + ], + [ + 364, + 137 + ] + ], + [ + [ + 375, + 173 + ], + [ + 380, + 184 + ], + [ + 364, + 170 + ], + [ + 386, + 176 + ], + [ + 359, + 160 + ], + [ + 381, + 165 + ] + ], + [ + [ + 376, + 162 + ], + [ + 386, + 153 + ], + [ + 385, + 164 + ], + [ + 366, + 160 + ], + [ + 375, + 173 + ], + [ + 356, + 169 + ] + ], + [ + [ + 370, + 156 + ], + [ + 361, + 149 + ], + [ + 360, + 152 + ], + [ + 379, + 159 + ], + [ + 368, + 159 + ], + [ + 388, + 165 + ] + ] + ] + }, + { + "task_id": "0001517", + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122060811#236", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001517/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 356, + 174 + ], + [ + 354, + 177 + ], + [ + 363, + 162 + ], + [ + 350, + 187 + ], + [ + 365, + 158 + ], + [ + 352, + 185 + ] + ], + [ + [ + 373, + 172 + ], + [ + 384, + 163 + ], + [ + 382, + 174 + ], + [ + 364, + 170 + ], + [ + 372, + 182 + ], + [ + 353, + 178 + ] + ], + [ + [ + 369, + 156 + ], + [ + 365, + 145 + ], + [ + 373, + 145 + ], + [ + 365, + 166 + ], + [ + 376, + 156 + ], + [ + 368, + 177 + ] + ], + [ + [ + 365, + 169 + ], + [ + 368, + 182 + ], + [ + 377, + 164 + ], + [ + 352, + 174 + ], + [ + 374, + 151 + ], + [ + 349, + 160 + ] + ] + ] + }, + { + "task_id": "0001519", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_104122061850#432", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001519/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 327, + 112 + ], + [ + 323, + 96 + ], + [ + 305, + 108 + ], + [ + 350, + 117 + ], + [ + 309, + 122 + ], + [ + 353, + 131 + ] + ], + [ + [ + 333, + 111 + ], + [ + 314, + 104 + ], + [ + 352, + 107 + ], + [ + 315, + 115 + ], + [ + 369, + 114 + ], + [ + 332, + 122 + ] + ], + [ + [ + 325, + 89 + ], + [ + 328, + 78 + ], + [ + 342, + 73 + ], + [ + 307, + 105 + ], + [ + 338, + 84 + ], + [ + 305, + 115 + ] + ], + [ + [ + 316, + 102 + ], + [ + 306, + 120 + ], + [ + 294, + 95 + ], + [ + 340, + 108 + ], + [ + 304, + 78 + ], + [ + 350, + 91 + ] + ] + ] + }, + { + "task_id": "0001520", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_104122061850#511", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001520/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 332, + 107 + ], + [ + 312, + 121 + ], + [ + 335, + 116 + ], + [ + 329, + 99 + ], + [ + 356, + 100 + ], + [ + 348, + 84 + ] + ], + [ + [ + 349, + 152 + ], + [ + 325, + 150 + ], + [ + 350, + 171 + ], + [ + 348, + 134 + ], + [ + 376, + 172 + ], + [ + 372, + 135 + ] + ], + [ + [ + 322, + 110 + ], + [ + 326, + 94 + ], + [ + 326, + 129 + ], + [ + 318, + 92 + ], + [ + 322, + 144 + ], + [ + 314, + 107 + ] + ], + [ + [ + 330, + 129 + ], + [ + 327, + 114 + ], + [ + 309, + 121 + ], + [ + 352, + 137 + ], + [ + 313, + 134 + ], + [ + 354, + 150 + ] + ] + ] + }, + { + "task_id": "0001522", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_104122061850#685", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001522/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 303, + 158 + ], + [ + 297, + 144 + ], + [ + 283, + 148 + ], + [ + 324, + 167 + ], + [ + 289, + 161 + ], + [ + 329, + 179 + ] + ], + [ + [ + 322, + 146 + ], + [ + 338, + 126 + ], + [ + 305, + 130 + ], + [ + 339, + 161 + ], + [ + 288, + 150 + ], + [ + 324, + 180 + ] + ], + [ + [ + 306, + 170 + ], + [ + 330, + 172 + ], + [ + 313, + 165 + ], + [ + 299, + 175 + ], + [ + 291, + 163 + ], + [ + 274, + 173 + ] + ], + [ + [ + 334, + 149 + ], + [ + 350, + 137 + ], + [ + 332, + 166 + ], + [ + 336, + 130 + ], + [ + 315, + 179 + ], + [ + 318, + 143 + ] + ] + ] + }, + { + "task_id": "0001523", + "id": "RH20T_cfg5_task_0119_user_0007_scene_0002_cfg_0005_104122061850#177", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn off the water tap'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001523/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 401, + 35 + ], + [ + 396, + 17 + ], + [ + 383, + 50 + ], + [ + 419, + 20 + ], + [ + 388, + 65 + ], + [ + 422, + 36 + ] + ], + [ + [ + 417, + 15 + ], + [ + 407, + 8 + ], + [ + 433, + 31 + ], + [ + 402, + 0 + ], + [ + 441, + 36 + ], + [ + 412, + 6 + ] + ], + [ + [ + 405, + 45 + ], + [ + 384, + 37 + ], + [ + 398, + 67 + ], + [ + 412, + 22 + ], + [ + 418, + 74 + ], + [ + 432, + 29 + ] + ], + [ + [ + 401, + 0 + ], + [ + 405, + 9 + ], + [ + 377, + 7 + ], + [ + 425, + 0 + ], + [ + 374, + 0 + ], + [ + 420, + 0 + ] + ] + ] + }, + { + "task_id": "0001525", + "id": "RH20T_cfg5_task_0044_user_0010_scene_0002_cfg_0005_036422060215#261", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001525/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 288, + 137 + ], + [ + 280, + 123 + ], + [ + 291, + 120 + ], + [ + 285, + 153 + ], + [ + 300, + 134 + ], + [ + 293, + 167 + ] + ], + [ + [ + 263, + 99 + ], + [ + 260, + 120 + ], + [ + 254, + 95 + ], + [ + 273, + 104 + ], + [ + 257, + 74 + ], + [ + 277, + 81 + ] + ], + [ + [ + 291, + 137 + ], + [ + 273, + 126 + ], + [ + 291, + 130 + ], + [ + 291, + 143 + ], + [ + 309, + 141 + ], + [ + 308, + 154 + ] + ], + [ + [ + 279, + 122 + ], + [ + 283, + 104 + ], + [ + 298, + 120 + ], + [ + 259, + 123 + ], + [ + 294, + 136 + ], + [ + 256, + 140 + ] + ] + ] + }, + { + "task_id": "0001533", + "id": "RH20T_cfg5_task_0118_user_0007_scene_0009_cfg_0005_036422060215#215", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the water tap'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001533/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 108 + ], + [ + 319, + 93 + ], + [ + 332, + 111 + ], + [ + 294, + 106 + ], + [ + 327, + 125 + ], + [ + 290, + 121 + ] + ], + [ + [ + 313, + 115 + ], + [ + 295, + 117 + ], + [ + 305, + 108 + ], + [ + 322, + 122 + ], + [ + 322, + 106 + ], + [ + 340, + 120 + ] + ], + [ + [ + 300, + 101 + ], + [ + 297, + 89 + ], + [ + 284, + 91 + ], + [ + 315, + 110 + ], + [ + 286, + 104 + ], + [ + 318, + 123 + ] + ], + [ + [ + 308, + 96 + ], + [ + 324, + 107 + ], + [ + 317, + 78 + ], + [ + 300, + 113 + ], + [ + 301, + 67 + ], + [ + 285, + 103 + ] + ] + ] + }, + { + "task_id": "0001535", + "id": "RH20T_cfg3_task_0008_user_0016_scene_0005_cfg_0003_104122062823#2", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001535/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 437, + 111 + ], + [ + 457, + 102 + ], + [ + 416, + 122 + ], + [ + 460, + 99 + ], + [ + 396, + 131 + ], + [ + 439, + 108 + ] + ], + [ + [ + 403, + 144 + ], + [ + 420, + 128 + ], + [ + 430, + 149 + ], + [ + 379, + 141 + ], + [ + 412, + 167 + ], + [ + 361, + 157 + ] + ], + [ + [ + 428, + 138 + ], + [ + 435, + 124 + ], + [ + 431, + 166 + ], + [ + 425, + 108 + ], + [ + 424, + 183 + ], + [ + 416, + 122 + ] + ], + [ + [ + 411, + 115 + ], + [ + 414, + 90 + ], + [ + 441, + 124 + ], + [ + 383, + 105 + ], + [ + 437, + 147 + ], + [ + 381, + 127 + ] + ] + ] + }, + { + "task_id": "0001538", + "id": "RH20T_cfg4_task_0055_user_0010_scene_0005_cfg_0004_104122063550#95", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001538/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 103 + ], + [ + 262, + 130 + ], + [ + 241, + 106 + ], + [ + 293, + 99 + ], + [ + 245, + 80 + ], + [ + 297, + 73 + ] + ], + [ + [ + 259, + 102 + ], + [ + 269, + 115 + ], + [ + 269, + 74 + ], + [ + 249, + 129 + ], + [ + 258, + 64 + ], + [ + 240, + 117 + ] + ], + [ + [ + 264, + 123 + ], + [ + 252, + 113 + ], + [ + 272, + 149 + ], + [ + 257, + 98 + ], + [ + 283, + 156 + ], + [ + 268, + 108 + ] + ], + [ + [ + 289, + 121 + ], + [ + 284, + 147 + ], + [ + 272, + 131 + ], + [ + 307, + 110 + ], + [ + 276, + 108 + ], + [ + 312, + 85 + ] + ] + ] + }, + { + "task_id": "0001539", + "id": "RH20T_cfg5_task_0058_user_0007_scene_0002_cfg_0005_104422070044#101", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001539/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 413, + 190 + ], + [ + 399, + 203 + ], + [ + 417, + 197 + ], + [ + 408, + 183 + ], + [ + 430, + 185 + ], + [ + 422, + 171 + ] + ], + [ + [ + 407, + 188 + ], + [ + 418, + 179 + ], + [ + 408, + 175 + ], + [ + 405, + 201 + ], + [ + 396, + 184 + ], + [ + 393, + 210 + ] + ], + [ + [ + 414, + 181 + ], + [ + 420, + 171 + ], + [ + 429, + 192 + ], + [ + 399, + 170 + ], + [ + 423, + 201 + ], + [ + 394, + 180 + ] + ], + [ + [ + 409, + 168 + ], + [ + 424, + 164 + ], + [ + 419, + 171 + ], + [ + 400, + 166 + ], + [ + 402, + 175 + ], + [ + 384, + 170 + ] + ] + ] + }, + { + "task_id": "0001543", + "id": "RH20T_cfg3_task_0032_user_0011_scene_0002_cfg_0003_104122062823#282", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001543/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 182, + 114 + ], + [ + 170, + 88 + ], + [ + 192, + 125 + ], + [ + 172, + 104 + ], + [ + 203, + 150 + ], + [ + 183, + 128 + ] + ], + [ + [ + 206, + 120 + ], + [ + 213, + 100 + ], + [ + 180, + 125 + ], + [ + 233, + 115 + ], + [ + 171, + 147 + ], + [ + 226, + 138 + ] + ], + [ + [ + 179, + 127 + ], + [ + 179, + 124 + ], + [ + 185, + 156 + ], + [ + 173, + 98 + ], + [ + 185, + 157 + ], + [ + 174, + 103 + ] + ], + [ + [ + 187, + 123 + ], + [ + 209, + 142 + ], + [ + 175, + 149 + ], + [ + 200, + 97 + ], + [ + 154, + 131 + ], + [ + 178, + 79 + ] + ] + ] + }, + { + "task_id": "0001544", + "id": "RH20T_cfg3_task_0032_user_0011_scene_0002_cfg_0003_104122062823#367", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001544/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 263, + 118 + ], + [ + 248, + 89 + ], + [ + 285, + 123 + ], + [ + 243, + 114 + ], + [ + 298, + 151 + ], + [ + 256, + 140 + ] + ], + [ + [ + 262, + 131 + ], + [ + 273, + 102 + ], + [ + 244, + 110 + ], + [ + 278, + 149 + ], + [ + 230, + 142 + ], + [ + 267, + 178 + ] + ], + [ + [ + 243, + 129 + ], + [ + 220, + 102 + ], + [ + 271, + 106 + ], + [ + 215, + 151 + ], + [ + 294, + 133 + ], + [ + 238, + 179 + ] + ], + [ + [ + 280, + 126 + ], + [ + 254, + 146 + ], + [ + 298, + 153 + ], + [ + 262, + 100 + ], + [ + 325, + 133 + ], + [ + 288, + 80 + ] + ] + ] + }, + { + "task_id": "0001545", + "id": "RH20T_cfg1_task_0064_user_0015_scene_0010_cfg_0001_819112070044#144", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001545/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 325, + 195 + ], + [ + 329, + 175 + ], + [ + 330, + 186 + ], + [ + 319, + 204 + ], + [ + 326, + 204 + ], + [ + 315, + 224 + ] + ], + [ + [ + 330, + 194 + ], + [ + 345, + 193 + ], + [ + 346, + 196 + ], + [ + 315, + 193 + ], + [ + 330, + 198 + ], + [ + 299, + 195 + ] + ], + [ + [ + 336, + 191 + ], + [ + 358, + 194 + ], + [ + 333, + 213 + ], + [ + 339, + 170 + ], + [ + 312, + 210 + ], + [ + 318, + 167 + ] + ], + [ + [ + 334, + 188 + ], + [ + 327, + 171 + ], + [ + 354, + 183 + ], + [ + 313, + 192 + ], + [ + 362, + 200 + ], + [ + 320, + 210 + ] + ] + ] + }, + { + "task_id": "0001547", + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122061330#149", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001547/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 329, + 182 + ], + [ + 356, + 189 + ], + [ + 330, + 168 + ], + [ + 328, + 195 + ], + [ + 301, + 161 + ], + [ + 302, + 188 + ] + ], + [ + [ + 327, + 230 + ], + [ + 336, + 203 + ], + [ + 336, + 231 + ], + [ + 318, + 229 + ], + [ + 328, + 257 + ], + [ + 309, + 258 + ] + ], + [ + [ + 303, + 201 + ], + [ + 293, + 203 + ], + [ + 296, + 229 + ], + [ + 311, + 173 + ], + [ + 305, + 226 + ], + [ + 320, + 173 + ] + ], + [ + [ + 309, + 215 + ], + [ + 311, + 206 + ], + [ + 338, + 204 + ], + [ + 280, + 226 + ], + [ + 338, + 213 + ], + [ + 275, + 236 + ] + ] + ] + }, + { + "task_id": "0001548", + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122061330#188", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001548/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 357, + 176 + ], + [ + 386, + 182 + ], + [ + 361, + 150 + ], + [ + 353, + 204 + ], + [ + 333, + 145 + ], + [ + 324, + 198 + ] + ], + [ + [ + 325, + 173 + ], + [ + 297, + 167 + ], + [ + 331, + 173 + ], + [ + 318, + 172 + ], + [ + 358, + 179 + ], + [ + 349, + 178 + ] + ], + [ + [ + 343, + 161 + ], + [ + 334, + 159 + ], + [ + 334, + 188 + ], + [ + 353, + 134 + ], + [ + 342, + 189 + ], + [ + 360, + 138 + ] + ], + [ + [ + 350, + 180 + ], + [ + 336, + 179 + ], + [ + 336, + 202 + ], + [ + 366, + 157 + ], + [ + 350, + 203 + ], + [ + 378, + 160 + ] + ] + ] + }, + { + "task_id": "0001550", + "id": "RH20T_cfg6_task_0078_user_0014_scene_0008_cfg_0006_104122063633#111", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close the microwave door'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001550/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 188, + 76 + ], + [ + 183, + 52 + ], + [ + 189, + 62 + ], + [ + 186, + 90 + ], + [ + 193, + 83 + ], + [ + 191, + 112 + ] + ], + [ + [ + 211, + 79 + ], + [ + 192, + 65 + ], + [ + 225, + 78 + ], + [ + 198, + 79 + ], + [ + 243, + 92 + ], + [ + 216, + 92 + ] + ], + [ + [ + 168, + 100 + ], + [ + 190, + 115 + ], + [ + 161, + 102 + ], + [ + 176, + 98 + ], + [ + 139, + 87 + ], + [ + 152, + 81 + ] + ], + [ + [ + 188, + 70 + ], + [ + 189, + 61 + ], + [ + 173, + 91 + ], + [ + 202, + 49 + ], + [ + 171, + 102 + ], + [ + 202, + 58 + ] + ] + ] + }, + { + "task_id": "0001551", + "id": "RH20T_cfg7_task_0063_user_0020_scene_0008_cfg_0007_104122061330#145", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a shovel to scoop up an object'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001551/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 351, + 89 + ], + [ + 371, + 99 + ], + [ + 339, + 107 + ], + [ + 364, + 71 + ], + [ + 319, + 97 + ], + [ + 343, + 60 + ] + ], + [ + [ + 329, + 138 + ], + [ + 349, + 146 + ], + [ + 315, + 142 + ], + [ + 343, + 135 + ], + [ + 296, + 134 + ], + [ + 325, + 128 + ] + ], + [ + [ + 352, + 113 + ], + [ + 338, + 123 + ], + [ + 354, + 134 + ], + [ + 349, + 90 + ], + [ + 366, + 124 + ], + [ + 362, + 81 + ] + ], + [ + [ + 344, + 115 + ], + [ + 346, + 91 + ], + [ + 338, + 119 + ], + [ + 352, + 111 + ], + [ + 336, + 142 + ], + [ + 351, + 137 + ] + ] + ] + }, + { + "task_id": "0001555", + "id": "RH20T_cfg6_task_0038_user_0020_scene_0010_cfg_0006_104122061330#333", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001555/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 227 + ], + [ + 363, + 203 + ], + [ + 326, + 228 + ], + [ + 366, + 226 + ], + [ + 310, + 251 + ], + [ + 351, + 248 + ] + ], + [ + [ + 327, + 223 + ], + [ + 316, + 223 + ], + [ + 316, + 252 + ], + [ + 339, + 195 + ], + [ + 327, + 250 + ], + [ + 348, + 197 + ] + ], + [ + [ + 295, + 243 + ], + [ + 299, + 269 + ], + [ + 264, + 249 + ], + [ + 326, + 236 + ], + [ + 262, + 226 + ], + [ + 321, + 213 + ] + ], + [ + [ + 353, + 201 + ], + [ + 345, + 185 + ], + [ + 379, + 185 + ], + [ + 327, + 216 + ], + [ + 389, + 202 + ], + [ + 334, + 234 + ] + ] + ] + }, + { + "task_id": "0001558", + "id": "RH20T_cfg2_task_0002_user_0013_scene_0006_cfg_0002_104122063678#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001558/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 444, + 272 + ], + [ + 424, + 284 + ], + [ + 457, + 301 + ], + [ + 432, + 245 + ], + [ + 480, + 289 + ], + [ + 453, + 232 + ] + ], + [ + [ + 464, + 268 + ], + [ + 457, + 289 + ], + [ + 493, + 279 + ], + [ + 436, + 258 + ], + [ + 500, + 256 + ], + [ + 441, + 236 + ] + ], + [ + [ + 486, + 270 + ], + [ + 478, + 266 + ], + [ + 514, + 260 + ], + [ + 458, + 280 + ], + [ + 526, + 264 + ], + [ + 465, + 286 + ] + ], + [ + [ + 445, + 259 + ], + [ + 452, + 244 + ], + [ + 451, + 284 + ], + [ + 439, + 236 + ], + [ + 444, + 297 + ], + [ + 433, + 250 + ] + ] + ] + }, + { + "task_id": "0001561", + "id": "RH20T_cfg2_task_0057_user_0016_scene_0006_cfg_0002_037522062165#225", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001561/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 207, + 227 + ], + [ + 202, + 229 + ], + [ + 230, + 211 + ], + [ + 183, + 242 + ], + [ + 233, + 210 + ], + [ + 189, + 239 + ] + ], + [ + [ + 234, + 274 + ], + [ + 252, + 271 + ], + [ + 253, + 288 + ], + [ + 216, + 261 + ], + [ + 235, + 292 + ], + [ + 196, + 264 + ] + ], + [ + [ + 240, + 239 + ], + [ + 238, + 223 + ], + [ + 264, + 229 + ], + [ + 215, + 251 + ], + [ + 265, + 244 + ], + [ + 219, + 266 + ] + ], + [ + [ + 226, + 247 + ], + [ + 234, + 260 + ], + [ + 251, + 232 + ], + [ + 201, + 262 + ], + [ + 243, + 216 + ], + [ + 190, + 248 + ] + ] + ] + }, + { + "task_id": "0001564", + "id": "RH20T_cfg2_task_0057_user_0016_scene_0006_cfg_0002_037522062165#616", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001564/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 230, + 302 + ], + [ + 226, + 299 + ], + [ + 208, + 324 + ], + [ + 252, + 281 + ], + [ + 211, + 328 + ], + [ + 258, + 282 + ] + ], + [ + [ + 257, + 306 + ], + [ + 251, + 292 + ], + [ + 279, + 289 + ], + [ + 233, + 325 + ], + [ + 284, + 303 + ], + [ + 240, + 338 + ] + ], + [ + [ + 228, + 312 + ], + [ + 225, + 287 + ], + [ + 216, + 302 + ], + [ + 240, + 322 + ], + [ + 219, + 331 + ], + [ + 244, + 349 + ] + ], + [ + [ + 241, + 290 + ], + [ + 236, + 270 + ], + [ + 240, + 270 + ], + [ + 242, + 311 + ], + [ + 245, + 289 + ], + [ + 247, + 330 + ] + ] + ] + }, + { + "task_id": "0001565", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_750612070851#368", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001565/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 308 + ], + [ + 346, + 272 + ], + [ + 353, + 313 + ], + [ + 325, + 303 + ], + [ + 344, + 353 + ], + [ + 318, + 338 + ] + ], + [ + [ + 291, + 315 + ], + [ + 307, + 280 + ], + [ + 292, + 317 + ], + [ + 290, + 313 + ], + [ + 276, + 351 + ], + [ + 272, + 353 + ] + ], + [ + [ + 358, + 290 + ], + [ + 325, + 298 + ], + [ + 357, + 329 + ], + [ + 358, + 255 + ], + [ + 394, + 322 + ], + [ + 392, + 246 + ] + ], + [ + [ + 326, + 325 + ], + [ + 325, + 309 + ], + [ + 288, + 319 + ], + [ + 365, + 331 + ], + [ + 291, + 333 + ], + [ + 363, + 344 + ] + ] + ] + }, + { + "task_id": "0001566", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_750612070851#902", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001566/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 345 + ], + [ + 328, + 316 + ], + [ + 298, + 345 + ], + [ + 373, + 346 + ], + [ + 305, + 360 + ], + [ + 377, + 360 + ] + ], + [ + [ + 372, + 331 + ], + [ + 374, + 314 + ], + [ + 333, + 323 + ], + [ + 411, + 339 + ], + [ + 333, + 339 + ], + [ + 406, + 353 + ] + ], + [ + [ + 396, + 343 + ], + [ + 429, + 355 + ], + [ + 423, + 325 + ], + [ + 365, + 360 + ], + [ + 394, + 317 + ], + [ + 337, + 350 + ] + ], + [ + [ + 362, + 346 + ], + [ + 338, + 318 + ], + [ + 398, + 324 + ], + [ + 327, + 360 + ], + [ + 420, + 352 + ], + [ + 351, + 360 + ] + ] + ] + }, + { + "task_id": "0001569", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_750612070851#1730", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001569/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 507, + 341 + ], + [ + 550, + 341 + ], + [ + 522, + 360 + ], + [ + 495, + 318 + ], + [ + 476, + 360 + ], + [ + 457, + 318 + ] + ], + [ + [ + 477, + 334 + ], + [ + 497, + 331 + ], + [ + 513, + 353 + ], + [ + 442, + 316 + ], + [ + 493, + 354 + ], + [ + 428, + 320 + ] + ], + [ + [ + 506, + 352 + ], + [ + 524, + 317 + ], + [ + 539, + 353 + ], + [ + 478, + 351 + ], + [ + 519, + 360 + ], + [ + 458, + 360 + ] + ], + [ + [ + 469, + 344 + ], + [ + 479, + 329 + ], + [ + 428, + 334 + ], + [ + 511, + 353 + ], + [ + 421, + 347 + ], + [ + 498, + 360 + ] + ] + ] + }, + { + "task_id": "0001571", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0006_cfg_0004_038522062288#159", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001571/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 271, + 144 + ], + [ + 264, + 119 + ], + [ + 246, + 129 + ], + [ + 298, + 161 + ], + [ + 253, + 151 + ], + [ + 303, + 183 + ] + ], + [ + [ + 282, + 119 + ], + [ + 305, + 98 + ], + [ + 267, + 126 + ], + [ + 298, + 111 + ], + [ + 243, + 148 + ], + [ + 272, + 135 + ] + ], + [ + [ + 268, + 134 + ], + [ + 295, + 145 + ], + [ + 287, + 112 + ], + [ + 251, + 154 + ], + [ + 258, + 97 + ], + [ + 222, + 143 + ] + ], + [ + [ + 275, + 132 + ], + [ + 297, + 115 + ], + [ + 253, + 106 + ], + [ + 297, + 158 + ], + [ + 228, + 123 + ], + [ + 275, + 178 + ] + ] + ] + }, + { + "task_id": "0001573", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0003_cfg_0003_104122062295#276", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001573/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 260, + 151 + ], + [ + 249, + 180 + ], + [ + 278, + 180 + ], + [ + 239, + 119 + ], + [ + 286, + 153 + ], + [ + 251, + 93 + ] + ], + [ + [ + 237, + 155 + ], + [ + 273, + 140 + ], + [ + 235, + 169 + ], + [ + 239, + 144 + ], + [ + 196, + 185 + ], + [ + 206, + 158 + ] + ], + [ + [ + 217, + 219 + ], + [ + 216, + 183 + ], + [ + 170, + 228 + ], + [ + 260, + 211 + ], + [ + 173, + 261 + ], + [ + 259, + 243 + ] + ], + [ + [ + 229, + 169 + ], + [ + 230, + 131 + ], + [ + 208, + 149 + ], + [ + 254, + 191 + ], + [ + 208, + 182 + ], + [ + 251, + 226 + ] + ] + ] + }, + { + "task_id": "0001574", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0003_cfg_0003_104122062295#311", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001574/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 247, + 214 + ], + [ + 289, + 193 + ], + [ + 236, + 164 + ], + [ + 258, + 259 + ], + [ + 185, + 185 + ], + [ + 213, + 285 + ] + ], + [ + [ + 304, + 260 + ], + [ + 310, + 221 + ], + [ + 274, + 231 + ], + [ + 338, + 293 + ], + [ + 270, + 265 + ], + [ + 330, + 327 + ] + ], + [ + [ + 294, + 211 + ], + [ + 310, + 216 + ], + [ + 286, + 258 + ], + [ + 303, + 165 + ], + [ + 266, + 257 + ], + [ + 285, + 154 + ] + ], + [ + [ + 268, + 283 + ], + [ + 253, + 246 + ], + [ + 315, + 280 + ], + [ + 224, + 285 + ], + [ + 327, + 314 + ], + [ + 240, + 317 + ] + ] + ] + }, + { + "task_id": "0001576", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104122062823#296", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001576/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 364, + 160 + ], + [ + 338, + 181 + ], + [ + 368, + 164 + ], + [ + 360, + 157 + ], + [ + 396, + 141 + ], + [ + 384, + 137 + ] + ], + [ + [ + 330, + 121 + ], + [ + 296, + 114 + ], + [ + 318, + 130 + ], + [ + 341, + 114 + ], + [ + 357, + 138 + ], + [ + 375, + 121 + ] + ], + [ + [ + 350, + 133 + ], + [ + 347, + 164 + ], + [ + 370, + 124 + ], + [ + 333, + 142 + ], + [ + 374, + 93 + ], + [ + 337, + 114 + ] + ], + [ + [ + 363, + 123 + ], + [ + 362, + 97 + ], + [ + 397, + 120 + ], + [ + 329, + 127 + ], + [ + 396, + 143 + ], + [ + 332, + 150 + ] + ] + ] + }, + { + "task_id": "0001581", + "id": "RH20T_cfg5_task_0008_user_0007_scene_0001_cfg_0005_105422061350#325", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001581/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 341, + 40 + ], + [ + 335, + 56 + ], + [ + 358, + 29 + ], + [ + 324, + 51 + ], + [ + 365, + 11 + ], + [ + 330, + 33 + ] + ], + [ + [ + 348, + 32 + ], + [ + 356, + 13 + ], + [ + 357, + 19 + ], + [ + 338, + 47 + ], + [ + 350, + 36 + ], + [ + 332, + 64 + ] + ], + [ + [ + 354, + 8 + ], + [ + 339, + 25 + ], + [ + 348, + 0 + ], + [ + 360, + 28 + ], + [ + 363, + 0 + ], + [ + 374, + 12 + ] + ], + [ + [ + 341, + 38 + ], + [ + 332, + 61 + ], + [ + 360, + 44 + ], + [ + 323, + 33 + ], + [ + 369, + 20 + ], + [ + 331, + 10 + ] + ] + ] + }, + { + "task_id": "0001582", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0002_cfg_0006_104122061602#247", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001582/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 486, + 212 + ], + [ + 511, + 216 + ], + [ + 482, + 190 + ], + [ + 491, + 233 + ], + [ + 459, + 187 + ], + [ + 468, + 228 + ] + ], + [ + [ + 480, + 207 + ], + [ + 497, + 197 + ], + [ + 493, + 232 + ], + [ + 467, + 184 + ], + [ + 476, + 240 + ], + [ + 452, + 194 + ] + ], + [ + [ + 475, + 185 + ], + [ + 465, + 178 + ], + [ + 488, + 209 + ], + [ + 463, + 164 + ], + [ + 497, + 215 + ], + [ + 472, + 172 + ] + ], + [ + [ + 468, + 213 + ], + [ + 468, + 220 + ], + [ + 464, + 237 + ], + [ + 473, + 189 + ], + [ + 465, + 231 + ], + [ + 474, + 179 + ] + ] + ] + }, + { + "task_id": "0001583", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0002_cfg_0006_104122061602#402", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001583/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 518, + 219 + ], + [ + 538, + 202 + ], + [ + 508, + 197 + ], + [ + 528, + 240 + ], + [ + 486, + 214 + ], + [ + 508, + 257 + ] + ], + [ + [ + 511, + 207 + ], + [ + 488, + 213 + ], + [ + 494, + 186 + ], + [ + 529, + 229 + ], + [ + 515, + 181 + ], + [ + 551, + 222 + ] + ], + [ + [ + 496, + 229 + ], + [ + 512, + 217 + ], + [ + 513, + 254 + ], + [ + 479, + 205 + ], + [ + 496, + 263 + ], + [ + 465, + 216 + ] + ], + [ + [ + 524, + 247 + ], + [ + 552, + 256 + ], + [ + 535, + 244 + ], + [ + 513, + 250 + ], + [ + 505, + 234 + ], + [ + 487, + 241 + ] + ] + ] + }, + { + "task_id": "0001584", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0005_cfg_0004_104122062823#235", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001584/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 290, + 63 + ], + [ + 320, + 63 + ], + [ + 301, + 84 + ], + [ + 280, + 43 + ], + [ + 269, + 84 + ], + [ + 250, + 42 + ] + ], + [ + [ + 314, + 104 + ], + [ + 312, + 74 + ], + [ + 316, + 88 + ], + [ + 311, + 122 + ], + [ + 318, + 115 + ], + [ + 312, + 151 + ] + ], + [ + [ + 343, + 90 + ], + [ + 380, + 96 + ], + [ + 341, + 116 + ], + [ + 345, + 61 + ], + [ + 308, + 110 + ], + [ + 307, + 55 + ] + ], + [ + [ + 306, + 85 + ], + [ + 274, + 80 + ], + [ + 301, + 102 + ], + [ + 310, + 70 + ], + [ + 335, + 108 + ], + [ + 341, + 75 + ] + ] + ] + }, + { + "task_id": "0001585", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0005_cfg_0004_104122062823#437", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001585/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 289, + 71 + ], + [ + 263, + 81 + ], + [ + 304, + 92 + ], + [ + 276, + 51 + ], + [ + 329, + 83 + ], + [ + 301, + 43 + ] + ], + [ + [ + 290, + 78 + ], + [ + 305, + 91 + ], + [ + 294, + 50 + ], + [ + 286, + 104 + ], + [ + 280, + 38 + ], + [ + 273, + 91 + ] + ], + [ + [ + 295, + 77 + ], + [ + 287, + 51 + ], + [ + 320, + 81 + ], + [ + 271, + 74 + ], + [ + 327, + 106 + ], + [ + 279, + 98 + ] + ], + [ + [ + 318, + 68 + ], + [ + 312, + 66 + ], + [ + 314, + 37 + ], + [ + 321, + 97 + ], + [ + 321, + 37 + ], + [ + 328, + 101 + ] + ] + ] + }, + { + "task_id": "0001587", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0002_cfg_0001_038522062547#292", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001587/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 466, + 65 + ], + [ + 456, + 76 + ], + [ + 465, + 41 + ], + [ + 467, + 89 + ], + [ + 477, + 27 + ], + [ + 480, + 78 + ] + ], + [ + [ + 451, + 88 + ], + [ + 458, + 71 + ], + [ + 455, + 110 + ], + [ + 447, + 66 + ], + [ + 449, + 124 + ], + [ + 442, + 81 + ] + ], + [ + [ + 475, + 60 + ], + [ + 468, + 78 + ], + [ + 496, + 49 + ], + [ + 454, + 70 + ], + [ + 505, + 28 + ], + [ + 460, + 49 + ] + ], + [ + [ + 456, + 89 + ], + [ + 436, + 108 + ], + [ + 440, + 71 + ], + [ + 472, + 107 + ], + [ + 460, + 51 + ], + [ + 492, + 89 + ] + ] + ] + }, + { + "task_id": "0001588", + "id": "RH20T_cfg1_task_0001_user_0001_scene_0002_cfg_0001_039422060546#205", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001588/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 252, + 149 + ], + [ + 243, + 127 + ], + [ + 240, + 177 + ], + [ + 263, + 123 + ], + [ + 249, + 195 + ], + [ + 270, + 143 + ] + ], + [ + [ + 230, + 124 + ], + [ + 197, + 129 + ], + [ + 227, + 94 + ], + [ + 233, + 155 + ], + [ + 259, + 89 + ], + [ + 266, + 150 + ] + ], + [ + [ + 250, + 108 + ], + [ + 271, + 125 + ], + [ + 220, + 118 + ], + [ + 278, + 99 + ], + [ + 202, + 102 + ], + [ + 259, + 84 + ] + ], + [ + [ + 272, + 180 + ], + [ + 296, + 157 + ], + [ + 246, + 155 + ], + [ + 299, + 206 + ], + [ + 220, + 180 + ], + [ + 274, + 231 + ] + ] + ] + }, + { + "task_id": "0001589", + "id": "RH20T_cfg7_task_0004_user_0020_scene_0005_cfg_0007_104122060811#145", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001589/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 408, + 135 + ], + [ + 399, + 125 + ], + [ + 402, + 145 + ], + [ + 414, + 125 + ], + [ + 410, + 155 + ], + [ + 423, + 135 + ] + ], + [ + [ + 414, + 119 + ], + [ + 401, + 115 + ], + [ + 415, + 108 + ], + [ + 413, + 129 + ], + [ + 429, + 111 + ], + [ + 426, + 133 + ] + ], + [ + [ + 406, + 130 + ], + [ + 416, + 122 + ], + [ + 416, + 130 + ], + [ + 396, + 129 + ], + [ + 407, + 137 + ], + [ + 388, + 137 + ] + ], + [ + [ + 391, + 137 + ], + [ + 400, + 140 + ], + [ + 380, + 142 + ], + [ + 402, + 132 + ], + [ + 372, + 139 + ], + [ + 393, + 130 + ] + ] + ] + }, + { + "task_id": "0001593", + "id": "RH20T_cfg5_task_0016_user_0010_scene_0001_cfg_0005_037522062165#225", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001593/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 280, + 188 + ], + [ + 280, + 173 + ], + [ + 257, + 177 + ], + [ + 305, + 200 + ], + [ + 258, + 190 + ], + [ + 304, + 212 + ] + ], + [ + [ + 276, + 185 + ], + [ + 293, + 176 + ], + [ + 266, + 160 + ], + [ + 286, + 209 + ], + [ + 246, + 169 + ], + [ + 268, + 220 + ] + ], + [ + [ + 271, + 170 + ], + [ + 289, + 168 + ], + [ + 257, + 153 + ], + [ + 286, + 188 + ], + [ + 238, + 155 + ], + [ + 266, + 191 + ] + ], + [ + [ + 272, + 170 + ], + [ + 285, + 188 + ], + [ + 267, + 156 + ], + [ + 276, + 183 + ], + [ + 254, + 139 + ], + [ + 265, + 166 + ] + ] + ] + }, + { + "task_id": "0001594", + "id": "RH20T_cfg6_task_0026_user_0014_scene_0007_cfg_0006_104122063633#141", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push and close the drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001594/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 201, + 171 + ], + [ + 174, + 184 + ], + [ + 191, + 156 + ], + [ + 212, + 188 + ], + [ + 216, + 144 + ], + [ + 239, + 174 + ] + ], + [ + [ + 210, + 184 + ], + [ + 206, + 159 + ], + [ + 218, + 168 + ], + [ + 201, + 200 + ], + [ + 221, + 190 + ], + [ + 205, + 223 + ] + ], + [ + [ + 227, + 184 + ], + [ + 253, + 181 + ], + [ + 221, + 196 + ], + [ + 233, + 173 + ], + [ + 194, + 198 + ], + [ + 208, + 175 + ] + ], + [ + [ + 228, + 158 + ], + [ + 218, + 179 + ], + [ + 205, + 141 + ], + [ + 251, + 174 + ], + [ + 215, + 121 + ], + [ + 259, + 154 + ] + ] + ] + }, + { + "task_id": "0001595", + "id": "RH20T_cfg1_task_0076_user_0010_scene_0001_cfg_0001_750612070851#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) one on top of the other every time'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001595/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 282, + 360 + ], + [ + 285, + 351 + ], + [ + 314, + 360 + ], + [ + 247, + 360 + ], + [ + 313, + 360 + ], + [ + 241, + 360 + ] + ], + [ + [ + 304, + 317 + ], + [ + 277, + 292 + ], + [ + 279, + 340 + ], + [ + 326, + 297 + ], + [ + 309, + 360 + ], + [ + 353, + 321 + ] + ], + [ + [ + 303, + 337 + ], + [ + 303, + 318 + ], + [ + 264, + 334 + ], + [ + 343, + 340 + ], + [ + 267, + 351 + ], + [ + 341, + 356 + ] + ], + [ + [ + 295, + 334 + ], + [ + 272, + 360 + ], + [ + 330, + 356 + ], + [ + 263, + 315 + ], + [ + 350, + 320 + ], + [ + 285, + 283 + ] + ] + ] + }, + { + "task_id": "0001598", + "id": "RH20T_cfg4_task_0046_user_0014_scene_0015_cfg_0004_038522062288#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001598/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 170, + 146 + ], + [ + 155, + 114 + ], + [ + 181, + 142 + ], + [ + 156, + 150 + ], + [ + 195, + 171 + ], + [ + 172, + 183 + ] + ], + [ + [ + 190, + 157 + ], + [ + 190, + 141 + ], + [ + 161, + 171 + ], + [ + 217, + 143 + ], + [ + 163, + 184 + ], + [ + 216, + 157 + ] + ], + [ + [ + 217, + 145 + ], + [ + 198, + 121 + ], + [ + 218, + 129 + ], + [ + 215, + 159 + ], + [ + 238, + 154 + ], + [ + 232, + 182 + ] + ], + [ + [ + 189, + 153 + ], + [ + 170, + 172 + ], + [ + 177, + 159 + ], + [ + 199, + 147 + ], + [ + 197, + 138 + ], + [ + 217, + 128 + ] + ] + ] + }, + { + "task_id": "0001600", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_039422060546#101", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001600/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 328, + 227 + ], + [ + 288, + 225 + ], + [ + 318, + 244 + ], + [ + 340, + 208 + ], + [ + 355, + 246 + ], + [ + 381, + 211 + ] + ], + [ + [ + 392, + 229 + ], + [ + 377, + 200 + ], + [ + 372, + 206 + ], + [ + 415, + 255 + ], + [ + 386, + 231 + ], + [ + 429, + 281 + ] + ], + [ + [ + 360, + 269 + ], + [ + 365, + 249 + ], + [ + 351, + 309 + ], + [ + 369, + 233 + ], + [ + 347, + 324 + ], + [ + 364, + 251 + ] + ], + [ + [ + 331, + 256 + ], + [ + 336, + 300 + ], + [ + 329, + 250 + ], + [ + 332, + 263 + ], + [ + 324, + 209 + ], + [ + 326, + 214 + ] + ] + ] + }, + { + "task_id": "0001604", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_039422060546#446", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001604/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 384, + 295 + ], + [ + 340, + 286 + ], + [ + 382, + 297 + ], + [ + 387, + 292 + ], + [ + 424, + 305 + ], + [ + 437, + 301 + ] + ], + [ + [ + 368, + 239 + ], + [ + 390, + 274 + ], + [ + 369, + 243 + ], + [ + 367, + 236 + ], + [ + 345, + 205 + ], + [ + 347, + 204 + ] + ], + [ + [ + 362, + 271 + ], + [ + 376, + 231 + ], + [ + 320, + 259 + ], + [ + 407, + 284 + ], + [ + 303, + 301 + ], + [ + 393, + 331 + ] + ], + [ + [ + 368, + 270 + ], + [ + 372, + 254 + ], + [ + 360, + 312 + ], + [ + 375, + 232 + ], + [ + 356, + 323 + ], + [ + 371, + 248 + ] + ] + ] + }, + { + "task_id": "0001608", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0007_cfg_0006_104122064161#178", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001608/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 345, + 158 + ], + [ + 320, + 157 + ], + [ + 352, + 152 + ], + [ + 338, + 165 + ], + [ + 376, + 152 + ], + [ + 365, + 166 + ] + ], + [ + [ + 333, + 178 + ], + [ + 335, + 179 + ], + [ + 330, + 204 + ], + [ + 336, + 152 + ], + [ + 329, + 202 + ], + [ + 335, + 154 + ] + ], + [ + [ + 317, + 165 + ], + [ + 305, + 145 + ], + [ + 301, + 181 + ], + [ + 332, + 149 + ], + [ + 313, + 200 + ], + [ + 343, + 169 + ] + ], + [ + [ + 324, + 172 + ], + [ + 298, + 176 + ], + [ + 321, + 165 + ], + [ + 328, + 180 + ], + [ + 345, + 161 + ], + [ + 355, + 175 + ] + ] + ] + }, + { + "task_id": "0001610", + "id": "RH20T_cfg7_task_0020_user_0014_scene_0010_cfg_0007_104122061330#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001610/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 323, + 157 + ], + [ + 304, + 153 + ], + [ + 312, + 178 + ], + [ + 334, + 136 + ], + [ + 332, + 183 + ], + [ + 354, + 140 + ] + ], + [ + [ + 349, + 178 + ], + [ + 376, + 178 + ], + [ + 346, + 168 + ], + [ + 352, + 186 + ], + [ + 318, + 168 + ], + [ + 327, + 186 + ] + ], + [ + [ + 341, + 162 + ], + [ + 328, + 179 + ], + [ + 340, + 180 + ], + [ + 341, + 143 + ], + [ + 352, + 165 + ], + [ + 353, + 128 + ] + ], + [ + [ + 334, + 126 + ], + [ + 309, + 131 + ], + [ + 340, + 150 + ], + [ + 329, + 102 + ], + [ + 364, + 145 + ], + [ + 353, + 97 + ] + ] + ] + }, + { + "task_id": "0001611", + "id": "RH20T_cfg7_task_0020_user_0014_scene_0010_cfg_0007_104122061330#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001611/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 341, + 163 + ], + [ + 328, + 179 + ], + [ + 341, + 180 + ], + [ + 342, + 145 + ], + [ + 353, + 165 + ], + [ + 355, + 130 + ] + ], + [ + [ + 335, + 186 + ], + [ + 332, + 187 + ], + [ + 334, + 160 + ], + [ + 336, + 211 + ], + [ + 337, + 156 + ], + [ + 340, + 211 + ] + ], + [ + [ + 361, + 152 + ], + [ + 363, + 160 + ], + [ + 336, + 146 + ], + [ + 386, + 158 + ], + [ + 333, + 136 + ], + [ + 385, + 149 + ] + ], + [ + [ + 337, + 164 + ], + [ + 327, + 156 + ], + [ + 338, + 141 + ], + [ + 335, + 186 + ], + [ + 349, + 148 + ], + [ + 345, + 195 + ] + ] + ] + }, + { + "task_id": "0001613", + "id": "RH20T_cfg4_task_0019_user_0016_scene_0005_cfg_0004_104122062295#494", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the first move as black on the 3-4 point in the upper right corner of the Go board'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001613/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 0, + 235 + ], + [ + 0, + 276 + ], + [ + 0, + 262 + ], + [ + 0, + 212 + ], + [ + 56, + 221 + ], + [ + 11, + 180 + ] + ], + [ + [ + 53, + 240 + ], + [ + 92, + 206 + ], + [ + 11, + 226 + ], + [ + 99, + 255 + ], + [ + 0, + 265 + ], + [ + 55, + 299 + ] + ], + [ + [ + 85, + 262 + ], + [ + 31, + 237 + ], + [ + 113, + 214 + ], + [ + 57, + 309 + ], + [ + 163, + 239 + ], + [ + 109, + 330 + ] + ], + [ + [ + 11, + 233 + ], + [ + 0, + 186 + ], + [ + 0, + 207 + ], + [ + 44, + 262 + ], + [ + 0, + 247 + ], + [ + 58, + 302 + ] + ] + ] + }, + { + "task_id": "0001615", + "id": "RH20T_cfg2_task_0128_user_0007_scene_0003_cfg_0002_104122061850#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the USB drive to the docking station'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001615/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 433, + 123 + ], + [ + 458, + 132 + ], + [ + 439, + 103 + ], + [ + 427, + 142 + ], + [ + 413, + 95 + ], + [ + 404, + 134 + ] + ], + [ + [ + 426, + 140 + ], + [ + 425, + 136 + ], + [ + 428, + 167 + ], + [ + 423, + 114 + ], + [ + 429, + 173 + ], + [ + 424, + 117 + ] + ], + [ + [ + 451, + 153 + ], + [ + 429, + 143 + ], + [ + 455, + 132 + ], + [ + 447, + 172 + ], + [ + 479, + 143 + ], + [ + 470, + 183 + ] + ], + [ + [ + 443, + 136 + ], + [ + 443, + 119 + ], + [ + 431, + 155 + ], + [ + 455, + 117 + ], + [ + 432, + 170 + ], + [ + 455, + 133 + ] + ] + ] + }, + { + "task_id": "0001617", + "id": "RH20T_cfg1_task_0070_user_0015_scene_0002_cfg_0001_038522062547#314", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001617/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 408, + 145 + ], + [ + 399, + 168 + ], + [ + 386, + 137 + ], + [ + 429, + 153 + ], + [ + 395, + 113 + ], + [ + 437, + 131 + ] + ], + [ + [ + 443, + 142 + ], + [ + 450, + 124 + ], + [ + 427, + 126 + ], + [ + 457, + 157 + ], + [ + 419, + 145 + ], + [ + 450, + 175 + ] + ], + [ + [ + 430, + 146 + ], + [ + 437, + 134 + ], + [ + 436, + 169 + ], + [ + 424, + 123 + ], + [ + 429, + 178 + ], + [ + 417, + 135 + ] + ], + [ + [ + 409, + 153 + ], + [ + 405, + 178 + ], + [ + 429, + 160 + ], + [ + 389, + 146 + ], + [ + 433, + 137 + ], + [ + 393, + 122 + ] + ] + ] + }, + { + "task_id": "0001618", + "id": "RH20T_cfg1_task_0001_user_0014_scene_0006_cfg_0001_750612070851#236", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001618/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 483, + 189 + ], + [ + 450, + 194 + ], + [ + 478, + 219 + ], + [ + 489, + 156 + ], + [ + 509, + 213 + ], + [ + 522, + 151 + ] + ], + [ + [ + 467, + 221 + ], + [ + 462, + 189 + ], + [ + 444, + 216 + ], + [ + 489, + 225 + ], + [ + 449, + 250 + ], + [ + 495, + 257 + ] + ], + [ + [ + 485, + 181 + ], + [ + 505, + 203 + ], + [ + 502, + 193 + ], + [ + 466, + 167 + ], + [ + 484, + 174 + ], + [ + 449, + 148 + ] + ], + [ + [ + 467, + 205 + ], + [ + 477, + 179 + ], + [ + 433, + 209 + ], + [ + 499, + 201 + ], + [ + 425, + 233 + ], + [ + 488, + 224 + ] + ] + ] + }, + { + "task_id": "0001620", + "id": "RH20T_cfg1_task_0016_user_0014_scene_0008_cfg_0001_750612070851#237", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001620/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 276 + ], + [ + 324, + 282 + ], + [ + 314, + 247 + ], + [ + 282, + 306 + ], + [ + 290, + 242 + ], + [ + 259, + 299 + ] + ], + [ + [ + 310, + 283 + ], + [ + 318, + 263 + ], + [ + 316, + 318 + ], + [ + 305, + 252 + ], + [ + 309, + 334 + ], + [ + 299, + 270 + ] + ], + [ + [ + 294, + 279 + ], + [ + 264, + 303 + ], + [ + 297, + 264 + ], + [ + 291, + 296 + ], + [ + 326, + 242 + ], + [ + 325, + 270 + ] + ], + [ + [ + 277, + 294 + ], + [ + 248, + 294 + ], + [ + 257, + 281 + ], + [ + 294, + 305 + ], + [ + 289, + 279 + ], + [ + 324, + 305 + ] + ] + ] + }, + { + "task_id": "0001621", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_105422061350#257", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001621/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 225, + 38 + ], + [ + 221, + 15 + ], + [ + 230, + 19 + ], + [ + 219, + 59 + ], + [ + 233, + 40 + ], + [ + 223, + 79 + ] + ], + [ + [ + 232, + 72 + ], + [ + 246, + 97 + ], + [ + 207, + 85 + ], + [ + 256, + 59 + ], + [ + 193, + 60 + ], + [ + 243, + 34 + ] + ], + [ + [ + 210, + 50 + ], + [ + 236, + 53 + ], + [ + 216, + 26 + ], + [ + 203, + 75 + ], + [ + 191, + 24 + ], + [ + 177, + 72 + ] + ], + [ + [ + 236, + 65 + ], + [ + 211, + 51 + ], + [ + 233, + 56 + ], + [ + 239, + 73 + ], + [ + 260, + 71 + ], + [ + 263, + 86 + ] + ] + ] + }, + { + "task_id": "0001623", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_105422061350#458", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001623/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 181, + 84 + ], + [ + 176, + 51 + ], + [ + 195, + 83 + ], + [ + 169, + 85 + ], + [ + 200, + 117 + ], + [ + 174, + 116 + ] + ], + [ + [ + 172, + 94 + ], + [ + 178, + 126 + ], + [ + 182, + 101 + ], + [ + 162, + 86 + ], + [ + 177, + 72 + ], + [ + 156, + 53 + ] + ], + [ + [ + 194, + 104 + ], + [ + 191, + 80 + ], + [ + 213, + 87 + ], + [ + 175, + 123 + ], + [ + 215, + 109 + ], + [ + 179, + 144 + ] + ], + [ + [ + 189, + 130 + ], + [ + 188, + 119 + ], + [ + 219, + 136 + ], + [ + 157, + 123 + ], + [ + 222, + 148 + ], + [ + 157, + 134 + ] + ] + ] + }, + { + "task_id": "0001625", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_750612070853#168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001625/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 140, + 234 + ], + [ + 181, + 217 + ], + [ + 127, + 241 + ], + [ + 154, + 226 + ], + [ + 83, + 260 + ], + [ + 103, + 248 + ] + ], + [ + [ + 156, + 129 + ], + [ + 136, + 172 + ], + [ + 113, + 110 + ], + [ + 199, + 148 + ], + [ + 132, + 67 + ], + [ + 218, + 105 + ] + ], + [ + [ + 121, + 200 + ], + [ + 151, + 159 + ], + [ + 103, + 184 + ], + [ + 141, + 218 + ], + [ + 74, + 223 + ], + [ + 108, + 263 + ] + ], + [ + [ + 146, + 181 + ], + [ + 124, + 142 + ], + [ + 130, + 164 + ], + [ + 164, + 200 + ], + [ + 150, + 197 + ], + [ + 184, + 236 + ] + ] + ] + }, + { + "task_id": "0001626", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_750612070853#261", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001626/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 144, + 183 + ], + [ + 121, + 150 + ], + [ + 101, + 203 + ], + [ + 182, + 165 + ], + [ + 126, + 239 + ], + [ + 206, + 197 + ] + ], + [ + [ + 160, + 153 + ], + [ + 173, + 171 + ], + [ + 200, + 119 + ], + [ + 118, + 188 + ], + [ + 189, + 94 + ], + [ + 96, + 169 + ] + ], + [ + [ + 151, + 158 + ], + [ + 124, + 137 + ], + [ + 118, + 141 + ], + [ + 186, + 176 + ], + [ + 142, + 159 + ], + [ + 207, + 193 + ] + ], + [ + [ + 146, + 180 + ], + [ + 128, + 141 + ], + [ + 124, + 164 + ], + [ + 170, + 198 + ], + [ + 140, + 198 + ], + [ + 185, + 234 + ] + ] + ] + }, + { + "task_id": "0001627", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_750612070853#348", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001627/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 150, + 176 + ], + [ + 131, + 136 + ], + [ + 127, + 163 + ], + [ + 176, + 192 + ], + [ + 144, + 197 + ], + [ + 192, + 228 + ] + ], + [ + [ + 192, + 184 + ], + [ + 151, + 212 + ], + [ + 219, + 194 + ], + [ + 170, + 174 + ], + [ + 258, + 166 + ], + [ + 206, + 151 + ] + ], + [ + [ + 150, + 157 + ], + [ + 158, + 172 + ], + [ + 120, + 200 + ], + [ + 181, + 114 + ], + [ + 116, + 184 + ], + [ + 172, + 105 + ] + ], + [ + [ + 194, + 188 + ], + [ + 195, + 147 + ], + [ + 240, + 185 + ], + [ + 144, + 191 + ], + [ + 237, + 220 + ], + [ + 146, + 229 + ] + ] + ] + }, + { + "task_id": "0001629", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_750612070853#525", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001629/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 183, + 194 + ], + [ + 188, + 233 + ], + [ + 185, + 169 + ], + [ + 180, + 225 + ], + [ + 180, + 128 + ], + [ + 174, + 180 + ] + ], + [ + [ + 167, + 188 + ], + [ + 134, + 193 + ], + [ + 142, + 216 + ], + [ + 189, + 164 + ], + [ + 180, + 212 + ], + [ + 223, + 158 + ] + ], + [ + [ + 146, + 181 + ], + [ + 127, + 143 + ], + [ + 116, + 169 + ], + [ + 180, + 194 + ], + [ + 134, + 202 + ], + [ + 196, + 229 + ] + ], + [ + [ + 149, + 140 + ], + [ + 171, + 119 + ], + [ + 115, + 156 + ], + [ + 186, + 122 + ], + [ + 86, + 182 + ], + [ + 162, + 147 + ] + ] + ] + }, + { + "task_id": "0001630", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_039422060546#85", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001630/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 246, + 218 + ], + [ + 279, + 225 + ], + [ + 249, + 212 + ], + [ + 241, + 224 + ], + [ + 218, + 206 + ], + [ + 205, + 216 + ] + ], + [ + [ + 239, + 203 + ], + [ + 253, + 217 + ], + [ + 269, + 182 + ], + [ + 208, + 224 + ], + [ + 254, + 165 + ], + [ + 189, + 211 + ] + ], + [ + [ + 239, + 205 + ], + [ + 261, + 183 + ], + [ + 249, + 239 + ], + [ + 230, + 173 + ], + [ + 227, + 259 + ], + [ + 210, + 194 + ] + ], + [ + [ + 247, + 199 + ], + [ + 236, + 178 + ], + [ + 234, + 231 + ], + [ + 260, + 170 + ], + [ + 245, + 248 + ], + [ + 269, + 189 + ] + ] + ] + }, + { + "task_id": "0001631", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_039422060546#184", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001631/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 258, + 140 + ], + [ + 285, + 130 + ], + [ + 274, + 163 + ], + [ + 244, + 119 + ], + [ + 245, + 175 + ], + [ + 216, + 130 + ] + ], + [ + [ + 260, + 129 + ], + [ + 231, + 126 + ], + [ + 250, + 103 + ], + [ + 268, + 152 + ], + [ + 282, + 106 + ], + [ + 298, + 156 + ] + ], + [ + [ + 274, + 116 + ], + [ + 264, + 92 + ], + [ + 263, + 140 + ], + [ + 284, + 94 + ], + [ + 272, + 161 + ], + [ + 291, + 116 + ] + ], + [ + [ + 300, + 96 + ], + [ + 288, + 127 + ], + [ + 278, + 88 + ], + [ + 324, + 104 + ], + [ + 290, + 58 + ], + [ + 338, + 71 + ] + ] + ] + }, + { + "task_id": "0001632", + "id": "RH20T_cfg1_task_0119_user_0014_scene_0004_cfg_0001_750612070851#98", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn off the water tap'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001632/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 432, + 330 + ], + [ + 474, + 333 + ], + [ + 422, + 355 + ], + [ + 444, + 302 + ], + [ + 381, + 353 + ], + [ + 399, + 300 + ] + ], + [ + [ + 456, + 309 + ], + [ + 459, + 287 + ], + [ + 415, + 324 + ], + [ + 497, + 294 + ], + [ + 415, + 342 + ], + [ + 492, + 313 + ] + ], + [ + [ + 443, + 307 + ], + [ + 434, + 280 + ], + [ + 480, + 316 + ], + [ + 405, + 298 + ], + [ + 494, + 348 + ], + [ + 413, + 329 + ] + ], + [ + [ + 411, + 303 + ], + [ + 433, + 338 + ], + [ + 389, + 287 + ], + [ + 430, + 317 + ], + [ + 370, + 255 + ], + [ + 410, + 287 + ] + ] + ] + }, + { + "task_id": "0001633", + "id": "RH20T_cfg2_task_0080_user_0014_scene_0006_cfg_0002_105422061350#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Flip over and spread out the paper that is laid flat on the table'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001633/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 223, + 109 + ], + [ + 202, + 121 + ], + [ + 212, + 90 + ], + [ + 234, + 126 + ], + [ + 234, + 77 + ], + [ + 254, + 114 + ] + ], + [ + [ + 220, + 110 + ], + [ + 213, + 136 + ], + [ + 238, + 115 + ], + [ + 203, + 105 + ], + [ + 246, + 88 + ], + [ + 210, + 80 + ] + ], + [ + [ + 198, + 114 + ], + [ + 195, + 97 + ], + [ + 223, + 118 + ], + [ + 172, + 109 + ], + [ + 225, + 133 + ], + [ + 176, + 125 + ] + ], + [ + [ + 199, + 114 + ], + [ + 199, + 138 + ], + [ + 220, + 109 + ], + [ + 178, + 119 + ], + [ + 220, + 86 + ], + [ + 178, + 96 + ] + ] + ] + }, + { + "task_id": "0001635", + "id": "RH20T_cfg6_task_0008_user_0014_scene_0007_cfg_0006_104122063633#171", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001635/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 213, + 155 + ], + [ + 213, + 137 + ], + [ + 215, + 140 + ], + [ + 210, + 170 + ], + [ + 214, + 156 + ], + [ + 209, + 187 + ] + ], + [ + [ + 187, + 167 + ], + [ + 173, + 150 + ], + [ + 168, + 177 + ], + [ + 205, + 159 + ], + [ + 183, + 195 + ], + [ + 219, + 176 + ] + ], + [ + [ + 216, + 176 + ], + [ + 238, + 174 + ], + [ + 222, + 185 + ], + [ + 210, + 167 + ], + [ + 197, + 187 + ], + [ + 188, + 169 + ] + ], + [ + [ + 225, + 172 + ], + [ + 233, + 187 + ], + [ + 234, + 155 + ], + [ + 216, + 188 + ], + [ + 226, + 140 + ], + [ + 209, + 173 + ] + ] + ] + }, + { + "task_id": "0001637", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_037522062165#189", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001637/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 198, + 170 + ], + [ + 169, + 185 + ], + [ + 203, + 184 + ], + [ + 193, + 158 + ], + [ + 233, + 169 + ], + [ + 220, + 145 + ] + ], + [ + [ + 219, + 149 + ], + [ + 236, + 128 + ], + [ + 200, + 129 + ], + [ + 238, + 167 + ], + [ + 182, + 151 + ], + [ + 222, + 188 + ] + ], + [ + [ + 190, + 145 + ], + [ + 188, + 128 + ], + [ + 166, + 128 + ], + [ + 215, + 162 + ], + [ + 169, + 143 + ], + [ + 215, + 176 + ] + ], + [ + [ + 174, + 157 + ], + [ + 184, + 147 + ], + [ + 199, + 178 + ], + [ + 148, + 135 + ], + [ + 190, + 190 + ], + [ + 135, + 144 + ] + ] + ] + }, + { + "task_id": "0001642", + "id": "RH20T_cfg1_task_0072_user_0010_scene_0010_cfg_0001_035622060973#566", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drop coins into a piggy bank'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001642/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 140, + 101 + ], + [ + 142, + 83 + ], + [ + 142, + 93 + ], + [ + 137, + 109 + ], + [ + 140, + 112 + ], + [ + 135, + 127 + ] + ], + [ + [ + 161, + 100 + ], + [ + 152, + 116 + ], + [ + 153, + 105 + ], + [ + 169, + 95 + ], + [ + 162, + 90 + ], + [ + 178, + 79 + ] + ], + [ + [ + 150, + 112 + ], + [ + 164, + 111 + ], + [ + 135, + 101 + ], + [ + 164, + 123 + ], + [ + 120, + 103 + ], + [ + 150, + 124 + ] + ], + [ + [ + 137, + 115 + ], + [ + 148, + 120 + ], + [ + 152, + 127 + ], + [ + 122, + 104 + ], + [ + 140, + 122 + ], + [ + 109, + 99 + ] + ] + ] + }, + { + "task_id": "0001643", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0002_cfg_0006_104122061330#287", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001643/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 374, + 244 + ], + [ + 343, + 240 + ], + [ + 372, + 231 + ], + [ + 376, + 258 + ], + [ + 401, + 234 + ], + [ + 409, + 262 + ] + ], + [ + [ + 341, + 257 + ], + [ + 327, + 281 + ], + [ + 341, + 271 + ], + [ + 340, + 240 + ], + [ + 353, + 250 + ], + [ + 354, + 217 + ] + ], + [ + [ + 326, + 281 + ], + [ + 322, + 257 + ], + [ + 336, + 298 + ], + [ + 317, + 266 + ], + [ + 340, + 322 + ], + [ + 320, + 288 + ] + ], + [ + [ + 335, + 237 + ], + [ + 320, + 213 + ], + [ + 330, + 228 + ], + [ + 340, + 247 + ], + [ + 343, + 250 + ], + [ + 355, + 271 + ] + ] + ] + }, + { + "task_id": "0001644", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0002_cfg_0006_104122061330#728", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001644/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 346, + 273 + ], + [ + 365, + 297 + ], + [ + 346, + 275 + ], + [ + 347, + 271 + ], + [ + 326, + 250 + ], + [ + 330, + 249 + ] + ], + [ + [ + 289, + 285 + ], + [ + 264, + 290 + ], + [ + 280, + 257 + ], + [ + 299, + 313 + ], + [ + 307, + 251 + ], + [ + 326, + 309 + ] + ], + [ + [ + 281, + 300 + ], + [ + 266, + 315 + ], + [ + 285, + 327 + ], + [ + 275, + 272 + ], + [ + 298, + 312 + ], + [ + 289, + 260 + ] + ], + [ + [ + 309, + 280 + ], + [ + 294, + 308 + ], + [ + 314, + 285 + ], + [ + 303, + 274 + ], + [ + 327, + 260 + ], + [ + 318, + 246 + ] + ] + ] + }, + { + "task_id": "0001645", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0002_cfg_0006_104122061330#1015", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001645/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 395, + 174 + ], + [ + 384, + 198 + ], + [ + 396, + 184 + ], + [ + 393, + 163 + ], + [ + 406, + 162 + ], + [ + 405, + 139 + ] + ], + [ + [ + 372, + 179 + ], + [ + 391, + 192 + ], + [ + 353, + 193 + ], + [ + 391, + 166 + ], + [ + 335, + 180 + ], + [ + 373, + 155 + ] + ], + [ + [ + 397, + 193 + ], + [ + 398, + 219 + ], + [ + 425, + 192 + ], + [ + 369, + 194 + ], + [ + 425, + 165 + ], + [ + 368, + 167 + ] + ], + [ + [ + 378, + 185 + ], + [ + 375, + 200 + ], + [ + 364, + 164 + ], + [ + 392, + 205 + ], + [ + 368, + 150 + ], + [ + 395, + 191 + ] + ] + ] + }, + { + "task_id": "0001647", + "id": "RH20T_cfg5_task_0008_user_0010_scene_0010_cfg_0005_036422060215#301", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001647/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 336, + 130 + ], + [ + 319, + 134 + ], + [ + 331, + 143 + ], + [ + 341, + 119 + ], + [ + 350, + 139 + ], + [ + 359, + 115 + ] + ], + [ + [ + 358, + 154 + ], + [ + 346, + 141 + ], + [ + 371, + 148 + ], + [ + 345, + 159 + ], + [ + 383, + 161 + ], + [ + 358, + 173 + ] + ], + [ + [ + 341, + 135 + ], + [ + 352, + 141 + ], + [ + 326, + 141 + ], + [ + 356, + 129 + ], + [ + 314, + 135 + ], + [ + 345, + 123 + ] + ], + [ + [ + 347, + 144 + ], + [ + 353, + 128 + ], + [ + 363, + 141 + ], + [ + 331, + 147 + ], + [ + 357, + 156 + ], + [ + 326, + 162 + ] + ] + ] + }, + { + "task_id": "0001649", + "id": "RH20T_cfg3_task_0091_user_0010_scene_0002_cfg_0003_036422060909#155", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001649/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 185, + 133 + ], + [ + 206, + 136 + ], + [ + 199, + 115 + ], + [ + 170, + 151 + ], + [ + 179, + 112 + ], + [ + 150, + 147 + ] + ], + [ + [ + 208, + 172 + ], + [ + 198, + 153 + ], + [ + 208, + 158 + ], + [ + 207, + 185 + ], + [ + 219, + 179 + ], + [ + 217, + 204 + ] + ], + [ + [ + 212, + 148 + ], + [ + 208, + 131 + ], + [ + 234, + 136 + ], + [ + 189, + 160 + ], + [ + 236, + 151 + ], + [ + 194, + 175 + ] + ], + [ + [ + 190, + 156 + ], + [ + 180, + 141 + ], + [ + 193, + 138 + ], + [ + 188, + 174 + ], + [ + 204, + 154 + ], + [ + 199, + 190 + ] + ] + ] + }, + { + "task_id": "0001651", + "id": "RH20T_cfg3_task_0091_user_0010_scene_0002_cfg_0003_036422060909#505", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001651/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 201, + 213 + ], + [ + 205, + 210 + ], + [ + 196, + 242 + ], + [ + 205, + 185 + ], + [ + 193, + 243 + ], + [ + 201, + 188 + ] + ], + [ + [ + 209, + 233 + ], + [ + 183, + 225 + ], + [ + 195, + 242 + ], + [ + 220, + 224 + ], + [ + 223, + 251 + ], + [ + 245, + 231 + ] + ], + [ + [ + 193, + 216 + ], + [ + 194, + 192 + ], + [ + 218, + 219 + ], + [ + 166, + 214 + ], + [ + 218, + 244 + ], + [ + 165, + 240 + ] + ], + [ + [ + 201, + 203 + ], + [ + 198, + 181 + ], + [ + 214, + 186 + ], + [ + 187, + 221 + ], + [ + 216, + 206 + ], + [ + 191, + 241 + ] + ] + ] + }, + { + "task_id": "0001652", + "id": "RH20T_cfg6_task_0051_user_0014_scene_0002_cfg_0006_104122064161#125", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the dish off the dish rack'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001652/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 382, + 250 + ], + [ + 362, + 260 + ], + [ + 358, + 243 + ], + [ + 404, + 257 + ], + [ + 379, + 231 + ], + [ + 426, + 247 + ] + ], + [ + [ + 425, + 233 + ], + [ + 440, + 249 + ], + [ + 449, + 234 + ], + [ + 399, + 231 + ], + [ + 435, + 219 + ], + [ + 386, + 216 + ] + ], + [ + [ + 379, + 208 + ], + [ + 399, + 219 + ], + [ + 384, + 228 + ], + [ + 374, + 189 + ], + [ + 363, + 216 + ], + [ + 354, + 177 + ] + ], + [ + [ + 395, + 206 + ], + [ + 398, + 183 + ], + [ + 410, + 226 + ], + [ + 382, + 188 + ], + [ + 407, + 247 + ], + [ + 381, + 209 + ] + ] + ] + }, + { + "task_id": "0001653", + "id": "RH20T_cfg6_task_0002_user_0014_scene_0006_cfg_0006_037522061512#232", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001653/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 223, + 129 + ], + [ + 219, + 110 + ], + [ + 237, + 118 + ], + [ + 208, + 141 + ], + [ + 240, + 136 + ], + [ + 212, + 160 + ] + ], + [ + [ + 216, + 138 + ], + [ + 214, + 114 + ], + [ + 238, + 137 + ], + [ + 195, + 140 + ], + [ + 240, + 161 + ], + [ + 197, + 163 + ] + ], + [ + [ + 248, + 128 + ], + [ + 266, + 132 + ], + [ + 237, + 118 + ], + [ + 258, + 137 + ], + [ + 218, + 114 + ], + [ + 240, + 132 + ] + ], + [ + [ + 209, + 144 + ], + [ + 224, + 150 + ], + [ + 215, + 122 + ], + [ + 204, + 165 + ], + [ + 201, + 117 + ], + [ + 190, + 158 + ] + ] + ] + }, + { + "task_id": "0001655", + "id": "RH20T_cfg1_task_0096_user_0015_scene_0005_cfg_0001_038522063145#221", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001655/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 430, + 62 + ], + [ + 415, + 65 + ], + [ + 448, + 74 + ], + [ + 413, + 50 + ], + [ + 462, + 71 + ], + [ + 428, + 48 + ] + ], + [ + [ + 451, + 49 + ], + [ + 444, + 72 + ], + [ + 459, + 61 + ], + [ + 443, + 39 + ], + [ + 467, + 36 + ], + [ + 449, + 15 + ] + ], + [ + [ + 432, + 44 + ], + [ + 410, + 44 + ], + [ + 442, + 33 + ], + [ + 420, + 55 + ], + [ + 465, + 33 + ], + [ + 444, + 56 + ] + ], + [ + [ + 431, + 49 + ], + [ + 438, + 27 + ], + [ + 453, + 58 + ], + [ + 410, + 40 + ], + [ + 446, + 78 + ], + [ + 405, + 61 + ] + ] + ] + }, + { + "task_id": "0001656", + "id": "RH20T_cfg1_task_0096_user_0015_scene_0005_cfg_0001_038522063145#284", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001656/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 430, + 62 + ], + [ + 438, + 42 + ], + [ + 452, + 71 + ], + [ + 409, + 53 + ], + [ + 445, + 89 + ], + [ + 404, + 72 + ] + ], + [ + [ + 433, + 39 + ], + [ + 448, + 40 + ], + [ + 432, + 14 + ], + [ + 434, + 62 + ], + [ + 418, + 14 + ], + [ + 420, + 60 + ] + ], + [ + [ + 429, + 41 + ], + [ + 450, + 46 + ], + [ + 419, + 54 + ], + [ + 440, + 28 + ], + [ + 399, + 49 + ], + [ + 417, + 23 + ] + ], + [ + [ + 443, + 77 + ], + [ + 458, + 58 + ], + [ + 439, + 74 + ], + [ + 448, + 80 + ], + [ + 426, + 91 + ], + [ + 433, + 99 + ] + ] + ] + }, + { + "task_id": "0001657", + "id": "RH20T_cfg4_task_0074_user_0015_scene_0006_cfg_0004_038522062288#112", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button horizontally'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001657/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 110, + 107 + ], + [ + 132, + 100 + ], + [ + 93, + 112 + ], + [ + 129, + 101 + ], + [ + 69, + 118 + ], + [ + 104, + 108 + ] + ], + [ + [ + 121, + 107 + ], + [ + 118, + 90 + ], + [ + 93, + 98 + ], + [ + 149, + 116 + ], + [ + 97, + 114 + ], + [ + 150, + 130 + ] + ], + [ + [ + 146, + 105 + ], + [ + 133, + 116 + ], + [ + 143, + 78 + ], + [ + 148, + 130 + ], + [ + 155, + 69 + ], + [ + 159, + 119 + ] + ], + [ + [ + 154, + 111 + ], + [ + 150, + 136 + ], + [ + 127, + 104 + ], + [ + 179, + 118 + ], + [ + 131, + 78 + ], + [ + 182, + 94 + ] + ] + ] + }, + { + "task_id": "0001658", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0001_cfg_0004_036422060909#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001658/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 372, + 58 + ], + [ + 394, + 49 + ], + [ + 365, + 35 + ], + [ + 380, + 80 + ], + [ + 344, + 44 + ], + [ + 359, + 87 + ] + ], + [ + [ + 377, + 61 + ], + [ + 381, + 44 + ], + [ + 367, + 48 + ], + [ + 389, + 76 + ], + [ + 364, + 64 + ], + [ + 385, + 91 + ] + ], + [ + [ + 369, + 40 + ], + [ + 378, + 40 + ], + [ + 378, + 60 + ], + [ + 360, + 20 + ], + [ + 369, + 61 + ], + [ + 350, + 19 + ] + ], + [ + [ + 398, + 57 + ], + [ + 377, + 56 + ], + [ + 402, + 71 + ], + [ + 395, + 45 + ], + [ + 424, + 72 + ], + [ + 416, + 46 + ] + ] + ] + }, + { + "task_id": "0001659", + "id": "RH20T_cfg1_task_0005_user_0015_scene_0007_cfg_0001_038522062547#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001659/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 397, + 140 + ], + [ + 418, + 150 + ], + [ + 388, + 124 + ], + [ + 404, + 154 + ], + [ + 368, + 115 + ], + [ + 385, + 145 + ] + ], + [ + [ + 409, + 127 + ], + [ + 433, + 114 + ], + [ + 405, + 144 + ], + [ + 413, + 111 + ], + [ + 382, + 155 + ], + [ + 391, + 123 + ] + ], + [ + [ + 378, + 170 + ], + [ + 404, + 165 + ], + [ + 380, + 155 + ], + [ + 376, + 184 + ], + [ + 352, + 159 + ], + [ + 351, + 188 + ] + ], + [ + [ + 390, + 157 + ], + [ + 390, + 140 + ], + [ + 398, + 179 + ], + [ + 382, + 137 + ], + [ + 398, + 194 + ], + [ + 383, + 153 + ] + ] + ] + }, + { + "task_id": "0001664", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_038522063145#150", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001664/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 420, + 177 + ], + [ + 436, + 167 + ], + [ + 401, + 167 + ], + [ + 439, + 187 + ], + [ + 384, + 176 + ], + [ + 421, + 198 + ] + ], + [ + [ + 418, + 180 + ], + [ + 425, + 162 + ], + [ + 443, + 183 + ], + [ + 393, + 177 + ], + [ + 436, + 199 + ], + [ + 388, + 193 + ] + ], + [ + [ + 409, + 173 + ], + [ + 417, + 147 + ], + [ + 432, + 180 + ], + [ + 385, + 166 + ], + [ + 424, + 205 + ], + [ + 377, + 192 + ] + ], + [ + [ + 424, + 193 + ], + [ + 442, + 204 + ], + [ + 418, + 215 + ], + [ + 429, + 171 + ], + [ + 399, + 204 + ], + [ + 411, + 159 + ] + ] + ] + }, + { + "task_id": "0001665", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_038522063145#265", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001665/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 352, + 141 + ], + [ + 362, + 124 + ], + [ + 338, + 148 + ], + [ + 367, + 135 + ], + [ + 328, + 165 + ], + [ + 356, + 153 + ] + ], + [ + [ + 369, + 160 + ], + [ + 381, + 144 + ], + [ + 380, + 153 + ], + [ + 359, + 166 + ], + [ + 367, + 170 + ], + [ + 347, + 182 + ] + ], + [ + [ + 371, + 124 + ], + [ + 387, + 106 + ], + [ + 387, + 140 + ], + [ + 355, + 107 + ], + [ + 371, + 157 + ], + [ + 340, + 124 + ] + ], + [ + [ + 361, + 141 + ], + [ + 366, + 125 + ], + [ + 381, + 153 + ], + [ + 342, + 129 + ], + [ + 376, + 167 + ], + [ + 338, + 144 + ] + ] + ] + }, + { + "task_id": "0001666", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_038522063145#433", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001666/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 355, + 67 + ], + [ + 337, + 64 + ], + [ + 363, + 70 + ], + [ + 347, + 63 + ], + [ + 381, + 72 + ], + [ + 367, + 65 + ] + ], + [ + [ + 348, + 83 + ], + [ + 351, + 65 + ], + [ + 365, + 92 + ], + [ + 330, + 74 + ], + [ + 361, + 108 + ], + [ + 327, + 90 + ] + ], + [ + [ + 356, + 92 + ], + [ + 361, + 91 + ], + [ + 358, + 113 + ], + [ + 355, + 71 + ], + [ + 353, + 115 + ], + [ + 349, + 72 + ] + ], + [ + [ + 356, + 88 + ], + [ + 375, + 94 + ], + [ + 360, + 96 + ], + [ + 351, + 80 + ], + [ + 340, + 90 + ], + [ + 332, + 74 + ] + ] + ] + }, + { + "task_id": "0001670", + "id": "RH20T_cfg2_task_0016_user_0003_scene_0009_cfg_0002_105422061350#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001670/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 327, + 130 + ], + [ + 305, + 135 + ], + [ + 317, + 107 + ], + [ + 337, + 153 + ], + [ + 337, + 102 + ], + [ + 357, + 147 + ] + ], + [ + [ + 312, + 155 + ], + [ + 317, + 141 + ], + [ + 335, + 167 + ], + [ + 289, + 144 + ], + [ + 329, + 180 + ], + [ + 286, + 157 + ] + ], + [ + [ + 302, + 168 + ], + [ + 313, + 144 + ], + [ + 290, + 166 + ], + [ + 313, + 170 + ], + [ + 279, + 191 + ], + [ + 303, + 192 + ] + ], + [ + [ + 286, + 158 + ], + [ + 301, + 174 + ], + [ + 269, + 177 + ], + [ + 303, + 139 + ], + [ + 252, + 160 + ], + [ + 288, + 120 + ] + ] + ] + }, + { + "task_id": "0001671", + "id": "RH20T_cfg1_task_0038_user_0014_scene_0004_cfg_0001_038522062547#69", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001671/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 358, + 67 + ], + [ + 381, + 56 + ], + [ + 350, + 42 + ], + [ + 365, + 90 + ], + [ + 326, + 53 + ], + [ + 342, + 100 + ] + ], + [ + [ + 348, + 59 + ], + [ + 359, + 63 + ], + [ + 324, + 67 + ], + [ + 372, + 50 + ], + [ + 315, + 62 + ], + [ + 362, + 46 + ] + ], + [ + [ + 337, + 58 + ], + [ + 348, + 68 + ], + [ + 361, + 56 + ], + [ + 316, + 60 + ], + [ + 350, + 43 + ], + [ + 303, + 48 + ] + ], + [ + [ + 366, + 50 + ], + [ + 370, + 33 + ], + [ + 378, + 71 + ], + [ + 354, + 28 + ], + [ + 373, + 85 + ], + [ + 350, + 44 + ] + ] + ] + }, + { + "task_id": "0001672", + "id": "RH20T_cfg4_task_0071_user_0015_scene_0010_cfg_0004_104122063550#52", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001672/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 228, + 309 + ], + [ + 224, + 307 + ], + [ + 249, + 332 + ], + [ + 208, + 286 + ], + [ + 251, + 332 + ], + [ + 212, + 289 + ] + ], + [ + [ + 263, + 314 + ], + [ + 261, + 296 + ], + [ + 233, + 315 + ], + [ + 293, + 312 + ], + [ + 236, + 330 + ], + [ + 294, + 328 + ] + ], + [ + [ + 221, + 295 + ], + [ + 195, + 312 + ], + [ + 212, + 272 + ], + [ + 231, + 320 + ], + [ + 237, + 257 + ], + [ + 257, + 303 + ] + ], + [ + [ + 223, + 333 + ], + [ + 193, + 337 + ], + [ + 227, + 310 + ], + [ + 219, + 355 + ], + [ + 256, + 307 + ], + [ + 247, + 351 + ] + ] + ] + }, + { + "task_id": "0001673", + "id": "RH20T_cfg4_task_0047_user_0014_scene_0005_cfg_0004_038522062288#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001673/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 195, + 265 + ], + [ + 205, + 302 + ], + [ + 214, + 271 + ], + [ + 173, + 258 + ], + [ + 205, + 238 + ], + [ + 164, + 220 + ] + ], + [ + [ + 163, + 224 + ], + [ + 152, + 204 + ], + [ + 126, + 221 + ], + [ + 199, + 227 + ], + [ + 138, + 239 + ], + [ + 206, + 245 + ] + ], + [ + [ + 166, + 256 + ], + [ + 137, + 242 + ], + [ + 151, + 290 + ], + [ + 182, + 223 + ], + [ + 181, + 306 + ], + [ + 212, + 237 + ] + ], + [ + [ + 189, + 245 + ], + [ + 190, + 229 + ], + [ + 159, + 269 + ], + [ + 215, + 224 + ], + [ + 159, + 283 + ], + [ + 212, + 239 + ] + ] + ] + }, + { + "task_id": "0001674", + "id": "RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_104122061602#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001674/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 491, + 162 + ], + [ + 483, + 141 + ], + [ + 472, + 174 + ], + [ + 511, + 149 + ], + [ + 479, + 196 + ], + [ + 521, + 172 + ] + ], + [ + [ + 505, + 169 + ], + [ + 514, + 149 + ], + [ + 526, + 185 + ], + [ + 485, + 154 + ], + [ + 517, + 204 + ], + [ + 478, + 172 + ] + ], + [ + [ + 483, + 160 + ], + [ + 490, + 182 + ], + [ + 476, + 148 + ], + [ + 488, + 171 + ], + [ + 469, + 127 + ], + [ + 481, + 152 + ] + ], + [ + [ + 513, + 157 + ], + [ + 539, + 155 + ], + [ + 518, + 184 + ], + [ + 509, + 131 + ], + [ + 492, + 187 + ], + [ + 484, + 133 + ] + ] + ] + }, + { + "task_id": "0001676", + "id": "RH20T_cfg2_task_0201_user_0007_scene_0008_cfg_0002_105422061350#341", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Insert the tip of a medium pipette into the holder for medium pipette tips'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001676/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 253, + 226 + ], + [ + 264, + 202 + ], + [ + 258, + 223 + ], + [ + 247, + 229 + ], + [ + 248, + 246 + ], + [ + 235, + 255 + ] + ], + [ + [ + 243, + 219 + ], + [ + 238, + 219 + ], + [ + 268, + 230 + ], + [ + 218, + 208 + ], + [ + 270, + 228 + ], + [ + 223, + 208 + ] + ], + [ + [ + 260, + 241 + ], + [ + 237, + 250 + ], + [ + 266, + 266 + ], + [ + 254, + 215 + ], + [ + 287, + 257 + ], + [ + 275, + 207 + ] + ], + [ + [ + 270, + 208 + ], + [ + 260, + 231 + ], + [ + 246, + 198 + ], + [ + 295, + 218 + ], + [ + 256, + 177 + ], + [ + 304, + 196 + ] + ] + ] + }, + { + "task_id": "0001677", + "id": "RH20T_cfg1_task_0053_user_0014_scene_0003_cfg_0001_039422060546#163", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a clamp'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001677/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 335, + 123 + ], + [ + 339, + 113 + ], + [ + 325, + 99 + ], + [ + 344, + 146 + ], + [ + 320, + 110 + ], + [ + 341, + 158 + ] + ], + [ + [ + 316, + 92 + ], + [ + 318, + 66 + ], + [ + 341, + 97 + ], + [ + 292, + 87 + ], + [ + 339, + 122 + ], + [ + 291, + 112 + ] + ], + [ + [ + 310, + 110 + ], + [ + 322, + 87 + ], + [ + 328, + 124 + ], + [ + 290, + 95 + ], + [ + 316, + 147 + ], + [ + 277, + 119 + ] + ], + [ + [ + 313, + 70 + ], + [ + 321, + 63 + ], + [ + 313, + 43 + ], + [ + 313, + 95 + ], + [ + 303, + 50 + ], + [ + 303, + 104 + ] + ] + ] + }, + { + "task_id": "0001679", + "id": "RH20T_cfg5_task_0001_user_0010_scene_0001_cfg_0005_105422061350#167", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001679/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 270, + 114 + ], + [ + 288, + 96 + ], + [ + 251, + 102 + ], + [ + 289, + 126 + ], + [ + 232, + 120 + ], + [ + 270, + 145 + ] + ], + [ + [ + 282, + 89 + ], + [ + 283, + 68 + ], + [ + 295, + 71 + ], + [ + 269, + 109 + ], + [ + 293, + 90 + ], + [ + 269, + 128 + ] + ], + [ + [ + 308, + 110 + ], + [ + 285, + 126 + ], + [ + 311, + 113 + ], + [ + 305, + 107 + ], + [ + 335, + 96 + ], + [ + 326, + 92 + ] + ], + [ + [ + 264, + 87 + ], + [ + 268, + 75 + ], + [ + 248, + 65 + ], + [ + 279, + 108 + ], + [ + 243, + 77 + ], + [ + 276, + 122 + ] + ] + ] + }, + { + "task_id": "0001680", + "id": "RH20T_cfg1_task_0053_user_0014_scene_0003_cfg_0001_819112070044#155", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a clamp'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001680/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 293, + 274 + ], + [ + 289, + 275 + ], + [ + 307, + 256 + ], + [ + 279, + 292 + ], + [ + 312, + 254 + ], + [ + 283, + 292 + ] + ], + [ + [ + 290, + 270 + ], + [ + 296, + 248 + ], + [ + 268, + 263 + ], + [ + 312, + 278 + ], + [ + 263, + 285 + ], + [ + 306, + 300 + ] + ], + [ + [ + 300, + 261 + ], + [ + 292, + 251 + ], + [ + 290, + 244 + ], + [ + 310, + 279 + ], + [ + 298, + 253 + ], + [ + 317, + 288 + ] + ], + [ + [ + 311, + 254 + ], + [ + 295, + 261 + ], + [ + 297, + 257 + ], + [ + 327, + 250 + ], + [ + 312, + 250 + ], + [ + 342, + 243 + ] + ] + ] + }, + { + "task_id": "0001683", + "id": "RH20T_cfg5_task_0001_user_0007_scene_0008_cfg_0005_036422060215#159", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001683/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 123 + ], + [ + 319, + 124 + ], + [ + 308, + 143 + ], + [ + 321, + 102 + ], + [ + 303, + 143 + ], + [ + 317, + 100 + ] + ], + [ + [ + 288, + 145 + ], + [ + 271, + 136 + ], + [ + 289, + 160 + ], + [ + 287, + 129 + ], + [ + 305, + 169 + ], + [ + 304, + 138 + ] + ], + [ + [ + 301, + 133 + ], + [ + 304, + 115 + ], + [ + 320, + 129 + ], + [ + 282, + 138 + ], + [ + 317, + 146 + ], + [ + 280, + 155 + ] + ], + [ + [ + 289, + 130 + ], + [ + 296, + 136 + ], + [ + 307, + 120 + ], + [ + 270, + 139 + ], + [ + 300, + 114 + ], + [ + 262, + 134 + ] + ] + ] + }, + { + "task_id": "0001685", + "id": "RH20T_cfg1_task_0066_user_0015_scene_0006_cfg_0001_035622060973#165", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001685/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 248, + 279 + ], + [ + 269, + 268 + ], + [ + 246, + 278 + ], + [ + 250, + 281 + ], + [ + 224, + 290 + ], + [ + 230, + 292 + ] + ], + [ + [ + 250, + 304 + ], + [ + 250, + 316 + ], + [ + 224, + 309 + ], + [ + 275, + 300 + ], + [ + 223, + 296 + ], + [ + 276, + 287 + ] + ], + [ + [ + 252, + 296 + ], + [ + 249, + 287 + ], + [ + 241, + 276 + ], + [ + 264, + 316 + ], + [ + 245, + 285 + ], + [ + 267, + 323 + ] + ], + [ + [ + 263, + 307 + ], + [ + 241, + 298 + ], + [ + 256, + 327 + ], + [ + 270, + 287 + ], + [ + 278, + 336 + ], + [ + 293, + 296 + ] + ] + ] + }, + { + "task_id": "0001687", + "id": "RH20T_cfg1_task_0066_user_0015_scene_0006_cfg_0001_035622060973#698", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001687/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 304, + 254 + ], + [ + 305, + 252 + ], + [ + 327, + 256 + ], + [ + 281, + 251 + ], + [ + 325, + 258 + ], + [ + 281, + 254 + ] + ], + [ + [ + 310, + 264 + ], + [ + 310, + 255 + ], + [ + 298, + 247 + ], + [ + 324, + 282 + ], + [ + 299, + 256 + ], + [ + 324, + 290 + ] + ], + [ + [ + 321, + 255 + ], + [ + 333, + 246 + ], + [ + 338, + 268 + ], + [ + 303, + 242 + ], + [ + 326, + 276 + ], + [ + 292, + 251 + ] + ], + [ + [ + 315, + 243 + ], + [ + 336, + 234 + ], + [ + 308, + 224 + ], + [ + 322, + 264 + ], + [ + 288, + 233 + ], + [ + 301, + 274 + ] + ] + ] + }, + { + "task_id": "0001688", + "id": "RH20T_cfg1_task_0066_user_0015_scene_0006_cfg_0001_035622060973#942", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001688/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 325, + 308 + ], + [ + 343, + 323 + ], + [ + 332, + 313 + ], + [ + 318, + 303 + ], + [ + 317, + 300 + ], + [ + 301, + 288 + ] + ], + [ + [ + 338, + 290 + ], + [ + 339, + 282 + ], + [ + 325, + 271 + ], + [ + 351, + 310 + ], + [ + 324, + 279 + ], + [ + 349, + 317 + ] + ], + [ + [ + 330, + 304 + ], + [ + 306, + 303 + ], + [ + 333, + 281 + ], + [ + 326, + 329 + ], + [ + 357, + 282 + ], + [ + 351, + 331 + ] + ], + [ + [ + 353, + 282 + ], + [ + 344, + 265 + ], + [ + 346, + 270 + ], + [ + 361, + 294 + ], + [ + 354, + 286 + ], + [ + 370, + 310 + ] + ] + ] + }, + { + "task_id": "0001690", + "id": "RH20T_cfg1_task_0066_user_0015_scene_0006_cfg_0001_035622060973#1511", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001690/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 390, + 266 + ], + [ + 392, + 257 + ], + [ + 377, + 248 + ], + [ + 405, + 285 + ], + [ + 375, + 257 + ], + [ + 402, + 293 + ] + ], + [ + [ + 372, + 269 + ], + [ + 389, + 252 + ], + [ + 376, + 267 + ], + [ + 369, + 270 + ], + [ + 358, + 285 + ], + [ + 352, + 287 + ] + ], + [ + [ + 392, + 258 + ], + [ + 404, + 245 + ], + [ + 394, + 274 + ], + [ + 390, + 241 + ], + [ + 382, + 288 + ], + [ + 376, + 255 + ] + ], + [ + [ + 375, + 259 + ], + [ + 360, + 272 + ], + [ + 391, + 276 + ], + [ + 359, + 243 + ], + [ + 406, + 263 + ], + [ + 374, + 231 + ] + ] + ] + }, + { + "task_id": "0001691", + "id": "RH20T_cfg5_task_0025_user_0010_scene_0006_cfg_0005_104422070044#97", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold a piece of paper'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001691/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 419, + 202 + ], + [ + 430, + 192 + ], + [ + 408, + 211 + ], + [ + 430, + 191 + ], + [ + 396, + 221 + ], + [ + 418, + 201 + ] + ], + [ + [ + 409, + 179 + ], + [ + 408, + 164 + ], + [ + 401, + 166 + ], + [ + 417, + 193 + ], + [ + 403, + 179 + ], + [ + 418, + 207 + ] + ], + [ + [ + 410, + 204 + ], + [ + 421, + 189 + ], + [ + 393, + 197 + ], + [ + 428, + 212 + ], + [ + 382, + 211 + ], + [ + 417, + 227 + ] + ], + [ + [ + 408, + 184 + ], + [ + 403, + 166 + ], + [ + 425, + 186 + ], + [ + 392, + 183 + ], + [ + 430, + 204 + ], + [ + 397, + 200 + ] + ] + ] + }, + { + "task_id": "0001692", + "id": "RH20T_cfg1_task_0110_user_0015_scene_0003_cfg_0001_038522062547#99", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the pencil out from the pencil sharpener'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001692/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 402, + 112 + ], + [ + 406, + 98 + ], + [ + 403, + 135 + ], + [ + 401, + 89 + ], + [ + 399, + 146 + ], + [ + 397, + 102 + ] + ], + [ + [ + 421, + 107 + ], + [ + 422, + 96 + ], + [ + 434, + 85 + ], + [ + 410, + 127 + ], + [ + 434, + 97 + ], + [ + 409, + 140 + ] + ], + [ + [ + 431, + 97 + ], + [ + 405, + 92 + ], + [ + 422, + 95 + ], + [ + 440, + 99 + ], + [ + 446, + 100 + ], + [ + 466, + 105 + ] + ], + [ + [ + 399, + 129 + ], + [ + 424, + 130 + ], + [ + 399, + 155 + ], + [ + 399, + 103 + ], + [ + 374, + 154 + ], + [ + 374, + 102 + ] + ] + ] + }, + { + "task_id": "0001695", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0002_cfg_0006_104122061018#413", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001695/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 252, + 100 + ], + [ + 238, + 74 + ], + [ + 277, + 88 + ], + [ + 226, + 112 + ], + [ + 290, + 113 + ], + [ + 240, + 137 + ] + ], + [ + [ + 247, + 88 + ], + [ + 239, + 60 + ], + [ + 241, + 95 + ], + [ + 255, + 79 + ], + [ + 249, + 121 + ], + [ + 263, + 108 + ] + ], + [ + [ + 261, + 68 + ], + [ + 247, + 43 + ], + [ + 244, + 79 + ], + [ + 276, + 58 + ], + [ + 258, + 104 + ], + [ + 288, + 82 + ] + ], + [ + [ + 270, + 103 + ], + [ + 272, + 118 + ], + [ + 298, + 108 + ], + [ + 242, + 97 + ], + [ + 298, + 91 + ], + [ + 238, + 80 + ] + ] + ] + }, + { + "task_id": "0001696", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0002_cfg_0006_104122061018#723", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001696/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 228, + 87 + ], + [ + 223, + 113 + ], + [ + 225, + 81 + ], + [ + 231, + 92 + ], + [ + 231, + 53 + ], + [ + 236, + 67 + ] + ], + [ + [ + 240, + 35 + ], + [ + 239, + 47 + ], + [ + 243, + 61 + ], + [ + 236, + 8 + ], + [ + 243, + 49 + ], + [ + 237, + 0 + ] + ], + [ + [ + 223, + 69 + ], + [ + 215, + 54 + ], + [ + 226, + 90 + ], + [ + 219, + 44 + ], + [ + 235, + 106 + ], + [ + 228, + 60 + ] + ], + [ + [ + 212, + 61 + ], + [ + 184, + 67 + ], + [ + 207, + 68 + ], + [ + 218, + 53 + ], + [ + 233, + 62 + ], + [ + 247, + 47 + ] + ] + ] + }, + { + "task_id": "0001703", + "id": "RH20T_cfg4_task_0089_user_0014_scene_0001_cfg_0004_104422070011#320", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push an object with obstacles'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001703/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 387, + 199 + ], + [ + 402, + 177 + ], + [ + 383, + 209 + ], + [ + 392, + 189 + ], + [ + 368, + 231 + ], + [ + 376, + 213 + ] + ], + [ + [ + 392, + 226 + ], + [ + 402, + 216 + ], + [ + 376, + 246 + ], + [ + 407, + 208 + ], + [ + 368, + 254 + ], + [ + 397, + 218 + ] + ], + [ + [ + 380, + 201 + ], + [ + 398, + 222 + ], + [ + 365, + 213 + ], + [ + 395, + 190 + ], + [ + 347, + 192 + ], + [ + 378, + 170 + ] + ], + [ + [ + 419, + 228 + ], + [ + 436, + 247 + ], + [ + 419, + 211 + ], + [ + 420, + 248 + ], + [ + 402, + 192 + ], + [ + 401, + 227 + ] + ] + ] + }, + { + "task_id": "0001704", + "id": "RH20T_cfg4_task_0089_user_0014_scene_0001_cfg_0004_104422070011#377", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push an object with obstacles'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001704/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 396, + 181 + ], + [ + 399, + 204 + ], + [ + 379, + 190 + ], + [ + 414, + 171 + ], + [ + 376, + 169 + ], + [ + 410, + 149 + ] + ], + [ + [ + 385, + 189 + ], + [ + 398, + 203 + ], + [ + 364, + 203 + ], + [ + 405, + 175 + ], + [ + 350, + 189 + ], + [ + 391, + 160 + ] + ], + [ + [ + 411, + 193 + ], + [ + 419, + 183 + ], + [ + 387, + 198 + ], + [ + 433, + 189 + ], + [ + 380, + 208 + ], + [ + 425, + 199 + ] + ], + [ + [ + 406, + 222 + ], + [ + 407, + 197 + ], + [ + 417, + 226 + ], + [ + 396, + 219 + ], + [ + 416, + 253 + ], + [ + 394, + 243 + ] + ] + ] + }, + { + "task_id": "0001705", + "id": "RH20T_cfg4_task_0089_user_0014_scene_0001_cfg_0004_104422070011#458", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push an object with obstacles'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001705/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 377, + 146 + ], + [ + 393, + 152 + ], + [ + 372, + 125 + ], + [ + 381, + 166 + ], + [ + 357, + 120 + ], + [ + 367, + 160 + ] + ], + [ + [ + 405, + 153 + ], + [ + 421, + 171 + ], + [ + 419, + 137 + ], + [ + 393, + 169 + ], + [ + 403, + 119 + ], + [ + 378, + 152 + ] + ], + [ + [ + 407, + 154 + ], + [ + 414, + 176 + ], + [ + 420, + 144 + ], + [ + 396, + 163 + ], + [ + 412, + 122 + ], + [ + 390, + 142 + ] + ], + [ + [ + 401, + 140 + ], + [ + 411, + 130 + ], + [ + 379, + 143 + ], + [ + 422, + 138 + ], + [ + 370, + 153 + ], + [ + 411, + 148 + ] + ] + ] + }, + { + "task_id": "0001709", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061018#563", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001709/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 250, + 122 + ], + [ + 279, + 126 + ], + [ + 243, + 153 + ], + [ + 258, + 89 + ], + [ + 212, + 150 + ], + [ + 227, + 82 + ] + ], + [ + [ + 229, + 129 + ], + [ + 227, + 104 + ], + [ + 218, + 108 + ], + [ + 241, + 151 + ], + [ + 221, + 131 + ], + [ + 242, + 173 + ] + ], + [ + [ + 267, + 123 + ], + [ + 278, + 96 + ], + [ + 281, + 147 + ], + [ + 254, + 100 + ], + [ + 270, + 170 + ], + [ + 245, + 124 + ] + ], + [ + [ + 212, + 172 + ], + [ + 182, + 155 + ], + [ + 226, + 146 + ], + [ + 197, + 200 + ], + [ + 254, + 163 + ], + [ + 227, + 217 + ] + ] + ] + }, + { + "task_id": "0001710", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061018#717", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001710/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 255, + 174 + ], + [ + 279, + 181 + ], + [ + 261, + 207 + ], + [ + 250, + 144 + ], + [ + 234, + 202 + ], + [ + 223, + 135 + ] + ], + [ + [ + 243, + 173 + ], + [ + 249, + 164 + ], + [ + 249, + 142 + ], + [ + 237, + 205 + ], + [ + 243, + 152 + ], + [ + 232, + 212 + ] + ], + [ + [ + 227, + 174 + ], + [ + 225, + 154 + ], + [ + 216, + 150 + ], + [ + 238, + 200 + ], + [ + 218, + 169 + ], + [ + 239, + 218 + ] + ], + [ + [ + 228, + 151 + ], + [ + 258, + 160 + ], + [ + 241, + 121 + ], + [ + 216, + 179 + ], + [ + 208, + 109 + ], + [ + 184, + 170 + ] + ] + ] + }, + { + "task_id": "0001711", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061018#860", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001711/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 256, + 159 + ], + [ + 256, + 132 + ], + [ + 245, + 135 + ], + [ + 269, + 185 + ], + [ + 246, + 159 + ], + [ + 269, + 208 + ] + ], + [ + [ + 272, + 176 + ], + [ + 293, + 169 + ], + [ + 264, + 208 + ], + [ + 279, + 147 + ], + [ + 244, + 214 + ], + [ + 260, + 155 + ] + ], + [ + [ + 252, + 122 + ], + [ + 289, + 129 + ], + [ + 262, + 96 + ], + [ + 241, + 150 + ], + [ + 227, + 89 + ], + [ + 202, + 143 + ] + ], + [ + [ + 247, + 151 + ], + [ + 251, + 114 + ], + [ + 282, + 151 + ], + [ + 208, + 150 + ], + [ + 277, + 184 + ], + [ + 205, + 185 + ] + ] + ] + }, + { + "task_id": "0001712", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122061018#1004", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001712/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 216, + 203 + ], + [ + 197, + 238 + ], + [ + 186, + 185 + ], + [ + 243, + 220 + ], + [ + 205, + 150 + ], + [ + 259, + 188 + ] + ], + [ + [ + 252, + 226 + ], + [ + 253, + 199 + ], + [ + 240, + 202 + ], + [ + 265, + 253 + ], + [ + 240, + 225 + ], + [ + 263, + 277 + ] + ], + [ + [ + 280, + 235 + ], + [ + 254, + 264 + ], + [ + 305, + 248 + ], + [ + 259, + 224 + ], + [ + 331, + 219 + ], + [ + 283, + 199 + ] + ], + [ + [ + 264, + 226 + ], + [ + 296, + 219 + ], + [ + 256, + 192 + ], + [ + 272, + 261 + ], + [ + 222, + 199 + ], + [ + 238, + 270 + ] + ] + ] + }, + { + "task_id": "0001713", + "id": "RH20T_cfg5_task_0010_user_0010_scene_0002_cfg_0005_037522062165#293", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001713/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 261, + 158 + ], + [ + 239, + 145 + ], + [ + 263, + 154 + ], + [ + 260, + 162 + ], + [ + 284, + 166 + ], + [ + 284, + 175 + ] + ], + [ + [ + 234, + 195 + ], + [ + 226, + 214 + ], + [ + 257, + 193 + ], + [ + 212, + 198 + ], + [ + 264, + 176 + ], + [ + 220, + 181 + ] + ], + [ + [ + 234, + 173 + ], + [ + 234, + 155 + ], + [ + 211, + 165 + ], + [ + 257, + 181 + ], + [ + 212, + 182 + ], + [ + 256, + 198 + ] + ], + [ + [ + 232, + 199 + ], + [ + 235, + 174 + ], + [ + 250, + 202 + ], + [ + 212, + 197 + ], + [ + 247, + 227 + ], + [ + 209, + 224 + ] + ] + ] + }, + { + "task_id": "0001714", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0003_cfg_0001_038522063145#152", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001714/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 432, + 164 + ], + [ + 454, + 173 + ], + [ + 420, + 170 + ], + [ + 444, + 158 + ], + [ + 398, + 161 + ], + [ + 423, + 150 + ] + ], + [ + [ + 419, + 172 + ], + [ + 438, + 183 + ], + [ + 410, + 192 + ], + [ + 429, + 151 + ], + [ + 393, + 180 + ], + [ + 411, + 140 + ] + ], + [ + [ + 434, + 120 + ], + [ + 418, + 130 + ], + [ + 416, + 114 + ], + [ + 451, + 126 + ], + [ + 433, + 103 + ], + [ + 469, + 115 + ] + ], + [ + [ + 436, + 152 + ], + [ + 443, + 134 + ], + [ + 452, + 143 + ], + [ + 419, + 160 + ], + [ + 445, + 160 + ], + [ + 413, + 177 + ] + ] + ] + }, + { + "task_id": "0001716", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0003_cfg_0001_038522063145#495", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001716/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 421, + 131 + ], + [ + 407, + 116 + ], + [ + 417, + 141 + ], + [ + 424, + 122 + ], + [ + 431, + 156 + ], + [ + 438, + 137 + ] + ], + [ + [ + 402, + 106 + ], + [ + 388, + 117 + ], + [ + 414, + 123 + ], + [ + 390, + 89 + ], + [ + 429, + 112 + ], + [ + 405, + 77 + ] + ], + [ + [ + 407, + 105 + ], + [ + 411, + 87 + ], + [ + 427, + 103 + ], + [ + 386, + 107 + ], + [ + 423, + 119 + ], + [ + 383, + 123 + ] + ], + [ + [ + 403, + 100 + ], + [ + 403, + 87 + ], + [ + 415, + 113 + ], + [ + 391, + 85 + ], + [ + 415, + 127 + ], + [ + 390, + 98 + ] + ] + ] + }, + { + "task_id": "0001717", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0003_cfg_0001_038522063145#755", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001717/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 368, + 76 + ], + [ + 360, + 78 + ], + [ + 376, + 96 + ], + [ + 361, + 56 + ], + [ + 384, + 95 + ], + [ + 370, + 52 + ] + ], + [ + [ + 364, + 90 + ], + [ + 364, + 74 + ], + [ + 383, + 84 + ], + [ + 344, + 95 + ], + [ + 382, + 99 + ], + [ + 345, + 109 + ] + ], + [ + [ + 348, + 93 + ], + [ + 356, + 109 + ], + [ + 331, + 91 + ], + [ + 365, + 96 + ], + [ + 323, + 76 + ], + [ + 357, + 81 + ] + ], + [ + [ + 358, + 98 + ], + [ + 372, + 97 + ], + [ + 344, + 98 + ], + [ + 372, + 98 + ], + [ + 329, + 100 + ], + [ + 357, + 100 + ] + ] + ] + }, + { + "task_id": "0001719", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0003_cfg_0001_038522063145#1599", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001719/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 471, + 127 + ], + [ + 476, + 110 + ], + [ + 488, + 120 + ], + [ + 453, + 134 + ], + [ + 482, + 136 + ], + [ + 448, + 150 + ] + ], + [ + [ + 472, + 116 + ], + [ + 481, + 109 + ], + [ + 451, + 113 + ], + [ + 493, + 119 + ], + [ + 443, + 119 + ], + [ + 484, + 125 + ] + ], + [ + [ + 470, + 129 + ], + [ + 484, + 141 + ], + [ + 474, + 110 + ], + [ + 466, + 149 + ], + [ + 460, + 98 + ], + [ + 452, + 138 + ] + ], + [ + [ + 470, + 108 + ], + [ + 454, + 101 + ], + [ + 482, + 117 + ], + [ + 459, + 100 + ], + [ + 498, + 123 + ], + [ + 474, + 106 + ] + ] + ] + }, + { + "task_id": "0001720", + "id": "RH20T_cfg5_task_0010_user_0007_scene_0003_cfg_0005_104422070044#109", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001720/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 344, + 158 + ], + [ + 342, + 146 + ], + [ + 350, + 170 + ], + [ + 339, + 146 + ], + [ + 352, + 180 + ], + [ + 342, + 157 + ] + ], + [ + [ + 351, + 152 + ], + [ + 357, + 167 + ], + [ + 342, + 149 + ], + [ + 361, + 156 + ], + [ + 336, + 134 + ], + [ + 355, + 141 + ] + ], + [ + [ + 357, + 159 + ], + [ + 347, + 155 + ], + [ + 354, + 175 + ], + [ + 361, + 142 + ], + [ + 364, + 179 + ], + [ + 371, + 146 + ] + ], + [ + [ + 339, + 162 + ], + [ + 337, + 149 + ], + [ + 326, + 158 + ], + [ + 351, + 166 + ], + [ + 328, + 171 + ], + [ + 354, + 180 + ] + ] + ] + }, + { + "task_id": "0001721", + "id": "RH20T_cfg6_task_0062_user_0014_scene_0003_cfg_0006_104122063633#108", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the toilet paper on its holder'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001721/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 186, + 107 + ], + [ + 204, + 102 + ], + [ + 182, + 120 + ], + [ + 189, + 93 + ], + [ + 163, + 126 + ], + [ + 169, + 97 + ] + ], + [ + [ + 194, + 131 + ], + [ + 171, + 129 + ], + [ + 196, + 110 + ], + [ + 192, + 152 + ], + [ + 218, + 113 + ], + [ + 213, + 153 + ] + ], + [ + [ + 178, + 114 + ], + [ + 160, + 101 + ], + [ + 184, + 117 + ], + [ + 172, + 111 + ], + [ + 201, + 130 + ], + [ + 191, + 124 + ] + ], + [ + [ + 194, + 135 + ], + [ + 196, + 115 + ], + [ + 196, + 127 + ], + [ + 191, + 144 + ], + [ + 194, + 145 + ], + [ + 188, + 163 + ] + ] + ] + }, + { + "task_id": "0001724", + "id": "RH20T_cfg3_task_0062_user_0010_scene_0008_cfg_0003_104122063550#99", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the toilet paper on its holder'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001724/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 293, + 253 + ], + [ + 298, + 242 + ], + [ + 305, + 280 + ], + [ + 281, + 226 + ], + [ + 300, + 288 + ], + [ + 277, + 237 + ] + ], + [ + [ + 292, + 273 + ], + [ + 299, + 271 + ], + [ + 311, + 250 + ], + [ + 272, + 295 + ], + [ + 304, + 250 + ], + [ + 263, + 298 + ] + ], + [ + [ + 305, + 251 + ], + [ + 334, + 257 + ], + [ + 309, + 235 + ], + [ + 299, + 270 + ], + [ + 282, + 229 + ], + [ + 268, + 264 + ] + ], + [ + [ + 285, + 289 + ], + [ + 262, + 310 + ], + [ + 265, + 268 + ], + [ + 306, + 312 + ], + [ + 288, + 250 + ], + [ + 328, + 292 + ] + ] + ] + }, + { + "task_id": "0001725", + "id": "RH20T_cfg7_task_0063_user_0020_scene_0008_cfg_0007_037522061512#142", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a shovel to scoop up an object'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001725/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 216, + 239 + ], + [ + 209, + 226 + ], + [ + 201, + 226 + ], + [ + 230, + 251 + ], + [ + 208, + 240 + ], + [ + 238, + 265 + ] + ], + [ + [ + 195, + 195 + ], + [ + 170, + 201 + ], + [ + 199, + 217 + ], + [ + 191, + 173 + ], + [ + 224, + 211 + ], + [ + 215, + 167 + ] + ], + [ + [ + 191, + 212 + ], + [ + 179, + 205 + ], + [ + 172, + 227 + ], + [ + 209, + 196 + ], + [ + 184, + 232 + ], + [ + 220, + 203 + ] + ], + [ + [ + 208, + 213 + ], + [ + 222, + 192 + ], + [ + 215, + 217 + ], + [ + 201, + 208 + ], + [ + 201, + 237 + ], + [ + 186, + 231 + ] + ] + ] + }, + { + "task_id": "0001727", + "id": "RH20T_cfg5_task_0045_user_0010_scene_0004_cfg_0005_104122063678#388", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the brush on the pen rack'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001727/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 237, + 243 + ], + [ + 248, + 221 + ], + [ + 247, + 255 + ], + [ + 226, + 230 + ], + [ + 236, + 278 + ], + [ + 214, + 253 + ] + ], + [ + [ + 247, + 218 + ], + [ + 255, + 203 + ], + [ + 253, + 240 + ], + [ + 241, + 198 + ], + [ + 245, + 254 + ], + [ + 234, + 213 + ] + ], + [ + [ + 248, + 218 + ], + [ + 223, + 209 + ], + [ + 255, + 196 + ], + [ + 241, + 242 + ], + [ + 279, + 205 + ], + [ + 267, + 252 + ] + ], + [ + [ + 229, + 219 + ], + [ + 207, + 233 + ], + [ + 233, + 208 + ], + [ + 226, + 229 + ], + [ + 255, + 195 + ], + [ + 246, + 216 + ] + ] + ] + }, + { + "task_id": "0001731", + "id": "RH20T_cfg5_task_0043_user_0007_scene_0002_cfg_0005_036422060215#259", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001731/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 276, + 154 + ], + [ + 275, + 136 + ], + [ + 290, + 165 + ], + [ + 262, + 143 + ], + [ + 290, + 183 + ], + [ + 263, + 161 + ] + ], + [ + [ + 295, + 180 + ], + [ + 290, + 156 + ], + [ + 308, + 175 + ], + [ + 282, + 184 + ], + [ + 314, + 200 + ], + [ + 287, + 207 + ] + ], + [ + [ + 297, + 169 + ], + [ + 313, + 161 + ], + [ + 282, + 158 + ], + [ + 313, + 181 + ], + [ + 265, + 167 + ], + [ + 295, + 190 + ] + ], + [ + [ + 289, + 165 + ], + [ + 295, + 148 + ], + [ + 304, + 180 + ], + [ + 275, + 151 + ], + [ + 298, + 196 + ], + [ + 271, + 167 + ] + ] + ] + }, + { + "task_id": "0001732", + "id": "RH20T_cfg5_task_0043_user_0007_scene_0002_cfg_0005_036422060215#332", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001732/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 301, + 175 + ], + [ + 318, + 178 + ], + [ + 285, + 174 + ], + [ + 316, + 175 + ], + [ + 269, + 171 + ], + [ + 300, + 172 + ] + ], + [ + [ + 281, + 174 + ], + [ + 298, + 166 + ], + [ + 268, + 171 + ], + [ + 296, + 178 + ], + [ + 250, + 179 + ], + [ + 278, + 187 + ] + ], + [ + [ + 303, + 176 + ], + [ + 317, + 164 + ], + [ + 287, + 176 + ], + [ + 320, + 175 + ], + [ + 272, + 187 + ], + [ + 306, + 188 + ] + ], + [ + [ + 292, + 171 + ], + [ + 295, + 154 + ], + [ + 308, + 185 + ], + [ + 276, + 158 + ], + [ + 304, + 201 + ], + [ + 274, + 174 + ] + ] + ] + }, + { + "task_id": "0001733", + "id": "RH20T_cfg5_task_0043_user_0007_scene_0002_cfg_0005_036422060215#515", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001733/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 302, + 199 + ], + [ + 296, + 217 + ], + [ + 314, + 192 + ], + [ + 290, + 207 + ], + [ + 320, + 173 + ], + [ + 297, + 188 + ] + ], + [ + [ + 273, + 170 + ], + [ + 280, + 155 + ], + [ + 271, + 154 + ], + [ + 275, + 186 + ], + [ + 263, + 170 + ], + [ + 267, + 202 + ] + ], + [ + [ + 292, + 175 + ], + [ + 297, + 158 + ], + [ + 301, + 191 + ], + [ + 284, + 159 + ], + [ + 296, + 207 + ], + [ + 279, + 175 + ] + ], + [ + [ + 274, + 169 + ], + [ + 270, + 171 + ], + [ + 278, + 146 + ], + [ + 270, + 192 + ], + [ + 283, + 142 + ], + [ + 274, + 190 + ] + ] + ] + }, + { + "task_id": "0001740", + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122063633#241", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001740/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 250, + 200 + ], + [ + 250, + 177 + ], + [ + 253, + 197 + ], + [ + 248, + 203 + ], + [ + 253, + 219 + ], + [ + 248, + 228 + ] + ], + [ + [ + 268, + 208 + ], + [ + 260, + 231 + ], + [ + 269, + 215 + ], + [ + 267, + 203 + ], + [ + 278, + 191 + ], + [ + 275, + 181 + ] + ], + [ + [ + 270, + 199 + ], + [ + 257, + 211 + ], + [ + 283, + 194 + ], + [ + 255, + 205 + ], + [ + 297, + 181 + ], + [ + 270, + 192 + ] + ], + [ + [ + 241, + 181 + ], + [ + 221, + 178 + ], + [ + 230, + 175 + ], + [ + 250, + 187 + ], + [ + 252, + 178 + ], + [ + 270, + 190 + ] + ] + ] + }, + { + "task_id": "0001744", + "id": "RH20T_cfg3_task_0023_user_0010_scene_0006_cfg_0003_104422070011#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001744/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 264, + 150 + ], + [ + 269, + 154 + ], + [ + 288, + 137 + ], + [ + 240, + 163 + ], + [ + 283, + 132 + ], + [ + 232, + 159 + ] + ], + [ + [ + 239, + 155 + ], + [ + 257, + 135 + ], + [ + 244, + 156 + ], + [ + 233, + 154 + ], + [ + 225, + 178 + ], + [ + 216, + 174 + ] + ], + [ + [ + 263, + 170 + ], + [ + 271, + 162 + ], + [ + 271, + 196 + ], + [ + 255, + 145 + ], + [ + 263, + 203 + ], + [ + 247, + 154 + ] + ], + [ + [ + 262, + 171 + ], + [ + 282, + 187 + ], + [ + 274, + 153 + ], + [ + 250, + 191 + ], + [ + 254, + 136 + ], + [ + 228, + 174 + ] + ] + ] + }, + { + "task_id": "0001745", + "id": "RH20T_cfg2_task_0101_user_0007_scene_0002_cfg_0002_104122061850#183", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the table with a cloth'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001745/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 247 + ], + [ + 363, + 271 + ], + [ + 374, + 238 + ], + [ + 369, + 255 + ], + [ + 382, + 215 + ], + [ + 377, + 234 + ] + ], + [ + [ + 373, + 227 + ], + [ + 356, + 247 + ], + [ + 360, + 211 + ], + [ + 388, + 245 + ], + [ + 376, + 191 + ], + [ + 407, + 223 + ] + ], + [ + [ + 409, + 208 + ], + [ + 436, + 217 + ], + [ + 397, + 216 + ], + [ + 420, + 200 + ], + [ + 371, + 207 + ], + [ + 396, + 192 + ] + ], + [ + [ + 395, + 229 + ], + [ + 393, + 216 + ], + [ + 382, + 251 + ], + [ + 408, + 208 + ], + [ + 386, + 263 + ], + [ + 409, + 220 + ] + ] + ] + }, + { + "task_id": "0001749", + "id": "RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_750612070853#212", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001749/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 265, + 178 + ], + [ + 268, + 218 + ], + [ + 254, + 196 + ], + [ + 279, + 157 + ], + [ + 253, + 161 + ], + [ + 276, + 118 + ] + ], + [ + [ + 264, + 184 + ], + [ + 244, + 215 + ], + [ + 240, + 185 + ], + [ + 292, + 183 + ], + [ + 259, + 159 + ], + [ + 310, + 153 + ] + ], + [ + [ + 266, + 165 + ], + [ + 264, + 195 + ], + [ + 281, + 188 + ], + [ + 249, + 139 + ], + [ + 282, + 161 + ], + [ + 252, + 111 + ] + ], + [ + [ + 262, + 172 + ], + [ + 253, + 142 + ], + [ + 231, + 165 + ], + [ + 295, + 180 + ], + [ + 240, + 192 + ], + [ + 301, + 208 + ] + ] + ] + }, + { + "task_id": "0001752", + "id": "RH20T_cfg4_task_0019_user_0016_scene_0005_cfg_0004_038522062288#201", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the first move as black on the 3-4 point in the upper right corner of the Go board'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001752/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 46, + 311 + ], + [ + 17, + 344 + ], + [ + 52, + 314 + ], + [ + 38, + 306 + ], + [ + 78, + 285 + ], + [ + 68, + 273 + ] + ], + [ + [ + 102, + 307 + ], + [ + 86, + 295 + ], + [ + 73, + 295 + ], + [ + 132, + 318 + ], + [ + 88, + 306 + ], + [ + 144, + 328 + ] + ], + [ + [ + 140, + 286 + ], + [ + 143, + 251 + ], + [ + 102, + 285 + ], + [ + 176, + 287 + ], + [ + 98, + 320 + ], + [ + 173, + 321 + ] + ], + [ + [ + 132, + 279 + ], + [ + 101, + 263 + ], + [ + 127, + 278 + ], + [ + 136, + 280 + ], + [ + 160, + 295 + ], + [ + 165, + 295 + ] + ] + ] + }, + { + "task_id": "0001754", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0007_cfg_0004_104122062823#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001754/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 253, + 63 + ], + [ + 260, + 43 + ], + [ + 237, + 38 + ], + [ + 267, + 85 + ], + [ + 227, + 59 + ], + [ + 259, + 106 + ] + ], + [ + [ + 280, + 75 + ], + [ + 278, + 50 + ], + [ + 309, + 75 + ], + [ + 252, + 75 + ], + [ + 310, + 98 + ], + [ + 255, + 98 + ] + ], + [ + [ + 255, + 87 + ], + [ + 275, + 89 + ], + [ + 276, + 90 + ], + [ + 235, + 84 + ], + [ + 254, + 87 + ], + [ + 213, + 82 + ] + ], + [ + [ + 298, + 99 + ], + [ + 301, + 73 + ], + [ + 270, + 99 + ], + [ + 327, + 99 + ], + [ + 266, + 126 + ], + [ + 325, + 127 + ] + ] + ] + }, + { + "task_id": "0001755", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0007_cfg_0004_104122062823#217", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001755/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 155 + ], + [ + 270, + 129 + ], + [ + 311, + 153 + ], + [ + 236, + 157 + ], + [ + 312, + 176 + ], + [ + 242, + 179 + ] + ], + [ + [ + 258, + 166 + ], + [ + 238, + 196 + ], + [ + 276, + 177 + ], + [ + 237, + 153 + ], + [ + 294, + 148 + ], + [ + 258, + 121 + ] + ], + [ + [ + 236, + 184 + ], + [ + 227, + 147 + ], + [ + 268, + 176 + ], + [ + 207, + 192 + ], + [ + 278, + 215 + ], + [ + 216, + 229 + ] + ], + [ + [ + 252, + 160 + ], + [ + 247, + 181 + ], + [ + 249, + 126 + ], + [ + 255, + 192 + ], + [ + 254, + 109 + ], + [ + 260, + 172 + ] + ] + ] + }, + { + "task_id": "0001758", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0007_cfg_0004_104122062823#436", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001758/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 313, + 112 + ], + [ + 312, + 83 + ], + [ + 346, + 108 + ], + [ + 280, + 115 + ], + [ + 345, + 135 + ], + [ + 283, + 141 + ] + ], + [ + [ + 318, + 126 + ], + [ + 330, + 145 + ], + [ + 348, + 119 + ], + [ + 288, + 134 + ], + [ + 336, + 103 + ], + [ + 278, + 116 + ] + ], + [ + [ + 322, + 91 + ], + [ + 307, + 107 + ], + [ + 296, + 105 + ], + [ + 345, + 79 + ], + [ + 312, + 88 + ], + [ + 362, + 61 + ] + ], + [ + [ + 305, + 126 + ], + [ + 315, + 149 + ], + [ + 338, + 128 + ], + [ + 275, + 124 + ], + [ + 329, + 101 + ], + [ + 263, + 98 + ] + ] + ] + }, + { + "task_id": "0001760", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_819112071037#311", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001760/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 416, + 164 + ], + [ + 414, + 133 + ], + [ + 483, + 169 + ], + [ + 354, + 158 + ], + [ + 478, + 193 + ], + [ + 363, + 182 + ] + ], + [ + [ + 294, + 184 + ], + [ + 332, + 233 + ], + [ + 344, + 140 + ], + [ + 245, + 227 + ], + [ + 307, + 98 + ], + [ + 215, + 181 + ] + ], + [ + [ + 357, + 183 + ], + [ + 359, + 118 + ], + [ + 307, + 198 + ], + [ + 398, + 171 + ], + [ + 307, + 262 + ], + [ + 394, + 226 + ] + ], + [ + [ + 345, + 165 + ], + [ + 329, + 149 + ], + [ + 394, + 182 + ], + [ + 290, + 147 + ], + [ + 419, + 204 + ], + [ + 303, + 166 + ] + ] + ] + }, + { + "task_id": "0001761", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_819112071037#442", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001761/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 440, + 137 + ], + [ + 440, + 81 + ], + [ + 391, + 146 + ], + [ + 484, + 129 + ], + [ + 393, + 200 + ], + [ + 482, + 180 + ] + ], + [ + [ + 454, + 104 + ], + [ + 417, + 118 + ], + [ + 479, + 162 + ], + [ + 429, + 45 + ], + [ + 531, + 152 + ], + [ + 474, + 17 + ] + ], + [ + [ + 422, + 156 + ], + [ + 429, + 137 + ], + [ + 445, + 108 + ], + [ + 399, + 206 + ], + [ + 438, + 128 + ], + [ + 396, + 216 + ] + ], + [ + [ + 493, + 174 + ], + [ + 538, + 193 + ], + [ + 491, + 225 + ], + [ + 495, + 119 + ], + [ + 456, + 205 + ], + [ + 457, + 109 + ] + ] + ] + }, + { + "task_id": "0001762", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_819112071037#670", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001762/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 253, + 99 + ], + [ + 194, + 114 + ], + [ + 256, + 143 + ], + [ + 248, + 46 + ], + [ + 305, + 128 + ], + [ + 308, + 33 + ] + ], + [ + [ + 252, + 119 + ], + [ + 275, + 135 + ], + [ + 228, + 166 + ], + [ + 277, + 69 + ], + [ + 211, + 149 + ], + [ + 256, + 60 + ] + ], + [ + [ + 295, + 209 + ], + [ + 340, + 242 + ], + [ + 261, + 257 + ], + [ + 327, + 163 + ], + [ + 212, + 223 + ], + [ + 282, + 129 + ] + ], + [ + [ + 299, + 147 + ], + [ + 301, + 92 + ], + [ + 273, + 160 + ], + [ + 320, + 136 + ], + [ + 272, + 217 + ], + [ + 318, + 183 + ] + ] + ] + }, + { + "task_id": "0001763", + "id": "RH20T_cfg3_task_0070_user_0010_scene_0008_cfg_0003_036422060909#116", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001763/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 228, + 125 + ], + [ + 237, + 113 + ], + [ + 247, + 141 + ], + [ + 207, + 110 + ], + [ + 238, + 155 + ], + [ + 196, + 123 + ] + ], + [ + [ + 253, + 108 + ], + [ + 269, + 119 + ], + [ + 236, + 113 + ], + [ + 272, + 104 + ], + [ + 218, + 102 + ], + [ + 254, + 92 + ] + ], + [ + [ + 241, + 136 + ], + [ + 250, + 119 + ], + [ + 249, + 123 + ], + [ + 232, + 151 + ], + [ + 241, + 140 + ], + [ + 224, + 167 + ] + ], + [ + [ + 236, + 110 + ], + [ + 212, + 104 + ], + [ + 228, + 130 + ], + [ + 243, + 91 + ], + [ + 252, + 137 + ], + [ + 267, + 98 + ] + ] + ] + }, + { + "task_id": "0001767", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0006_cfg_0004_038522062288#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001767/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 209, + 109 + ], + [ + 202, + 86 + ], + [ + 185, + 130 + ], + [ + 231, + 89 + ], + [ + 193, + 150 + ], + [ + 236, + 110 + ] + ], + [ + [ + 208, + 144 + ], + [ + 214, + 111 + ], + [ + 234, + 148 + ], + [ + 185, + 140 + ], + [ + 228, + 182 + ], + [ + 179, + 172 + ] + ], + [ + [ + 222, + 128 + ], + [ + 197, + 110 + ], + [ + 212, + 121 + ], + [ + 230, + 134 + ], + [ + 238, + 141 + ], + [ + 254, + 151 + ] + ], + [ + [ + 192, + 126 + ], + [ + 193, + 140 + ], + [ + 177, + 152 + ], + [ + 205, + 102 + ], + [ + 175, + 137 + ], + [ + 205, + 85 + ] + ] + ] + }, + { + "task_id": "0001769", + "id": "RH20T_cfg7_task_0004_user_0020_scene_0005_cfg_0007_104122064161#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001769/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 254 + ], + [ + 374, + 250 + ], + [ + 371, + 281 + ], + [ + 372, + 228 + ], + [ + 369, + 283 + ], + [ + 370, + 232 + ] + ], + [ + [ + 376, + 246 + ], + [ + 360, + 228 + ], + [ + 386, + 227 + ], + [ + 367, + 263 + ], + [ + 404, + 246 + ], + [ + 384, + 282 + ] + ], + [ + [ + 358, + 272 + ], + [ + 350, + 256 + ], + [ + 356, + 253 + ], + [ + 360, + 291 + ], + [ + 364, + 270 + ], + [ + 368, + 309 + ] + ], + [ + [ + 357, + 225 + ], + [ + 334, + 216 + ], + [ + 366, + 200 + ], + [ + 348, + 251 + ], + [ + 390, + 208 + ], + [ + 372, + 261 + ] + ] + ] + }, + { + "task_id": "0001773", + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522062547#257", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001773/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 485, + 228 + ], + [ + 509, + 239 + ], + [ + 496, + 234 + ], + [ + 472, + 222 + ], + [ + 474, + 224 + ], + [ + 449, + 212 + ] + ], + [ + [ + 492, + 229 + ], + [ + 470, + 209 + ], + [ + 506, + 215 + ], + [ + 477, + 243 + ], + [ + 529, + 234 + ], + [ + 501, + 265 + ] + ], + [ + [ + 472, + 210 + ], + [ + 478, + 200 + ], + [ + 499, + 209 + ], + [ + 444, + 211 + ], + [ + 492, + 218 + ], + [ + 440, + 221 + ] + ], + [ + [ + 468, + 206 + ], + [ + 466, + 232 + ], + [ + 473, + 202 + ], + [ + 464, + 210 + ], + [ + 476, + 176 + ], + [ + 467, + 187 + ] + ] + ] + }, + { + "task_id": "0001775", + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522062547#560", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001775/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 460, + 181 + ], + [ + 462, + 168 + ], + [ + 487, + 179 + ], + [ + 433, + 182 + ], + [ + 484, + 190 + ], + [ + 432, + 193 + ] + ], + [ + [ + 468, + 180 + ], + [ + 475, + 206 + ], + [ + 449, + 189 + ], + [ + 487, + 171 + ], + [ + 442, + 162 + ], + [ + 480, + 146 + ] + ], + [ + [ + 464, + 205 + ], + [ + 443, + 215 + ], + [ + 473, + 230 + ], + [ + 456, + 181 + ], + [ + 497, + 221 + ], + [ + 478, + 170 + ] + ], + [ + [ + 479, + 168 + ], + [ + 495, + 187 + ], + [ + 463, + 170 + ], + [ + 497, + 166 + ], + [ + 448, + 152 + ], + [ + 480, + 146 + ] + ] + ] + }, + { + "task_id": "0001777", + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522062547#889", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001777/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 446, + 159 + ], + [ + 452, + 165 + ], + [ + 424, + 172 + ], + [ + 468, + 146 + ], + [ + 420, + 166 + ], + [ + 462, + 142 + ] + ], + [ + [ + 461, + 149 + ], + [ + 463, + 136 + ], + [ + 487, + 146 + ], + [ + 434, + 151 + ], + [ + 484, + 158 + ], + [ + 433, + 163 + ] + ], + [ + [ + 474, + 124 + ], + [ + 486, + 122 + ], + [ + 453, + 139 + ], + [ + 495, + 110 + ], + [ + 443, + 140 + ], + [ + 483, + 113 + ] + ], + [ + [ + 455, + 131 + ], + [ + 463, + 105 + ], + [ + 440, + 125 + ], + [ + 472, + 138 + ], + [ + 432, + 149 + ], + [ + 463, + 165 + ] + ] + ] + }, + { + "task_id": "0001779", + "id": "RH20T_cfg5_task_0123_user_0010_scene_0004_cfg_0005_105422061350#285", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the knife on the knife rack'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001779/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 184, + 141 + ], + [ + 177, + 123 + ], + [ + 166, + 162 + ], + [ + 203, + 120 + ], + [ + 173, + 181 + ], + [ + 212, + 138 + ] + ], + [ + [ + 200, + 152 + ], + [ + 177, + 140 + ], + [ + 215, + 133 + ], + [ + 183, + 172 + ], + [ + 238, + 145 + ], + [ + 207, + 185 + ] + ], + [ + [ + 210, + 164 + ], + [ + 206, + 147 + ], + [ + 223, + 142 + ], + [ + 196, + 187 + ], + [ + 226, + 158 + ], + [ + 201, + 201 + ] + ], + [ + [ + 221, + 162 + ], + [ + 228, + 147 + ], + [ + 193, + 156 + ], + [ + 248, + 168 + ], + [ + 184, + 172 + ], + [ + 242, + 184 + ] + ] + ] + }, + { + "task_id": "0001780", + "id": "RH20T_cfg1_task_0066_user_0010_scene_0004_cfg_0001_039422060546#359", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001780/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 332, + 77 + ], + [ + 334, + 56 + ], + [ + 326, + 96 + ], + [ + 338, + 60 + ], + [ + 325, + 115 + ], + [ + 336, + 80 + ] + ], + [ + [ + 313, + 100 + ], + [ + 309, + 125 + ], + [ + 320, + 111 + ], + [ + 307, + 91 + ], + [ + 325, + 84 + ], + [ + 311, + 66 + ] + ], + [ + [ + 310, + 99 + ], + [ + 305, + 121 + ], + [ + 306, + 84 + ], + [ + 315, + 114 + ], + [ + 311, + 61 + ], + [ + 321, + 89 + ] + ], + [ + [ + 350, + 109 + ], + [ + 324, + 104 + ], + [ + 355, + 82 + ], + [ + 345, + 134 + ], + [ + 382, + 88 + ], + [ + 371, + 140 + ] + ] + ] + }, + { + "task_id": "0001783", + "id": "RH20T_cfg1_task_0066_user_0010_scene_0004_cfg_0001_039422060546#871", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001783/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 432, + 36 + ], + [ + 409, + 33 + ], + [ + 431, + 53 + ], + [ + 432, + 19 + ], + [ + 454, + 56 + ], + [ + 456, + 22 + ] + ], + [ + [ + 413, + 20 + ], + [ + 432, + 33 + ], + [ + 413, + 26 + ], + [ + 413, + 16 + ], + [ + 393, + 12 + ], + [ + 394, + 3 + ] + ], + [ + [ + 408, + 11 + ], + [ + 431, + 17 + ], + [ + 403, + 20 + ], + [ + 413, + 4 + ], + [ + 379, + 14 + ], + [ + 391, + 0 + ] + ], + [ + [ + 424, + 23 + ], + [ + 429, + 3 + ], + [ + 419, + 39 + ], + [ + 429, + 9 + ], + [ + 415, + 57 + ], + [ + 425, + 28 + ] + ] + ] + }, + { + "task_id": "0001786", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_037522061512#175", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001786/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 208, + 230 + ], + [ + 233, + 236 + ], + [ + 211, + 245 + ], + [ + 204, + 213 + ], + [ + 187, + 239 + ], + [ + 178, + 207 + ] + ], + [ + [ + 227, + 218 + ], + [ + 209, + 209 + ], + [ + 233, + 236 + ], + [ + 222, + 199 + ], + [ + 251, + 246 + ], + [ + 241, + 209 + ] + ], + [ + [ + 200, + 199 + ], + [ + 180, + 191 + ], + [ + 184, + 204 + ], + [ + 215, + 194 + ], + [ + 205, + 213 + ], + [ + 235, + 202 + ] + ], + [ + [ + 206, + 217 + ], + [ + 204, + 195 + ], + [ + 222, + 206 + ], + [ + 188, + 229 + ], + [ + 223, + 226 + ], + [ + 190, + 251 + ] + ] + ] + }, + { + "task_id": "0001787", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_037522061512#268", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001787/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 195, + 220 + ], + [ + 202, + 240 + ], + [ + 215, + 205 + ], + [ + 174, + 236 + ], + [ + 208, + 187 + ], + [ + 168, + 217 + ] + ], + [ + [ + 222, + 240 + ], + [ + 202, + 224 + ], + [ + 209, + 260 + ], + [ + 235, + 219 + ], + [ + 230, + 276 + ], + [ + 256, + 235 + ] + ], + [ + [ + 212, + 229 + ], + [ + 212, + 208 + ], + [ + 228, + 216 + ], + [ + 195, + 243 + ], + [ + 228, + 235 + ], + [ + 196, + 263 + ] + ], + [ + [ + 213, + 232 + ], + [ + 237, + 237 + ], + [ + 225, + 224 + ], + [ + 200, + 241 + ], + [ + 203, + 220 + ], + [ + 176, + 236 + ] + ] + ] + }, + { + "task_id": "0001788", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_037522061512#387", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001788/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 216, + 228 + ], + [ + 229, + 247 + ], + [ + 205, + 217 + ], + [ + 226, + 238 + ], + [ + 193, + 199 + ], + [ + 214, + 221 + ] + ], + [ + [ + 240, + 240 + ], + [ + 225, + 228 + ], + [ + 244, + 261 + ], + [ + 237, + 220 + ], + [ + 259, + 271 + ], + [ + 251, + 231 + ] + ], + [ + [ + 216, + 238 + ], + [ + 213, + 216 + ], + [ + 231, + 225 + ], + [ + 199, + 253 + ], + [ + 232, + 246 + ], + [ + 202, + 275 + ] + ], + [ + [ + 212, + 235 + ], + [ + 191, + 237 + ], + [ + 208, + 209 + ], + [ + 216, + 261 + ], + [ + 230, + 207 + ], + [ + 238, + 259 + ] + ] + ] + }, + { + "task_id": "0001791", + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_819112070044#101", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001791/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 489, + 161 + ], + [ + 473, + 153 + ], + [ + 494, + 174 + ], + [ + 483, + 148 + ], + [ + 510, + 184 + ], + [ + 500, + 157 + ] + ], + [ + [ + 504, + 171 + ], + [ + 479, + 171 + ], + [ + 501, + 179 + ], + [ + 507, + 164 + ], + [ + 528, + 179 + ], + [ + 532, + 163 + ] + ], + [ + [ + 505, + 167 + ], + [ + 512, + 162 + ], + [ + 496, + 145 + ], + [ + 514, + 188 + ], + [ + 487, + 149 + ], + [ + 507, + 193 + ] + ], + [ + [ + 482, + 164 + ], + [ + 488, + 146 + ], + [ + 483, + 150 + ], + [ + 481, + 178 + ], + [ + 478, + 167 + ], + [ + 476, + 195 + ] + ] + ] + }, + { + "task_id": "0001792", + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_819112070044#232", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001792/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 472, + 144 + ], + [ + 489, + 159 + ], + [ + 483, + 134 + ], + [ + 461, + 155 + ], + [ + 467, + 119 + ], + [ + 444, + 139 + ] + ], + [ + [ + 483, + 157 + ], + [ + 491, + 139 + ], + [ + 483, + 145 + ], + [ + 483, + 170 + ], + [ + 476, + 162 + ], + [ + 476, + 187 + ] + ], + [ + [ + 478, + 163 + ], + [ + 461, + 167 + ], + [ + 473, + 141 + ], + [ + 484, + 185 + ], + [ + 492, + 135 + ], + [ + 503, + 182 + ] + ], + [ + [ + 471, + 144 + ], + [ + 450, + 153 + ], + [ + 473, + 144 + ], + [ + 469, + 145 + ], + [ + 495, + 135 + ], + [ + 490, + 137 + ] + ] + ] + }, + { + "task_id": "0001794", + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_819112070044#533", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001794/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 472, + 174 + ], + [ + 473, + 161 + ], + [ + 463, + 187 + ], + [ + 481, + 160 + ], + [ + 461, + 202 + ], + [ + 480, + 175 + ] + ], + [ + [ + 474, + 168 + ], + [ + 496, + 175 + ], + [ + 470, + 179 + ], + [ + 479, + 156 + ], + [ + 450, + 173 + ], + [ + 456, + 149 + ] + ], + [ + [ + 490, + 170 + ], + [ + 498, + 153 + ], + [ + 492, + 159 + ], + [ + 487, + 182 + ], + [ + 484, + 175 + ], + [ + 479, + 198 + ] + ], + [ + [ + 485, + 168 + ], + [ + 473, + 160 + ], + [ + 476, + 154 + ], + [ + 495, + 183 + ], + [ + 487, + 162 + ], + [ + 507, + 191 + ] + ] + ] + }, + { + "task_id": "0001797", + "id": "RH20T_cfg6_task_0007_user_0014_scene_0007_cfg_0006_104122061330#183", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001797/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 243, + 290 + ], + [ + 238, + 314 + ], + [ + 248, + 271 + ], + [ + 238, + 311 + ], + [ + 254, + 247 + ], + [ + 244, + 285 + ] + ], + [ + [ + 271, + 285 + ], + [ + 283, + 259 + ], + [ + 246, + 277 + ], + [ + 297, + 294 + ], + [ + 233, + 304 + ], + [ + 285, + 322 + ] + ], + [ + [ + 253, + 261 + ], + [ + 235, + 264 + ], + [ + 246, + 290 + ], + [ + 260, + 232 + ], + [ + 262, + 285 + ], + [ + 275, + 230 + ] + ], + [ + [ + 261, + 280 + ], + [ + 283, + 294 + ], + [ + 279, + 259 + ], + [ + 243, + 303 + ], + [ + 259, + 247 + ], + [ + 222, + 289 + ] + ] + ] + }, + { + "task_id": "0001798", + "id": "RH20T_cfg6_task_0007_user_0014_scene_0007_cfg_0006_104122061330#201", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001798/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 254, + 239 + ], + [ + 257, + 259 + ], + [ + 282, + 238 + ], + [ + 225, + 240 + ], + [ + 280, + 217 + ], + [ + 221, + 218 + ] + ], + [ + [ + 224, + 263 + ], + [ + 240, + 265 + ], + [ + 232, + 291 + ], + [ + 216, + 235 + ], + [ + 217, + 288 + ], + [ + 202, + 234 + ] + ], + [ + [ + 293, + 243 + ], + [ + 317, + 256 + ], + [ + 283, + 269 + ], + [ + 304, + 217 + ], + [ + 261, + 256 + ], + [ + 281, + 205 + ] + ], + [ + [ + 257, + 262 + ], + [ + 240, + 265 + ], + [ + 250, + 290 + ], + [ + 265, + 234 + ], + [ + 265, + 286 + ], + [ + 280, + 233 + ] + ] + ] + }, + { + "task_id": "0001800", + "id": "RH20T_cfg5_task_0123_user_0010_scene_0004_cfg_0005_104422070044#366", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the knife on the knife rack'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001800/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 415, + 164 + ], + [ + 408, + 150 + ], + [ + 422, + 172 + ], + [ + 409, + 156 + ], + [ + 430, + 186 + ], + [ + 416, + 170 + ] + ], + [ + [ + 416, + 161 + ], + [ + 409, + 176 + ], + [ + 412, + 151 + ], + [ + 419, + 169 + ], + [ + 420, + 136 + ], + [ + 426, + 155 + ] + ], + [ + [ + 423, + 164 + ], + [ + 427, + 171 + ], + [ + 429, + 179 + ], + [ + 417, + 149 + ], + [ + 424, + 171 + ], + [ + 412, + 141 + ] + ], + [ + [ + 395, + 159 + ], + [ + 412, + 153 + ], + [ + 388, + 142 + ], + [ + 403, + 175 + ], + [ + 372, + 148 + ], + [ + 387, + 179 + ] + ] + ] + }, + { + "task_id": "0001801", + "id": "RH20T_cfg6_task_0046_user_0020_scene_0010_cfg_0006_104122061330#309", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001801/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 345, + 175 + ], + [ + 331, + 201 + ], + [ + 346, + 182 + ], + [ + 345, + 167 + ], + [ + 359, + 159 + ], + [ + 360, + 141 + ] + ], + [ + [ + 362, + 165 + ], + [ + 338, + 160 + ], + [ + 355, + 194 + ], + [ + 368, + 137 + ], + [ + 380, + 200 + ], + [ + 394, + 141 + ] + ], + [ + [ + 359, + 157 + ], + [ + 382, + 156 + ], + [ + 375, + 176 + ], + [ + 345, + 139 + ], + [ + 350, + 178 + ], + [ + 322, + 140 + ] + ], + [ + [ + 368, + 190 + ], + [ + 345, + 185 + ], + [ + 353, + 206 + ], + [ + 385, + 172 + ], + [ + 374, + 209 + ], + [ + 408, + 177 + ] + ] + ] + }, + { + "task_id": "0001802", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0001_cfg_0004_104122062823#103", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001802/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 292, + 19 + ], + [ + 260, + 40 + ], + [ + 311, + 51 + ], + [ + 272, + 0 + ], + [ + 344, + 30 + ], + [ + 304, + 0 + ] + ], + [ + [ + 302, + 30 + ], + [ + 295, + 5 + ], + [ + 312, + 3 + ], + [ + 293, + 54 + ], + [ + 321, + 31 + ], + [ + 300, + 80 + ] + ], + [ + [ + 296, + 15 + ], + [ + 296, + 0 + ], + [ + 333, + 18 + ], + [ + 260, + 11 + ], + [ + 331, + 50 + ], + [ + 262, + 44 + ] + ], + [ + [ + 299, + 26 + ], + [ + 295, + 35 + ], + [ + 330, + 42 + ], + [ + 268, + 10 + ], + [ + 332, + 34 + ], + [ + 273, + 4 + ] + ] + ] + }, + { + "task_id": "0001807", + "id": "RH20T_cfg2_task_0042_user_0007_scene_0002_cfg_0002_036422060215#700", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001807/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 308, + 206 + ], + [ + 326, + 215 + ], + [ + 313, + 217 + ], + [ + 304, + 194 + ], + [ + 296, + 208 + ], + [ + 287, + 186 + ] + ], + [ + [ + 319, + 230 + ], + [ + 326, + 245 + ], + [ + 300, + 233 + ], + [ + 338, + 227 + ], + [ + 292, + 218 + ], + [ + 330, + 211 + ] + ], + [ + [ + 318, + 224 + ], + [ + 323, + 208 + ], + [ + 338, + 227 + ], + [ + 297, + 222 + ], + [ + 332, + 242 + ], + [ + 292, + 237 + ] + ], + [ + [ + 321, + 228 + ], + [ + 328, + 246 + ], + [ + 338, + 229 + ], + [ + 303, + 226 + ], + [ + 331, + 213 + ], + [ + 296, + 209 + ] + ] + ] + }, + { + "task_id": "0001808", + "id": "RH20T_cfg2_task_0042_user_0007_scene_0002_cfg_0002_036422060215#885", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001808/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 312, + 222 + ], + [ + 317, + 205 + ], + [ + 331, + 220 + ], + [ + 292, + 223 + ], + [ + 326, + 235 + ], + [ + 287, + 239 + ] + ], + [ + [ + 306, + 218 + ], + [ + 314, + 206 + ], + [ + 310, + 234 + ], + [ + 302, + 201 + ], + [ + 302, + 247 + ], + [ + 293, + 213 + ] + ], + [ + [ + 327, + 210 + ], + [ + 338, + 224 + ], + [ + 340, + 194 + ], + [ + 313, + 226 + ], + [ + 329, + 182 + ], + [ + 303, + 213 + ] + ], + [ + [ + 304, + 229 + ], + [ + 291, + 232 + ], + [ + 318, + 220 + ], + [ + 289, + 238 + ], + [ + 333, + 216 + ], + [ + 304, + 234 + ] + ] + ] + }, + { + "task_id": "0001810", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_037522061512#213", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001810/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 182, + 162 + ], + [ + 166, + 155 + ], + [ + 188, + 141 + ], + [ + 176, + 183 + ], + [ + 205, + 148 + ], + [ + 192, + 191 + ] + ], + [ + [ + 210, + 151 + ], + [ + 206, + 135 + ], + [ + 223, + 138 + ], + [ + 195, + 165 + ], + [ + 226, + 153 + ], + [ + 199, + 180 + ] + ], + [ + [ + 225, + 171 + ], + [ + 218, + 181 + ], + [ + 205, + 175 + ], + [ + 245, + 166 + ], + [ + 212, + 165 + ], + [ + 250, + 156 + ] + ], + [ + [ + 207, + 143 + ], + [ + 191, + 136 + ], + [ + 204, + 162 + ], + [ + 211, + 123 + ], + [ + 221, + 169 + ], + [ + 229, + 130 + ] + ] + ] + }, + { + "task_id": "0001811", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_037522061512#561", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001811/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 263, + 110 + ], + [ + 259, + 106 + ], + [ + 269, + 88 + ], + [ + 257, + 131 + ], + [ + 274, + 91 + ], + [ + 261, + 136 + ] + ], + [ + [ + 247, + 109 + ], + [ + 264, + 106 + ], + [ + 234, + 110 + ], + [ + 260, + 108 + ], + [ + 216, + 114 + ], + [ + 242, + 112 + ] + ], + [ + [ + 273, + 123 + ], + [ + 293, + 117 + ], + [ + 275, + 106 + ], + [ + 272, + 140 + ], + [ + 257, + 112 + ], + [ + 253, + 146 + ] + ], + [ + [ + 256, + 124 + ], + [ + 263, + 104 + ], + [ + 267, + 128 + ], + [ + 243, + 121 + ], + [ + 260, + 147 + ], + [ + 235, + 141 + ] + ] + ] + }, + { + "task_id": "0001812", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_037522061512#852", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001812/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 245, + 132 + ], + [ + 264, + 138 + ], + [ + 245, + 148 + ], + [ + 246, + 115 + ], + [ + 227, + 143 + ], + [ + 227, + 109 + ] + ], + [ + [ + 228, + 98 + ], + [ + 232, + 117 + ], + [ + 208, + 102 + ], + [ + 246, + 94 + ], + [ + 204, + 83 + ], + [ + 242, + 74 + ] + ], + [ + [ + 232, + 106 + ], + [ + 225, + 97 + ], + [ + 244, + 90 + ], + [ + 221, + 123 + ], + [ + 250, + 99 + ], + [ + 228, + 131 + ] + ], + [ + [ + 247, + 114 + ], + [ + 250, + 116 + ], + [ + 236, + 132 + ], + [ + 258, + 97 + ], + [ + 231, + 131 + ], + [ + 254, + 94 + ] + ] + ] + }, + { + "task_id": "0001813", + "id": "RH20T_cfg5_task_0010_user_0010_scene_0009_cfg_0005_105422061350#335", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001813/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 218, + 104 + ], + [ + 211, + 82 + ], + [ + 201, + 124 + ], + [ + 237, + 82 + ], + [ + 208, + 148 + ], + [ + 245, + 106 + ] + ], + [ + [ + 195, + 145 + ], + [ + 194, + 175 + ], + [ + 226, + 147 + ], + [ + 163, + 143 + ], + [ + 228, + 117 + ], + [ + 165, + 111 + ] + ], + [ + [ + 199, + 90 + ], + [ + 177, + 70 + ], + [ + 212, + 99 + ], + [ + 187, + 81 + ], + [ + 232, + 117 + ], + [ + 206, + 99 + ] + ], + [ + [ + 195, + 109 + ], + [ + 197, + 91 + ], + [ + 209, + 86 + ], + [ + 180, + 133 + ], + [ + 207, + 102 + ], + [ + 180, + 148 + ] + ] + ] + }, + { + "task_id": "0001814", + "id": "RH20T_cfg3_task_0008_user_0016_scene_0008_cfg_0003_036422060909#173", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001814/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 356, + 72 + ], + [ + 360, + 55 + ], + [ + 371, + 62 + ], + [ + 342, + 82 + ], + [ + 367, + 78 + ], + [ + 339, + 98 + ] + ], + [ + [ + 374, + 72 + ], + [ + 359, + 59 + ], + [ + 381, + 55 + ], + [ + 368, + 87 + ], + [ + 397, + 68 + ], + [ + 383, + 100 + ] + ], + [ + [ + 368, + 83 + ], + [ + 367, + 98 + ], + [ + 356, + 69 + ], + [ + 379, + 95 + ], + [ + 357, + 54 + ], + [ + 380, + 81 + ] + ], + [ + [ + 338, + 79 + ], + [ + 329, + 60 + ], + [ + 337, + 74 + ], + [ + 339, + 84 + ], + [ + 347, + 94 + ], + [ + 349, + 102 + ] + ] + ] + }, + { + "task_id": "0001815", + "id": "RH20T_cfg3_task_0038_user_0010_scene_0010_cfg_0003_104122063550#97", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001815/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 436, + 169 + ], + [ + 408, + 180 + ], + [ + 441, + 197 + ], + [ + 431, + 138 + ], + [ + 467, + 186 + ], + [ + 459, + 127 + ] + ], + [ + [ + 408, + 188 + ], + [ + 417, + 175 + ], + [ + 417, + 217 + ], + [ + 399, + 160 + ], + [ + 408, + 228 + ], + [ + 392, + 173 + ] + ], + [ + [ + 430, + 170 + ], + [ + 462, + 169 + ], + [ + 437, + 191 + ], + [ + 424, + 150 + ], + [ + 403, + 192 + ], + [ + 394, + 151 + ] + ], + [ + [ + 378, + 191 + ], + [ + 352, + 204 + ], + [ + 392, + 216 + ], + [ + 365, + 165 + ], + [ + 419, + 204 + ], + [ + 392, + 151 + ] + ] + ] + }, + { + "task_id": "0001819", + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_035622060973#274", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001819/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 224, + 111 + ], + [ + 222, + 99 + ], + [ + 205, + 107 + ], + [ + 243, + 115 + ], + [ + 208, + 118 + ], + [ + 244, + 126 + ] + ], + [ + [ + 224, + 110 + ], + [ + 210, + 119 + ], + [ + 225, + 96 + ], + [ + 222, + 125 + ], + [ + 239, + 87 + ], + [ + 237, + 116 + ] + ], + [ + [ + 218, + 116 + ], + [ + 223, + 100 + ], + [ + 228, + 126 + ], + [ + 207, + 106 + ], + [ + 223, + 143 + ], + [ + 202, + 123 + ] + ], + [ + [ + 211, + 100 + ], + [ + 207, + 84 + ], + [ + 193, + 95 + ], + [ + 227, + 104 + ], + [ + 196, + 113 + ], + [ + 230, + 120 + ] + ] + ] + }, + { + "task_id": "0001820", + "id": "RH20T_cfg1_task_0001_user_0009_scene_0003_cfg_0001_039422060546#64", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001820/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 252, + 170 + ], + [ + 234, + 166 + ], + [ + 225, + 164 + ], + [ + 278, + 176 + ], + [ + 245, + 168 + ], + [ + 299, + 180 + ] + ], + [ + [ + 273, + 132 + ], + [ + 300, + 152 + ], + [ + 293, + 106 + ], + [ + 253, + 159 + ], + [ + 266, + 86 + ], + [ + 226, + 139 + ] + ], + [ + [ + 283, + 177 + ], + [ + 315, + 185 + ], + [ + 283, + 183 + ], + [ + 284, + 171 + ], + [ + 249, + 175 + ], + [ + 253, + 163 + ] + ], + [ + [ + 269, + 155 + ], + [ + 265, + 133 + ], + [ + 255, + 182 + ], + [ + 281, + 131 + ], + [ + 259, + 201 + ], + [ + 284, + 151 + ] + ] + ] + }, + { + "task_id": "0001825", + "id": "RH20T_cfg1_task_0041_user_0001_scene_0006_cfg_0001_819112070044#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the tabletop with a sponge'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001825/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 370, + 226 + ], + [ + 370, + 207 + ], + [ + 352, + 229 + ], + [ + 389, + 222 + ], + [ + 352, + 248 + ], + [ + 390, + 242 + ] + ], + [ + [ + 371, + 204 + ], + [ + 364, + 188 + ], + [ + 351, + 209 + ], + [ + 391, + 198 + ], + [ + 358, + 226 + ], + [ + 400, + 214 + ] + ], + [ + [ + 391, + 236 + ], + [ + 408, + 230 + ], + [ + 400, + 255 + ], + [ + 381, + 217 + ], + [ + 384, + 260 + ], + [ + 366, + 223 + ] + ], + [ + [ + 367, + 219 + ], + [ + 371, + 205 + ], + [ + 385, + 214 + ], + [ + 348, + 224 + ], + [ + 381, + 227 + ], + [ + 345, + 237 + ] + ] + ] + }, + { + "task_id": "0001826", + "id": "RH20T_cfg6_task_0125_user_0020_scene_0004_cfg_0006_104122064161#434", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull up the lever'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001826/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 462, + 137 + ], + [ + 440, + 134 + ], + [ + 468, + 119 + ], + [ + 457, + 154 + ], + [ + 491, + 123 + ], + [ + 480, + 158 + ] + ], + [ + [ + 437, + 136 + ], + [ + 439, + 122 + ], + [ + 441, + 152 + ], + [ + 434, + 120 + ], + [ + 439, + 165 + ], + [ + 432, + 133 + ] + ], + [ + [ + 447, + 116 + ], + [ + 457, + 98 + ], + [ + 456, + 113 + ], + [ + 437, + 120 + ], + [ + 446, + 129 + ], + [ + 427, + 138 + ] + ], + [ + [ + 451, + 131 + ], + [ + 472, + 126 + ], + [ + 460, + 143 + ], + [ + 443, + 120 + ], + [ + 438, + 148 + ], + [ + 423, + 125 + ] + ] + ] + }, + { + "task_id": "0001831", + "id": "RH20T_cfg6_task_0126_user_0014_scene_0006_cfg_0006_104122064161#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord to the socket'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001831/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 400, + 146 + ], + [ + 401, + 151 + ], + [ + 380, + 160 + ], + [ + 421, + 132 + ], + [ + 381, + 155 + ], + [ + 420, + 129 + ] + ], + [ + [ + 408, + 155 + ], + [ + 392, + 146 + ], + [ + 426, + 149 + ], + [ + 392, + 160 + ], + [ + 441, + 159 + ], + [ + 408, + 168 + ] + ], + [ + [ + 428, + 169 + ], + [ + 409, + 164 + ], + [ + 433, + 147 + ], + [ + 423, + 191 + ], + [ + 452, + 153 + ], + [ + 442, + 196 + ] + ], + [ + [ + 416, + 158 + ], + [ + 420, + 137 + ], + [ + 420, + 170 + ], + [ + 414, + 147 + ], + [ + 416, + 191 + ], + [ + 411, + 167 + ] + ] + ] + }, + { + "task_id": "0001833", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_750612070851#185", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001833/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 261 + ], + [ + 391, + 295 + ], + [ + 437, + 267 + ], + [ + 378, + 256 + ], + [ + 451, + 232 + ], + [ + 392, + 226 + ] + ], + [ + [ + 414, + 254 + ], + [ + 407, + 229 + ], + [ + 407, + 281 + ], + [ + 421, + 225 + ], + [ + 414, + 310 + ], + [ + 430, + 254 + ] + ], + [ + [ + 415, + 282 + ], + [ + 420, + 265 + ], + [ + 378, + 286 + ], + [ + 451, + 277 + ], + [ + 375, + 300 + ], + [ + 444, + 292 + ] + ], + [ + [ + 441, + 266 + ], + [ + 446, + 304 + ], + [ + 430, + 251 + ], + [ + 451, + 280 + ], + [ + 426, + 215 + ], + [ + 446, + 247 + ] + ] + ] + }, + { + "task_id": "0001834", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0010_cfg_0001_750612070851#279", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001834/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 492, + 299 + ], + [ + 481, + 265 + ], + [ + 530, + 291 + ], + [ + 456, + 308 + ], + [ + 544, + 328 + ], + [ + 466, + 344 + ] + ], + [ + [ + 453, + 339 + ], + [ + 421, + 360 + ], + [ + 466, + 360 + ], + [ + 440, + 306 + ], + [ + 495, + 349 + ], + [ + 471, + 285 + ] + ], + [ + [ + 453, + 292 + ], + [ + 455, + 266 + ], + [ + 440, + 267 + ], + [ + 465, + 315 + ], + [ + 438, + 295 + ], + [ + 465, + 344 + ] + ], + [ + [ + 477, + 301 + ], + [ + 488, + 284 + ], + [ + 441, + 309 + ], + [ + 513, + 294 + ], + [ + 433, + 324 + ], + [ + 501, + 309 + ] + ] + ] + }, + { + "task_id": "0001837", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0008_cfg_0006_104122061018#124", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001837/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 373, + 110 + ], + [ + 398, + 110 + ], + [ + 389, + 120 + ], + [ + 358, + 100 + ], + [ + 362, + 119 + ], + [ + 333, + 99 + ] + ], + [ + [ + 359, + 156 + ], + [ + 372, + 168 + ], + [ + 380, + 169 + ], + [ + 337, + 141 + ], + [ + 368, + 158 + ], + [ + 326, + 131 + ] + ], + [ + [ + 363, + 128 + ], + [ + 354, + 100 + ], + [ + 356, + 130 + ], + [ + 371, + 125 + ], + [ + 365, + 156 + ], + [ + 381, + 155 + ] + ], + [ + [ + 369, + 105 + ], + [ + 387, + 126 + ], + [ + 375, + 92 + ], + [ + 362, + 120 + ], + [ + 357, + 71 + ], + [ + 342, + 98 + ] + ] + ] + }, + { + "task_id": "0001839", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0008_cfg_0006_104122061018#421", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001839/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 158 + ], + [ + 338, + 129 + ], + [ + 342, + 161 + ], + [ + 356, + 154 + ], + [ + 351, + 189 + ], + [ + 367, + 185 + ] + ], + [ + [ + 360, + 175 + ], + [ + 364, + 157 + ], + [ + 388, + 166 + ], + [ + 330, + 184 + ], + [ + 382, + 183 + ], + [ + 328, + 201 + ] + ], + [ + [ + 345, + 136 + ], + [ + 353, + 128 + ], + [ + 373, + 144 + ], + [ + 318, + 127 + ], + [ + 365, + 154 + ], + [ + 307, + 136 + ] + ], + [ + [ + 370, + 168 + ], + [ + 360, + 180 + ], + [ + 382, + 196 + ], + [ + 359, + 139 + ], + [ + 390, + 183 + ], + [ + 369, + 129 + ] + ] + ] + }, + { + "task_id": "0001840", + "id": "RH20T_cfg1_task_0010_user_0001_scene_0005_cfg_0001_819112071037#372", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001840/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 483, + 270 + ], + [ + 574, + 234 + ], + [ + 426, + 303 + ], + [ + 576, + 217 + ], + [ + 330, + 342 + ], + [ + 434, + 271 + ] + ], + [ + [ + 632, + 331 + ], + [ + 635, + 360 + ], + [ + 640, + 304 + ], + [ + 531, + 360 + ], + [ + 640, + 196 + ], + [ + 522, + 242 + ] + ], + [ + [ + 564, + 360 + ], + [ + 545, + 360 + ], + [ + 640, + 360 + ], + [ + 415, + 360 + ], + [ + 640, + 360 + ], + [ + 452, + 360 + ] + ], + [ + [ + 622, + 360 + ], + [ + 640, + 291 + ], + [ + 640, + 360 + ], + [ + 524, + 289 + ], + [ + 640, + 360 + ], + [ + 491, + 360 + ] + ] + ] + }, + { + "task_id": "0001841", + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_038522062288#171", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001841/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 179, + 136 + ], + [ + 155, + 158 + ], + [ + 175, + 127 + ], + [ + 183, + 144 + ], + [ + 200, + 105 + ], + [ + 204, + 124 + ] + ], + [ + [ + 139, + 129 + ], + [ + 148, + 120 + ], + [ + 146, + 101 + ], + [ + 132, + 158 + ], + [ + 138, + 110 + ], + [ + 124, + 164 + ] + ], + [ + [ + 137, + 130 + ], + [ + 123, + 108 + ], + [ + 112, + 116 + ], + [ + 162, + 144 + ], + [ + 125, + 135 + ], + [ + 173, + 162 + ] + ], + [ + [ + 115, + 139 + ], + [ + 84, + 158 + ], + [ + 116, + 131 + ], + [ + 115, + 147 + ], + [ + 147, + 112 + ], + [ + 143, + 130 + ] + ] + ] + }, + { + "task_id": "0001842", + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_038522062288#212", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001842/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 142, + 135 + ], + [ + 123, + 112 + ], + [ + 116, + 127 + ], + [ + 169, + 143 + ], + [ + 134, + 147 + ], + [ + 185, + 162 + ] + ], + [ + [ + 132, + 138 + ], + [ + 136, + 160 + ], + [ + 93, + 135 + ], + [ + 169, + 141 + ], + [ + 85, + 110 + ], + [ + 166, + 118 + ] + ], + [ + [ + 146, + 103 + ], + [ + 127, + 104 + ], + [ + 111, + 106 + ], + [ + 182, + 101 + ], + [ + 130, + 105 + ], + [ + 196, + 100 + ] + ], + [ + [ + 114, + 164 + ], + [ + 121, + 131 + ], + [ + 136, + 177 + ], + [ + 93, + 151 + ], + [ + 128, + 209 + ], + [ + 87, + 181 + ] + ] + ] + }, + { + "task_id": "0001843", + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_038522062288#250", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001843/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 121, + 165 + ], + [ + 103, + 191 + ], + [ + 89, + 153 + ], + [ + 150, + 175 + ], + [ + 108, + 125 + ], + [ + 168, + 150 + ] + ], + [ + [ + 121, + 147 + ], + [ + 105, + 127 + ], + [ + 95, + 136 + ], + [ + 149, + 160 + ], + [ + 110, + 154 + ], + [ + 162, + 177 + ] + ], + [ + [ + 133, + 125 + ], + [ + 128, + 93 + ], + [ + 98, + 124 + ], + [ + 163, + 126 + ], + [ + 103, + 158 + ], + [ + 168, + 157 + ] + ], + [ + [ + 97, + 152 + ], + [ + 117, + 175 + ], + [ + 61, + 168 + ], + [ + 132, + 136 + ], + [ + 35, + 143 + ], + [ + 111, + 109 + ] + ] + ] + }, + { + "task_id": "0001844", + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_038522062288#318", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001844/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 83, + 124 + ], + [ + 105, + 97 + ], + [ + 63, + 106 + ], + [ + 100, + 141 + ], + [ + 38, + 136 + ], + [ + 79, + 168 + ] + ], + [ + [ + 128, + 134 + ], + [ + 120, + 150 + ], + [ + 90, + 124 + ], + [ + 164, + 144 + ], + [ + 98, + 105 + ], + [ + 175, + 127 + ] + ], + [ + [ + 118, + 145 + ], + [ + 100, + 124 + ], + [ + 91, + 135 + ], + [ + 146, + 157 + ], + [ + 108, + 153 + ], + [ + 160, + 175 + ] + ], + [ + [ + 124, + 128 + ], + [ + 135, + 154 + ], + [ + 148, + 136 + ], + [ + 100, + 120 + ], + [ + 137, + 107 + ], + [ + 87, + 92 + ] + ] + ] + }, + { + "task_id": "0001850", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_819112071037#963", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001850/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 294, + 211 + ], + [ + 296, + 141 + ], + [ + 219, + 207 + ], + [ + 367, + 214 + ], + [ + 221, + 270 + ], + [ + 360, + 275 + ] + ], + [ + [ + 324, + 259 + ], + [ + 377, + 209 + ], + [ + 277, + 198 + ], + [ + 367, + 315 + ], + [ + 213, + 253 + ], + [ + 312, + 360 + ] + ], + [ + [ + 240, + 180 + ], + [ + 308, + 161 + ], + [ + 177, + 162 + ], + [ + 287, + 192 + ], + [ + 116, + 182 + ], + [ + 233, + 205 + ] + ], + [ + [ + 209, + 256 + ], + [ + 217, + 217 + ], + [ + 171, + 187 + ], + [ + 242, + 316 + ], + [ + 150, + 231 + ], + [ + 238, + 360 + ] + ] + ] + }, + { + "task_id": "0001852", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_819112071037#1583", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001852/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 453, + 300 + ], + [ + 471, + 223 + ], + [ + 370, + 296 + ], + [ + 534, + 305 + ], + [ + 359, + 360 + ], + [ + 515, + 360 + ] + ], + [ + [ + 406, + 326 + ], + [ + 356, + 360 + ], + [ + 394, + 346 + ], + [ + 423, + 298 + ], + [ + 433, + 289 + ], + [ + 471, + 227 + ] + ], + [ + [ + 366, + 355 + ], + [ + 326, + 299 + ], + [ + 307, + 351 + ], + [ + 439, + 359 + ], + [ + 342, + 360 + ], + [ + 474, + 360 + ] + ], + [ + [ + 417, + 360 + ], + [ + 431, + 330 + ], + [ + 480, + 360 + ], + [ + 349, + 349 + ], + [ + 471, + 360 + ], + [ + 316, + 360 + ] + ] + ] + }, + { + "task_id": "0001855", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122063633#249", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001855/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 134, + 158 + ], + [ + 127, + 139 + ], + [ + 121, + 150 + ], + [ + 148, + 167 + ], + [ + 128, + 168 + ], + [ + 155, + 185 + ] + ], + [ + [ + 148, + 163 + ], + [ + 145, + 149 + ], + [ + 152, + 145 + ], + [ + 144, + 181 + ], + [ + 155, + 158 + ], + [ + 148, + 194 + ] + ], + [ + [ + 154, + 153 + ], + [ + 132, + 161 + ], + [ + 153, + 146 + ], + [ + 155, + 159 + ], + [ + 176, + 138 + ], + [ + 176, + 152 + ] + ], + [ + [ + 161, + 151 + ], + [ + 156, + 145 + ], + [ + 154, + 131 + ], + [ + 168, + 171 + ], + [ + 159, + 137 + ], + [ + 172, + 175 + ] + ] + ] + }, + { + "task_id": "0001859", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122061330#241", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001859/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 316, + 214 + ], + [ + 305, + 211 + ], + [ + 307, + 241 + ], + [ + 325, + 188 + ], + [ + 317, + 242 + ], + [ + 335, + 192 + ] + ], + [ + [ + 337, + 227 + ], + [ + 358, + 215 + ], + [ + 322, + 205 + ], + [ + 352, + 250 + ], + [ + 300, + 216 + ], + [ + 330, + 263 + ] + ], + [ + [ + 317, + 226 + ], + [ + 320, + 218 + ], + [ + 323, + 202 + ], + [ + 311, + 251 + ], + [ + 320, + 211 + ], + [ + 309, + 257 + ] + ], + [ + [ + 303, + 225 + ], + [ + 327, + 237 + ], + [ + 307, + 235 + ], + [ + 298, + 214 + ], + [ + 285, + 225 + ], + [ + 273, + 203 + ] + ] + ] + }, + { + "task_id": "0001860", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122061330#320", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001860/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 258, + 220 + ], + [ + 257, + 246 + ], + [ + 230, + 217 + ], + [ + 286, + 222 + ], + [ + 232, + 193 + ], + [ + 286, + 199 + ] + ], + [ + [ + 278, + 256 + ], + [ + 277, + 253 + ], + [ + 282, + 228 + ], + [ + 274, + 284 + ], + [ + 284, + 230 + ], + [ + 275, + 289 + ] + ], + [ + [ + 280, + 243 + ], + [ + 295, + 248 + ], + [ + 267, + 270 + ], + [ + 292, + 216 + ], + [ + 254, + 264 + ], + [ + 279, + 213 + ] + ], + [ + [ + 275, + 228 + ], + [ + 260, + 228 + ], + [ + 266, + 255 + ], + [ + 283, + 202 + ], + [ + 280, + 254 + ], + [ + 296, + 204 + ] + ] + ] + }, + { + "task_id": "0001861", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122061330#460", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001861/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 282, + 224 + ], + [ + 270, + 223 + ], + [ + 273, + 250 + ], + [ + 291, + 197 + ], + [ + 284, + 249 + ], + [ + 301, + 199 + ] + ], + [ + [ + 291, + 211 + ], + [ + 319, + 215 + ], + [ + 290, + 210 + ], + [ + 291, + 212 + ], + [ + 264, + 207 + ], + [ + 261, + 208 + ] + ], + [ + [ + 272, + 194 + ], + [ + 248, + 180 + ], + [ + 276, + 203 + ], + [ + 267, + 182 + ], + [ + 299, + 217 + ], + [ + 294, + 197 + ] + ], + [ + [ + 257, + 231 + ], + [ + 245, + 215 + ], + [ + 267, + 208 + ], + [ + 247, + 252 + ], + [ + 281, + 224 + ], + [ + 259, + 270 + ] + ] + ] + }, + { + "task_id": "0001862", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522062547#104", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001862/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 472, + 215 + ], + [ + 495, + 210 + ], + [ + 476, + 189 + ], + [ + 469, + 240 + ], + [ + 451, + 194 + ], + [ + 446, + 246 + ] + ], + [ + [ + 468, + 203 + ], + [ + 475, + 193 + ], + [ + 477, + 229 + ], + [ + 459, + 179 + ], + [ + 471, + 237 + ], + [ + 454, + 189 + ] + ], + [ + [ + 445, + 212 + ], + [ + 418, + 216 + ], + [ + 452, + 234 + ], + [ + 440, + 191 + ], + [ + 481, + 230 + ], + [ + 466, + 188 + ] + ], + [ + [ + 452, + 210 + ], + [ + 481, + 209 + ], + [ + 453, + 208 + ], + [ + 452, + 213 + ], + [ + 427, + 208 + ], + [ + 423, + 213 + ] + ] + ] + }, + { + "task_id": "0001863", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522062547#221", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001863/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 405, + 206 + ], + [ + 380, + 209 + ], + [ + 398, + 185 + ], + [ + 411, + 226 + ], + [ + 424, + 182 + ], + [ + 436, + 223 + ] + ], + [ + [ + 403, + 193 + ], + [ + 405, + 184 + ], + [ + 414, + 217 + ], + [ + 393, + 170 + ], + [ + 412, + 225 + ], + [ + 392, + 179 + ] + ], + [ + [ + 384, + 202 + ], + [ + 395, + 222 + ], + [ + 410, + 193 + ], + [ + 359, + 211 + ], + [ + 399, + 171 + ], + [ + 347, + 191 + ] + ], + [ + [ + 434, + 190 + ], + [ + 454, + 192 + ], + [ + 440, + 162 + ], + [ + 428, + 217 + ], + [ + 419, + 160 + ], + [ + 406, + 216 + ] + ] + ] + }, + { + "task_id": "0001866", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0006_cfg_0003_036422060909#178", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001866/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 94 + ], + [ + 362, + 71 + ], + [ + 349, + 103 + ], + [ + 376, + 87 + ], + [ + 350, + 125 + ], + [ + 376, + 108 + ] + ], + [ + [ + 362, + 86 + ], + [ + 369, + 68 + ], + [ + 377, + 76 + ], + [ + 346, + 96 + ], + [ + 371, + 93 + ], + [ + 340, + 113 + ] + ], + [ + [ + 359, + 65 + ], + [ + 343, + 52 + ], + [ + 343, + 76 + ], + [ + 375, + 55 + ], + [ + 360, + 91 + ], + [ + 392, + 68 + ] + ], + [ + [ + 382, + 66 + ], + [ + 366, + 61 + ], + [ + 394, + 44 + ], + [ + 369, + 88 + ], + [ + 412, + 48 + ], + [ + 386, + 95 + ] + ] + ] + }, + { + "task_id": "0001867", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0006_cfg_0003_036422060909#245", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001867/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 308, + 120 + ], + [ + 312, + 103 + ], + [ + 331, + 114 + ], + [ + 284, + 126 + ], + [ + 327, + 129 + ], + [ + 281, + 141 + ] + ], + [ + [ + 302, + 123 + ], + [ + 309, + 150 + ], + [ + 328, + 115 + ], + [ + 277, + 131 + ], + [ + 321, + 88 + ], + [ + 270, + 105 + ] + ], + [ + [ + 332, + 127 + ], + [ + 338, + 148 + ], + [ + 356, + 117 + ], + [ + 309, + 137 + ], + [ + 350, + 97 + ], + [ + 304, + 117 + ] + ], + [ + [ + 302, + 119 + ], + [ + 312, + 142 + ], + [ + 297, + 134 + ], + [ + 307, + 103 + ], + [ + 288, + 113 + ], + [ + 297, + 80 + ] + ] + ] + }, + { + "task_id": "0001869", + "id": "RH20T_cfg2_task_0104_user_0007_scene_0007_cfg_0002_104122061850#156", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate back after holding it down'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001869/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 352, + 187 + ], + [ + 367, + 193 + ], + [ + 337, + 171 + ], + [ + 366, + 203 + ], + [ + 325, + 166 + ], + [ + 353, + 197 + ] + ], + [ + [ + 344, + 196 + ], + [ + 341, + 179 + ], + [ + 329, + 214 + ], + [ + 359, + 180 + ], + [ + 333, + 229 + ], + [ + 361, + 195 + ] + ], + [ + [ + 344, + 179 + ], + [ + 327, + 160 + ], + [ + 363, + 166 + ], + [ + 323, + 194 + ], + [ + 381, + 185 + ], + [ + 341, + 214 + ] + ], + [ + [ + 320, + 191 + ], + [ + 340, + 203 + ], + [ + 325, + 205 + ], + [ + 315, + 176 + ], + [ + 307, + 194 + ], + [ + 296, + 165 + ] + ] + ] + }, + { + "task_id": "0001870", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_104122062823#223", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001870/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 495, + 60 + ], + [ + 495, + 37 + ], + [ + 500, + 81 + ], + [ + 490, + 37 + ], + [ + 500, + 106 + ], + [ + 490, + 63 + ] + ], + [ + [ + 465, + 79 + ], + [ + 470, + 53 + ], + [ + 496, + 87 + ], + [ + 437, + 71 + ], + [ + 490, + 110 + ], + [ + 434, + 94 + ] + ], + [ + [ + 444, + 56 + ], + [ + 459, + 41 + ], + [ + 435, + 26 + ], + [ + 452, + 83 + ], + [ + 418, + 41 + ], + [ + 436, + 100 + ] + ], + [ + [ + 454, + 75 + ], + [ + 471, + 57 + ], + [ + 434, + 54 + ], + [ + 475, + 96 + ], + [ + 421, + 71 + ], + [ + 458, + 111 + ] + ] + ] + }, + { + "task_id": "0001871", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_104122062823#403", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001871/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 417, + 103 + ], + [ + 398, + 116 + ], + [ + 406, + 119 + ], + [ + 427, + 88 + ], + [ + 426, + 105 + ], + [ + 447, + 73 + ] + ], + [ + [ + 429, + 62 + ], + [ + 415, + 55 + ], + [ + 455, + 48 + ], + [ + 405, + 74 + ], + [ + 467, + 55 + ], + [ + 418, + 79 + ] + ], + [ + [ + 455, + 64 + ], + [ + 468, + 48 + ], + [ + 431, + 76 + ], + [ + 479, + 52 + ], + [ + 421, + 89 + ], + [ + 466, + 66 + ] + ], + [ + [ + 441, + 73 + ], + [ + 444, + 47 + ], + [ + 470, + 81 + ], + [ + 415, + 65 + ], + [ + 466, + 105 + ], + [ + 414, + 88 + ] + ] + ] + }, + { + "task_id": "0001872", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_104122062823#782", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001872/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 413, + 69 + ], + [ + 402, + 62 + ], + [ + 386, + 66 + ], + [ + 439, + 73 + ], + [ + 396, + 75 + ], + [ + 452, + 81 + ] + ], + [ + [ + 436, + 44 + ], + [ + 431, + 35 + ], + [ + 462, + 23 + ], + [ + 411, + 65 + ], + [ + 468, + 33 + ], + [ + 415, + 76 + ] + ], + [ + [ + 447, + 74 + ], + [ + 452, + 43 + ], + [ + 438, + 76 + ], + [ + 454, + 73 + ], + [ + 433, + 107 + ], + [ + 449, + 101 + ] + ], + [ + [ + 421, + 73 + ], + [ + 423, + 47 + ], + [ + 451, + 75 + ], + [ + 392, + 71 + ], + [ + 448, + 99 + ], + [ + 391, + 94 + ] + ] + ] + }, + { + "task_id": "0001873", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_104122062823#948", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001873/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 355, + 66 + ], + [ + 340, + 76 + ], + [ + 344, + 39 + ], + [ + 364, + 92 + ], + [ + 358, + 31 + ], + [ + 377, + 83 + ] + ], + [ + [ + 392, + 79 + ], + [ + 410, + 60 + ], + [ + 372, + 59 + ], + [ + 412, + 100 + ], + [ + 353, + 78 + ], + [ + 394, + 120 + ] + ], + [ + [ + 383, + 62 + ], + [ + 385, + 37 + ], + [ + 408, + 71 + ], + [ + 359, + 54 + ], + [ + 406, + 95 + ], + [ + 359, + 78 + ] + ], + [ + [ + 355, + 81 + ], + [ + 360, + 55 + ], + [ + 374, + 90 + ], + [ + 333, + 70 + ], + [ + 370, + 116 + ], + [ + 328, + 97 + ] + ] + ] + }, + { + "task_id": "0001874", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0009_cfg_0004_104122062295#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001874/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 141, + 120 + ], + [ + 196, + 100 + ], + [ + 161, + 169 + ], + [ + 119, + 63 + ], + [ + 105, + 193 + ], + [ + 52, + 84 + ] + ], + [ + [ + 64, + 124 + ], + [ + 38, + 152 + ], + [ + 125, + 112 + ], + [ + 3, + 135 + ], + [ + 139, + 91 + ], + [ + 30, + 112 + ] + ], + [ + [ + 116, + 98 + ], + [ + 120, + 43 + ], + [ + 66, + 72 + ], + [ + 168, + 125 + ], + [ + 66, + 121 + ], + [ + 160, + 171 + ] + ], + [ + [ + 70, + 73 + ], + [ + 19, + 78 + ], + [ + 92, + 20 + ], + [ + 45, + 132 + ], + [ + 146, + 12 + ], + [ + 104, + 128 + ] + ] + ] + }, + { + "task_id": "0001875", + "id": "RH20T_cfg4_task_0067_user_0007_scene_0004_cfg_0004_104122062823#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button from top to bottom with obstacles'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001875/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 262, + 63 + ], + [ + 247, + 86 + ], + [ + 288, + 85 + ], + [ + 236, + 41 + ], + [ + 300, + 65 + ], + [ + 251, + 20 + ] + ], + [ + [ + 262, + 58 + ], + [ + 259, + 25 + ], + [ + 281, + 74 + ], + [ + 245, + 43 + ], + [ + 283, + 103 + ], + [ + 248, + 71 + ] + ], + [ + [ + 245, + 71 + ], + [ + 220, + 84 + ], + [ + 232, + 37 + ], + [ + 256, + 103 + ], + [ + 256, + 27 + ], + [ + 278, + 90 + ] + ], + [ + [ + 295, + 70 + ], + [ + 298, + 97 + ], + [ + 301, + 94 + ], + [ + 287, + 43 + ], + [ + 298, + 70 + ], + [ + 284, + 19 + ] + ] + ] + }, + { + "task_id": "0001878", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_104122061850#237", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001878/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 394, + 140 + ], + [ + 412, + 151 + ], + [ + 398, + 120 + ], + [ + 391, + 160 + ], + [ + 381, + 110 + ], + [ + 375, + 150 + ] + ], + [ + [ + 413, + 156 + ], + [ + 394, + 149 + ], + [ + 428, + 141 + ], + [ + 400, + 169 + ], + [ + 448, + 149 + ], + [ + 419, + 176 + ] + ], + [ + [ + 420, + 138 + ], + [ + 431, + 137 + ], + [ + 441, + 127 + ], + [ + 400, + 149 + ], + [ + 430, + 128 + ], + [ + 391, + 149 + ] + ], + [ + [ + 400, + 158 + ], + [ + 399, + 144 + ], + [ + 392, + 178 + ], + [ + 409, + 138 + ], + [ + 394, + 190 + ], + [ + 409, + 151 + ] + ] + ] + }, + { + "task_id": "0001879", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_104122061850#340", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001879/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 367, + 181 + ], + [ + 390, + 172 + ], + [ + 357, + 163 + ], + [ + 376, + 197 + ], + [ + 334, + 172 + ], + [ + 353, + 205 + ] + ], + [ + [ + 398, + 182 + ], + [ + 395, + 167 + ], + [ + 380, + 198 + ], + [ + 416, + 166 + ], + [ + 384, + 211 + ], + [ + 418, + 180 + ] + ], + [ + [ + 403, + 209 + ], + [ + 413, + 190 + ], + [ + 426, + 216 + ], + [ + 380, + 201 + ], + [ + 416, + 233 + ], + [ + 371, + 219 + ] + ], + [ + [ + 403, + 203 + ], + [ + 402, + 220 + ], + [ + 426, + 214 + ], + [ + 379, + 193 + ], + [ + 425, + 198 + ], + [ + 381, + 177 + ] + ] + ] + }, + { + "task_id": "0001880", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_104122061850#499", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001880/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 430, + 145 + ], + [ + 426, + 168 + ], + [ + 414, + 140 + ], + [ + 448, + 149 + ], + [ + 419, + 117 + ], + [ + 453, + 125 + ] + ], + [ + [ + 427, + 161 + ], + [ + 426, + 147 + ], + [ + 406, + 174 + ], + [ + 448, + 149 + ], + [ + 408, + 187 + ], + [ + 448, + 162 + ] + ], + [ + [ + 413, + 176 + ], + [ + 400, + 185 + ], + [ + 392, + 161 + ], + [ + 434, + 191 + ], + [ + 405, + 151 + ], + [ + 449, + 182 + ] + ], + [ + [ + 440, + 159 + ], + [ + 442, + 165 + ], + [ + 427, + 180 + ], + [ + 454, + 137 + ], + [ + 426, + 174 + ], + [ + 452, + 133 + ] + ] + ] + }, + { + "task_id": "0001887", + "id": "RH20T_cfg4_task_0003_user_0010_scene_0010_cfg_0004_036422060909#94", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001887/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 391, + 62 + ], + [ + 368, + 55 + ], + [ + 384, + 80 + ], + [ + 396, + 44 + ], + [ + 408, + 87 + ], + [ + 419, + 50 + ] + ], + [ + [ + 374, + 63 + ], + [ + 380, + 67 + ], + [ + 359, + 48 + ], + [ + 390, + 79 + ], + [ + 352, + 43 + ], + [ + 384, + 75 + ] + ], + [ + [ + 370, + 58 + ], + [ + 376, + 40 + ], + [ + 377, + 42 + ], + [ + 363, + 75 + ], + [ + 372, + 59 + ], + [ + 358, + 91 + ] + ], + [ + [ + 358, + 60 + ], + [ + 380, + 68 + ], + [ + 366, + 62 + ], + [ + 350, + 58 + ], + [ + 345, + 55 + ], + [ + 328, + 50 + ] + ] + ] + }, + { + "task_id": "0001888", + "id": "RH20T_cfg4_task_0003_user_0010_scene_0010_cfg_0004_036422060909#141", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001888/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 317, + 79 + ], + [ + 343, + 83 + ], + [ + 322, + 78 + ], + [ + 313, + 81 + ], + [ + 294, + 74 + ], + [ + 288, + 77 + ] + ], + [ + [ + 329, + 93 + ], + [ + 352, + 77 + ], + [ + 320, + 83 + ], + [ + 338, + 103 + ], + [ + 297, + 99 + ], + [ + 317, + 117 + ] + ], + [ + [ + 311, + 77 + ], + [ + 313, + 55 + ], + [ + 331, + 64 + ], + [ + 291, + 91 + ], + [ + 328, + 84 + ], + [ + 290, + 111 + ] + ], + [ + [ + 313, + 106 + ], + [ + 326, + 83 + ], + [ + 336, + 113 + ], + [ + 288, + 98 + ], + [ + 324, + 134 + ], + [ + 277, + 121 + ] + ] + ] + }, + { + "task_id": "0001891", + "id": "RH20T_cfg1_task_0038_user_0015_scene_0010_cfg_0001_038522063145#145", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001891/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 107 + ], + [ + 418, + 100 + ], + [ + 389, + 93 + ], + [ + 424, + 121 + ], + [ + 375, + 100 + ], + [ + 411, + 129 + ] + ], + [ + [ + 404, + 104 + ], + [ + 397, + 120 + ], + [ + 404, + 122 + ], + [ + 403, + 85 + ], + [ + 411, + 108 + ], + [ + 410, + 70 + ] + ], + [ + [ + 443, + 80 + ], + [ + 454, + 87 + ], + [ + 431, + 64 + ], + [ + 456, + 97 + ], + [ + 419, + 56 + ], + [ + 445, + 91 + ] + ], + [ + [ + 420, + 103 + ], + [ + 427, + 85 + ], + [ + 443, + 111 + ], + [ + 398, + 95 + ], + [ + 436, + 128 + ], + [ + 393, + 112 + ] + ] + ] + }, + { + "task_id": "0001892", + "id": "RH20T_cfg1_task_0046_user_0001_scene_0003_cfg_0001_038522063145#211", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001892/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 515, + 147 + ], + [ + 524, + 168 + ], + [ + 519, + 151 + ], + [ + 511, + 143 + ], + [ + 509, + 129 + ], + [ + 503, + 123 + ] + ], + [ + [ + 541, + 135 + ], + [ + 547, + 115 + ], + [ + 552, + 124 + ], + [ + 530, + 146 + ], + [ + 547, + 142 + ], + [ + 525, + 165 + ] + ], + [ + [ + 541, + 155 + ], + [ + 515, + 151 + ], + [ + 543, + 151 + ], + [ + 539, + 160 + ], + [ + 568, + 155 + ], + [ + 567, + 165 + ] + ], + [ + [ + 539, + 128 + ], + [ + 515, + 136 + ], + [ + 532, + 112 + ], + [ + 546, + 145 + ], + [ + 556, + 103 + ], + [ + 572, + 137 + ] + ] + ] + }, + { + "task_id": "0001894", + "id": "RH20T_cfg7_task_0044_user_0014_scene_0005_cfg_0007_104122061018#89", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001894/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 294, + 44 + ], + [ + 301, + 23 + ], + [ + 281, + 20 + ], + [ + 309, + 68 + ], + [ + 276, + 40 + ], + [ + 302, + 86 + ] + ], + [ + [ + 328, + 42 + ], + [ + 310, + 59 + ], + [ + 336, + 68 + ], + [ + 319, + 13 + ], + [ + 351, + 53 + ], + [ + 336, + 0 + ] + ], + [ + [ + 271, + 57 + ], + [ + 286, + 49 + ], + [ + 295, + 44 + ], + [ + 249, + 69 + ], + [ + 279, + 52 + ], + [ + 232, + 77 + ] + ], + [ + [ + 290, + 78 + ], + [ + 300, + 80 + ], + [ + 304, + 105 + ], + [ + 276, + 51 + ], + [ + 294, + 104 + ], + [ + 265, + 47 + ] + ] + ] + }, + { + "task_id": "0001895", + "id": "RH20T_cfg6_task_0038_user_0020_scene_0001_cfg_0006_104122064161#428", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001895/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 354, + 175 + ], + [ + 355, + 171 + ], + [ + 370, + 192 + ], + [ + 339, + 157 + ], + [ + 370, + 197 + ], + [ + 337, + 161 + ] + ], + [ + [ + 379, + 151 + ], + [ + 380, + 165 + ], + [ + 376, + 169 + ], + [ + 381, + 131 + ], + [ + 376, + 155 + ], + [ + 380, + 118 + ] + ], + [ + [ + 375, + 167 + ], + [ + 392, + 181 + ], + [ + 380, + 171 + ], + [ + 368, + 164 + ], + [ + 364, + 158 + ], + [ + 351, + 150 + ] + ], + [ + [ + 365, + 128 + ], + [ + 354, + 115 + ], + [ + 385, + 127 + ], + [ + 347, + 129 + ], + [ + 395, + 140 + ], + [ + 358, + 141 + ] + ] + ] + }, + { + "task_id": "0001896", + "id": "RH20T_cfg3_task_0038_user_0010_scene_0003_cfg_0003_038522062288#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001896/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 295, + 147 + ], + [ + 286, + 173 + ], + [ + 321, + 160 + ], + [ + 268, + 133 + ], + [ + 329, + 136 + ], + [ + 278, + 108 + ] + ], + [ + [ + 287, + 141 + ], + [ + 296, + 160 + ], + [ + 314, + 135 + ], + [ + 260, + 148 + ], + [ + 305, + 118 + ], + [ + 253, + 130 + ] + ], + [ + [ + 299, + 115 + ], + [ + 284, + 133 + ], + [ + 298, + 138 + ], + [ + 301, + 90 + ], + [ + 312, + 122 + ], + [ + 315, + 74 + ] + ], + [ + [ + 273, + 138 + ], + [ + 272, + 118 + ], + [ + 252, + 123 + ], + [ + 295, + 154 + ], + [ + 255, + 142 + ], + [ + 296, + 172 + ] + ] + ] + }, + { + "task_id": "0001899", + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_105422061350#341", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001899/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 191 + ], + [ + 278, + 196 + ], + [ + 311, + 211 + ], + [ + 300, + 170 + ], + [ + 338, + 207 + ], + [ + 328, + 166 + ] + ], + [ + [ + 303, + 206 + ], + [ + 306, + 190 + ], + [ + 327, + 195 + ], + [ + 278, + 218 + ], + [ + 324, + 209 + ], + [ + 277, + 232 + ] + ], + [ + [ + 316, + 211 + ], + [ + 337, + 223 + ], + [ + 309, + 237 + ], + [ + 323, + 186 + ], + [ + 286, + 225 + ], + [ + 301, + 173 + ] + ], + [ + [ + 289, + 192 + ], + [ + 269, + 179 + ], + [ + 275, + 183 + ], + [ + 305, + 202 + ], + [ + 294, + 195 + ], + [ + 324, + 214 + ] + ] + ] + }, + { + "task_id": "0001900", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061330#226", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001900/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 308, + 168 + ], + [ + 296, + 148 + ], + [ + 299, + 182 + ], + [ + 317, + 152 + ], + [ + 311, + 202 + ], + [ + 330, + 172 + ] + ], + [ + [ + 276, + 188 + ], + [ + 259, + 168 + ], + [ + 257, + 206 + ], + [ + 295, + 172 + ], + [ + 274, + 226 + ], + [ + 311, + 191 + ] + ], + [ + [ + 285, + 181 + ], + [ + 270, + 188 + ], + [ + 281, + 205 + ], + [ + 288, + 157 + ], + [ + 294, + 197 + ], + [ + 302, + 151 + ] + ], + [ + [ + 296, + 166 + ], + [ + 296, + 140 + ], + [ + 275, + 164 + ], + [ + 318, + 168 + ], + [ + 275, + 189 + ], + [ + 318, + 194 + ] + ] + ] + }, + { + "task_id": "0001901", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061330#564", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001901/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 325, + 118 + ], + [ + 345, + 109 + ], + [ + 334, + 137 + ], + [ + 314, + 98 + ], + [ + 314, + 146 + ], + [ + 294, + 107 + ] + ], + [ + [ + 334, + 84 + ], + [ + 321, + 66 + ], + [ + 325, + 76 + ], + [ + 342, + 91 + ], + [ + 338, + 96 + ], + [ + 354, + 109 + ] + ], + [ + [ + 338, + 107 + ], + [ + 330, + 101 + ], + [ + 330, + 130 + ], + [ + 346, + 84 + ], + [ + 338, + 135 + ], + [ + 353, + 91 + ] + ], + [ + [ + 316, + 88 + ], + [ + 294, + 81 + ], + [ + 323, + 64 + ], + [ + 309, + 112 + ], + [ + 346, + 72 + ], + [ + 331, + 120 + ] + ] + ] + }, + { + "task_id": "0001902", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061330#829", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001902/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 345, + 120 + ], + [ + 326, + 124 + ], + [ + 359, + 135 + ], + [ + 330, + 105 + ], + [ + 379, + 132 + ], + [ + 351, + 101 + ] + ], + [ + [ + 360, + 164 + ], + [ + 337, + 165 + ], + [ + 368, + 158 + ], + [ + 353, + 170 + ], + [ + 391, + 156 + ], + [ + 378, + 169 + ] + ], + [ + [ + 316, + 159 + ], + [ + 312, + 184 + ], + [ + 341, + 162 + ], + [ + 291, + 155 + ], + [ + 345, + 137 + ], + [ + 295, + 130 + ] + ], + [ + [ + 351, + 143 + ], + [ + 339, + 151 + ], + [ + 348, + 166 + ], + [ + 353, + 119 + ], + [ + 359, + 157 + ], + [ + 364, + 112 + ] + ] + ] + }, + { + "task_id": "0001905", + "id": "RH20T_cfg6_task_0104_user_0020_scene_0006_cfg_0006_104122061018#209", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate back after holding it down'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001905/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 354, + 247 + ], + [ + 337, + 217 + ], + [ + 354, + 248 + ], + [ + 353, + 245 + ], + [ + 370, + 277 + ], + [ + 372, + 278 + ] + ], + [ + [ + 376, + 245 + ], + [ + 358, + 240 + ], + [ + 402, + 226 + ], + [ + 349, + 264 + ], + [ + 424, + 230 + ], + [ + 369, + 272 + ] + ], + [ + [ + 369, + 215 + ], + [ + 357, + 243 + ], + [ + 350, + 217 + ], + [ + 392, + 213 + ], + [ + 361, + 192 + ], + [ + 403, + 185 + ] + ], + [ + [ + 355, + 244 + ], + [ + 324, + 260 + ], + [ + 366, + 245 + ], + [ + 345, + 244 + ], + [ + 399, + 229 + ], + [ + 374, + 230 + ] + ] + ] + }, + { + "task_id": "0001906", + "id": "RH20T_cfg4_task_0091_user_0014_scene_0007_cfg_0004_104122062295#114", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001906/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 331, + 84 + ], + [ + 297, + 88 + ], + [ + 334, + 116 + ], + [ + 328, + 54 + ], + [ + 370, + 111 + ], + [ + 362, + 49 + ] + ], + [ + [ + 299, + 89 + ], + [ + 303, + 58 + ], + [ + 264, + 84 + ], + [ + 333, + 93 + ], + [ + 262, + 113 + ], + [ + 327, + 120 + ] + ], + [ + [ + 321, + 108 + ], + [ + 306, + 84 + ], + [ + 316, + 94 + ], + [ + 328, + 125 + ], + [ + 329, + 116 + ], + [ + 342, + 148 + ] + ], + [ + [ + 276, + 76 + ], + [ + 289, + 100 + ], + [ + 311, + 77 + ], + [ + 244, + 75 + ], + [ + 299, + 48 + ], + [ + 228, + 48 + ] + ] + ] + }, + { + "task_id": "0001907", + "id": "RH20T_cfg3_task_0001_user_0016_scene_0007_cfg_0003_104122062295#219", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001907/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 258, + 282 + ], + [ + 267, + 245 + ], + [ + 230, + 249 + ], + [ + 290, + 320 + ], + [ + 225, + 281 + ], + [ + 280, + 351 + ] + ], + [ + [ + 279, + 334 + ], + [ + 289, + 283 + ], + [ + 328, + 344 + ], + [ + 230, + 324 + ], + [ + 319, + 360 + ], + [ + 218, + 360 + ] + ], + [ + [ + 265, + 324 + ], + [ + 290, + 301 + ], + [ + 237, + 276 + ], + [ + 294, + 360 + ], + [ + 203, + 299 + ], + [ + 266, + 360 + ] + ], + [ + [ + 236, + 274 + ], + [ + 265, + 253 + ], + [ + 248, + 240 + ], + [ + 226, + 306 + ], + [ + 214, + 263 + ], + [ + 194, + 332 + ] + ] + ] + }, + { + "task_id": "0001911", + "id": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122064161#392", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001911/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 362, + 192 + ], + [ + 367, + 175 + ], + [ + 363, + 211 + ], + [ + 361, + 174 + ], + [ + 357, + 226 + ], + [ + 357, + 190 + ] + ], + [ + [ + 386, + 204 + ], + [ + 411, + 209 + ], + [ + 390, + 199 + ], + [ + 382, + 209 + ], + [ + 367, + 194 + ], + [ + 356, + 204 + ] + ], + [ + [ + 353, + 195 + ], + [ + 328, + 201 + ], + [ + 354, + 213 + ], + [ + 351, + 179 + ], + [ + 380, + 207 + ], + [ + 375, + 173 + ] + ], + [ + [ + 383, + 174 + ], + [ + 396, + 192 + ], + [ + 373, + 188 + ], + [ + 392, + 161 + ], + [ + 360, + 168 + ], + [ + 379, + 143 + ] + ] + ] + }, + { + "task_id": "0001914", + "id": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122064161#742", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001914/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 330, + 177 + ], + [ + 338, + 158 + ], + [ + 332, + 192 + ], + [ + 329, + 162 + ], + [ + 325, + 210 + ], + [ + 322, + 180 + ] + ], + [ + [ + 325, + 187 + ], + [ + 342, + 196 + ], + [ + 332, + 167 + ], + [ + 317, + 207 + ], + [ + 315, + 156 + ], + [ + 299, + 198 + ] + ], + [ + [ + 340, + 153 + ], + [ + 320, + 143 + ], + [ + 348, + 132 + ], + [ + 333, + 175 + ], + [ + 368, + 142 + ], + [ + 354, + 184 + ] + ], + [ + [ + 327, + 167 + ], + [ + 313, + 172 + ], + [ + 342, + 184 + ], + [ + 313, + 150 + ], + [ + 355, + 178 + ], + [ + 326, + 146 + ] + ] + ] + }, + { + "task_id": "0001916", + "id": "RH20T_cfg7_task_0036_user_0020_scene_0001_cfg_0007_104122061602#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Shake the test tube'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001916/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 489, + 136 + ], + [ + 497, + 111 + ], + [ + 480, + 142 + ], + [ + 499, + 128 + ], + [ + 472, + 166 + ], + [ + 490, + 154 + ] + ], + [ + [ + 471, + 165 + ], + [ + 484, + 188 + ], + [ + 456, + 177 + ], + [ + 486, + 151 + ], + [ + 445, + 155 + ], + [ + 472, + 127 + ] + ], + [ + [ + 463, + 184 + ], + [ + 457, + 209 + ], + [ + 442, + 185 + ], + [ + 487, + 183 + ], + [ + 447, + 162 + ], + [ + 492, + 158 + ] + ], + [ + [ + 474, + 187 + ], + [ + 479, + 159 + ], + [ + 493, + 190 + ], + [ + 453, + 184 + ], + [ + 487, + 217 + ], + [ + 447, + 213 + ] + ] + ] + }, + { + "task_id": "0001919", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0007_cfg_0006_104122061602#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001919/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 451, + 213 + ], + [ + 434, + 231 + ], + [ + 430, + 190 + ], + [ + 472, + 235 + ], + [ + 447, + 174 + ], + [ + 488, + 217 + ] + ], + [ + [ + 450, + 180 + ], + [ + 471, + 183 + ], + [ + 454, + 208 + ], + [ + 445, + 152 + ], + [ + 435, + 204 + ], + [ + 427, + 152 + ] + ], + [ + [ + 432, + 206 + ], + [ + 420, + 221 + ], + [ + 408, + 205 + ], + [ + 459, + 206 + ], + [ + 420, + 191 + ], + [ + 470, + 191 + ] + ], + [ + [ + 454, + 187 + ], + [ + 430, + 192 + ], + [ + 474, + 187 + ], + [ + 435, + 187 + ], + [ + 497, + 183 + ], + [ + 457, + 183 + ] + ] + ] + }, + { + "task_id": "0001923", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122064161#701", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001923/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 391, + 122 + ], + [ + 393, + 127 + ], + [ + 388, + 142 + ], + [ + 394, + 101 + ], + [ + 386, + 136 + ], + [ + 392, + 97 + ] + ], + [ + [ + 385, + 140 + ], + [ + 367, + 130 + ], + [ + 375, + 159 + ], + [ + 396, + 121 + ], + [ + 394, + 169 + ], + [ + 415, + 131 + ] + ], + [ + [ + 374, + 136 + ], + [ + 359, + 125 + ], + [ + 360, + 142 + ], + [ + 388, + 130 + ], + [ + 374, + 153 + ], + [ + 403, + 142 + ] + ], + [ + [ + 388, + 141 + ], + [ + 379, + 152 + ], + [ + 407, + 144 + ], + [ + 369, + 138 + ], + [ + 415, + 134 + ], + [ + 378, + 128 + ] + ] + ] + }, + { + "task_id": "0001924", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122064161#929", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001924/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 362, + 96 + ], + [ + 347, + 89 + ], + [ + 351, + 106 + ], + [ + 374, + 85 + ], + [ + 366, + 113 + ], + [ + 389, + 92 + ] + ], + [ + [ + 361, + 97 + ], + [ + 364, + 85 + ], + [ + 361, + 114 + ], + [ + 361, + 81 + ], + [ + 358, + 124 + ], + [ + 359, + 92 + ] + ], + [ + [ + 376, + 94 + ], + [ + 386, + 81 + ], + [ + 391, + 106 + ], + [ + 361, + 82 + ], + [ + 380, + 120 + ], + [ + 350, + 95 + ] + ], + [ + [ + 346, + 87 + ], + [ + 361, + 82 + ], + [ + 350, + 107 + ], + [ + 342, + 66 + ], + [ + 335, + 112 + ], + [ + 326, + 71 + ] + ] + ] + }, + { + "task_id": "0001925", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122064161#1129", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001925/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 333, + 132 + ], + [ + 333, + 120 + ], + [ + 332, + 151 + ], + [ + 333, + 114 + ], + [ + 332, + 161 + ], + [ + 332, + 125 + ] + ], + [ + [ + 317, + 144 + ], + [ + 330, + 154 + ], + [ + 331, + 149 + ], + [ + 304, + 140 + ], + [ + 317, + 139 + ], + [ + 290, + 130 + ] + ], + [ + [ + 349, + 134 + ], + [ + 359, + 151 + ], + [ + 343, + 143 + ], + [ + 354, + 125 + ], + [ + 333, + 125 + ], + [ + 345, + 108 + ] + ], + [ + [ + 333, + 150 + ], + [ + 344, + 141 + ], + [ + 352, + 155 + ], + [ + 314, + 145 + ], + [ + 342, + 162 + ], + [ + 305, + 153 + ] + ] + ] + }, + { + "task_id": "0001927", + "id": "RH20T_cfg1_task_0120_user_0014_scene_0002_cfg_0001_035622060973#289", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001927/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 230, + 187 + ], + [ + 229, + 187 + ], + [ + 234, + 165 + ], + [ + 225, + 209 + ], + [ + 235, + 164 + ], + [ + 225, + 211 + ] + ], + [ + [ + 238, + 198 + ], + [ + 247, + 213 + ], + [ + 259, + 195 + ], + [ + 217, + 201 + ], + [ + 250, + 181 + ], + [ + 209, + 186 + ] + ], + [ + [ + 234, + 172 + ], + [ + 231, + 160 + ], + [ + 216, + 184 + ], + [ + 253, + 160 + ], + [ + 219, + 195 + ], + [ + 255, + 172 + ] + ], + [ + [ + 241, + 179 + ], + [ + 220, + 173 + ], + [ + 242, + 195 + ], + [ + 241, + 164 + ], + [ + 263, + 200 + ], + [ + 261, + 170 + ] + ] + ] + }, + { + "task_id": "0001932", + "id": "RH20T_cfg1_task_0089_user_0014_scene_0002_cfg_0001_819112070044#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push an object with obstacles'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001932/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 287, + 206 + ], + [ + 286, + 190 + ], + [ + 304, + 197 + ], + [ + 269, + 215 + ], + [ + 303, + 212 + ], + [ + 270, + 229 + ] + ], + [ + [ + 295, + 179 + ], + [ + 310, + 185 + ], + [ + 293, + 199 + ], + [ + 297, + 159 + ], + [ + 279, + 194 + ], + [ + 283, + 155 + ] + ], + [ + [ + 292, + 193 + ], + [ + 285, + 175 + ], + [ + 307, + 190 + ], + [ + 276, + 196 + ], + [ + 314, + 208 + ], + [ + 283, + 214 + ] + ], + [ + [ + 258, + 195 + ], + [ + 268, + 205 + ], + [ + 255, + 177 + ], + [ + 262, + 212 + ], + [ + 245, + 168 + ], + [ + 253, + 202 + ] + ] + ] + }, + { + "task_id": "0001933", + "id": "RH20T_cfg4_task_0005_user_0010_scene_0004_cfg_0004_038522062288#103", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001933/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 160, + 167 + ], + [ + 154, + 147 + ], + [ + 139, + 151 + ], + [ + 181, + 183 + ], + [ + 145, + 169 + ], + [ + 185, + 200 + ] + ], + [ + [ + 148, + 172 + ], + [ + 124, + 191 + ], + [ + 130, + 163 + ], + [ + 167, + 182 + ], + [ + 153, + 146 + ], + [ + 190, + 163 + ] + ], + [ + [ + 124, + 175 + ], + [ + 139, + 166 + ], + [ + 102, + 151 + ], + [ + 145, + 198 + ], + [ + 88, + 160 + ], + [ + 130, + 205 + ] + ], + [ + [ + 170, + 168 + ], + [ + 164, + 143 + ], + [ + 198, + 163 + ], + [ + 140, + 174 + ], + [ + 205, + 188 + ], + [ + 146, + 201 + ] + ] + ] + }, + { + "task_id": "0001935", + "id": "RH20T_cfg6_task_0129_user_0020_scene_0009_cfg_0006_104122060811#208", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001935/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 137 + ], + [ + 301, + 143 + ], + [ + 318, + 148 + ], + [ + 299, + 125 + ], + [ + 325, + 142 + ], + [ + 307, + 119 + ] + ], + [ + [ + 319, + 139 + ], + [ + 316, + 129 + ], + [ + 329, + 130 + ], + [ + 308, + 147 + ], + [ + 332, + 140 + ], + [ + 311, + 158 + ] + ], + [ + [ + 316, + 137 + ], + [ + 328, + 129 + ], + [ + 324, + 144 + ], + [ + 307, + 130 + ], + [ + 312, + 152 + ], + [ + 296, + 138 + ] + ], + [ + [ + 336, + 142 + ], + [ + 323, + 145 + ], + [ + 329, + 140 + ], + [ + 343, + 144 + ], + [ + 341, + 137 + ], + [ + 356, + 140 + ] + ] + ] + }, + { + "task_id": "0001939", + "id": "RH20T_cfg5_task_0075_user_0007_scene_0001_cfg_0005_104122063678#1032", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Finish setting up the starting position of a chessboard that is almost arranged'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001939/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 396, + 296 + ], + [ + 423, + 298 + ], + [ + 396, + 321 + ], + [ + 395, + 274 + ], + [ + 368, + 319 + ], + [ + 369, + 272 + ] + ], + [ + [ + 395, + 282 + ], + [ + 377, + 300 + ], + [ + 378, + 271 + ], + [ + 414, + 294 + ], + [ + 395, + 255 + ], + [ + 432, + 276 + ] + ], + [ + [ + 375, + 272 + ], + [ + 373, + 246 + ], + [ + 368, + 272 + ], + [ + 382, + 272 + ], + [ + 370, + 297 + ], + [ + 384, + 300 + ] + ], + [ + [ + 388, + 270 + ], + [ + 394, + 257 + ], + [ + 384, + 294 + ], + [ + 392, + 248 + ], + [ + 379, + 305 + ], + [ + 387, + 261 + ] + ] + ] + }, + { + "task_id": "0001940", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_037522061512#297", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001940/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 264, + 108 + ], + [ + 269, + 127 + ], + [ + 254, + 103 + ], + [ + 274, + 112 + ], + [ + 249, + 84 + ], + [ + 269, + 92 + ] + ], + [ + [ + 269, + 109 + ], + [ + 267, + 87 + ], + [ + 249, + 111 + ], + [ + 289, + 108 + ], + [ + 251, + 133 + ], + [ + 290, + 131 + ] + ], + [ + [ + 284, + 84 + ], + [ + 305, + 82 + ], + [ + 284, + 70 + ], + [ + 284, + 100 + ], + [ + 264, + 73 + ], + [ + 263, + 103 + ] + ], + [ + [ + 270, + 107 + ], + [ + 268, + 89 + ], + [ + 282, + 96 + ], + [ + 258, + 119 + ], + [ + 283, + 113 + ], + [ + 261, + 136 + ] + ] + ] + }, + { + "task_id": "0001941", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_037522061512#736", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001941/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 236, + 164 + ], + [ + 255, + 177 + ], + [ + 225, + 187 + ], + [ + 248, + 142 + ], + [ + 204, + 175 + ], + [ + 228, + 129 + ] + ], + [ + [ + 253, + 177 + ], + [ + 253, + 156 + ], + [ + 267, + 167 + ], + [ + 237, + 188 + ], + [ + 266, + 186 + ], + [ + 237, + 208 + ] + ], + [ + [ + 229, + 179 + ], + [ + 252, + 170 + ], + [ + 220, + 172 + ], + [ + 238, + 186 + ], + [ + 196, + 181 + ], + [ + 213, + 195 + ] + ], + [ + [ + 276, + 193 + ], + [ + 256, + 206 + ], + [ + 279, + 207 + ], + [ + 272, + 179 + ], + [ + 297, + 194 + ], + [ + 291, + 166 + ] + ] + ] + }, + { + "task_id": "0001942", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_037522061512#1128", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001942/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 222, + 142 + ], + [ + 239, + 138 + ], + [ + 205, + 130 + ], + [ + 238, + 152 + ], + [ + 189, + 133 + ], + [ + 223, + 155 + ] + ], + [ + [ + 220, + 148 + ], + [ + 240, + 156 + ], + [ + 207, + 168 + ], + [ + 232, + 128 + ], + [ + 188, + 160 + ], + [ + 213, + 121 + ] + ], + [ + [ + 218, + 143 + ], + [ + 212, + 126 + ], + [ + 231, + 128 + ], + [ + 204, + 158 + ], + [ + 235, + 143 + ], + [ + 210, + 173 + ] + ], + [ + [ + 207, + 126 + ], + [ + 194, + 112 + ], + [ + 210, + 109 + ], + [ + 204, + 144 + ], + [ + 220, + 123 + ], + [ + 215, + 157 + ] + ] + ] + }, + { + "task_id": "0001944", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_104422070044#260", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001944/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 372, + 193 + ], + [ + 378, + 177 + ], + [ + 360, + 185 + ], + [ + 384, + 200 + ], + [ + 353, + 201 + ], + [ + 378, + 216 + ] + ], + [ + [ + 373, + 168 + ], + [ + 364, + 152 + ], + [ + 388, + 159 + ], + [ + 358, + 178 + ], + [ + 396, + 175 + ], + [ + 366, + 194 + ] + ], + [ + [ + 391, + 194 + ], + [ + 409, + 193 + ], + [ + 393, + 183 + ], + [ + 389, + 206 + ], + [ + 376, + 184 + ], + [ + 371, + 207 + ] + ], + [ + [ + 368, + 190 + ], + [ + 366, + 178 + ], + [ + 359, + 203 + ], + [ + 377, + 177 + ], + [ + 360, + 214 + ], + [ + 378, + 188 + ] + ] + ] + }, + { + "task_id": "0001945", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_104422070044#298", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001945/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 182 + ], + [ + 396, + 167 + ], + [ + 379, + 188 + ], + [ + 407, + 175 + ], + [ + 375, + 203 + ], + [ + 403, + 191 + ] + ], + [ + [ + 381, + 216 + ], + [ + 367, + 210 + ], + [ + 389, + 199 + ], + [ + 374, + 233 + ], + [ + 404, + 206 + ], + [ + 388, + 240 + ] + ], + [ + [ + 374, + 181 + ], + [ + 392, + 186 + ], + [ + 380, + 164 + ], + [ + 369, + 198 + ], + [ + 362, + 159 + ], + [ + 351, + 193 + ] + ], + [ + [ + 381, + 189 + ], + [ + 380, + 177 + ], + [ + 369, + 201 + ], + [ + 393, + 178 + ], + [ + 371, + 212 + ], + [ + 393, + 190 + ] + ] + ] + }, + { + "task_id": "0001946", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_104422070044#345", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001946/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 391, + 198 + ], + [ + 390, + 186 + ], + [ + 378, + 209 + ], + [ + 403, + 186 + ], + [ + 379, + 221 + ], + [ + 403, + 198 + ] + ], + [ + [ + 370, + 192 + ], + [ + 357, + 185 + ], + [ + 382, + 178 + ], + [ + 357, + 207 + ], + [ + 396, + 185 + ], + [ + 371, + 215 + ] + ], + [ + [ + 385, + 211 + ], + [ + 381, + 219 + ], + [ + 389, + 227 + ], + [ + 380, + 194 + ], + [ + 393, + 220 + ], + [ + 384, + 187 + ] + ], + [ + [ + 390, + 195 + ], + [ + 372, + 188 + ], + [ + 394, + 178 + ], + [ + 385, + 211 + ], + [ + 413, + 185 + ], + [ + 403, + 218 + ] + ] + ] + }, + { + "task_id": "0001947", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_104422070044#368", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001947/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 402, + 207 + ], + [ + 401, + 196 + ], + [ + 389, + 220 + ], + [ + 414, + 195 + ], + [ + 390, + 230 + ], + [ + 414, + 206 + ] + ], + [ + [ + 393, + 201 + ], + [ + 402, + 203 + ], + [ + 376, + 197 + ], + [ + 410, + 206 + ], + [ + 369, + 196 + ], + [ + 402, + 204 + ] + ], + [ + [ + 408, + 195 + ], + [ + 398, + 179 + ], + [ + 391, + 204 + ], + [ + 424, + 185 + ], + [ + 401, + 221 + ], + [ + 435, + 201 + ] + ], + [ + [ + 414, + 209 + ], + [ + 416, + 191 + ], + [ + 426, + 211 + ], + [ + 402, + 207 + ], + [ + 424, + 230 + ], + [ + 400, + 227 + ] + ] + ] + }, + { + "task_id": "0001950", + "id": "RH20T_cfg3_task_0008_user_0016_scene_0009_cfg_0003_104122062823#144", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001950/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 214, + 38 + ], + [ + 223, + 48 + ], + [ + 201, + 64 + ], + [ + 226, + 11 + ], + [ + 190, + 55 + ], + [ + 217, + 0 + ] + ], + [ + [ + 227, + 43 + ], + [ + 223, + 19 + ], + [ + 249, + 53 + ], + [ + 206, + 34 + ], + [ + 252, + 76 + ], + [ + 211, + 56 + ] + ], + [ + [ + 225, + 38 + ], + [ + 220, + 61 + ], + [ + 220, + 50 + ], + [ + 231, + 24 + ], + [ + 225, + 28 + ], + [ + 237, + 1 + ] + ], + [ + [ + 217, + 24 + ], + [ + 222, + 40 + ], + [ + 192, + 40 + ], + [ + 240, + 9 + ], + [ + 185, + 23 + ], + [ + 237, + 0 + ] + ] + ] + }, + { + "task_id": "0001952", + "id": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_104122062823#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001952/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 206, + 42 + ], + [ + 208, + 42 + ], + [ + 179, + 30 + ], + [ + 232, + 55 + ], + [ + 176, + 29 + ], + [ + 232, + 55 + ] + ], + [ + [ + 213, + 50 + ], + [ + 206, + 71 + ], + [ + 203, + 32 + ], + [ + 223, + 67 + ], + [ + 210, + 11 + ], + [ + 229, + 47 + ] + ], + [ + [ + 226, + 42 + ], + [ + 223, + 16 + ], + [ + 248, + 51 + ], + [ + 205, + 34 + ], + [ + 250, + 75 + ], + [ + 208, + 57 + ] + ], + [ + [ + 232, + 22 + ], + [ + 217, + 6 + ], + [ + 239, + 41 + ], + [ + 224, + 1 + ], + [ + 254, + 58 + ], + [ + 241, + 20 + ] + ] + ] + }, + { + "task_id": "0001955", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_819112071037#117", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001955/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 233, + 96 + ], + [ + 272, + 108 + ], + [ + 210, + 150 + ], + [ + 258, + 38 + ], + [ + 160, + 142 + ], + [ + 211, + 15 + ] + ], + [ + [ + 213, + 80 + ], + [ + 207, + 8 + ], + [ + 141, + 82 + ], + [ + 281, + 78 + ], + [ + 149, + 149 + ], + [ + 283, + 143 + ] + ], + [ + [ + 254, + 27 + ], + [ + 275, + 38 + ], + [ + 213, + 76 + ], + [ + 294, + 0 + ], + [ + 200, + 62 + ], + [ + 273, + 0 + ] + ], + [ + [ + 184, + 82 + ], + [ + 186, + 65 + ], + [ + 235, + 33 + ], + [ + 135, + 128 + ], + [ + 239, + 48 + ], + [ + 125, + 156 + ] + ] + ] + }, + { + "task_id": "0001957", + "id": "RH20T_cfg3_task_0071_user_0010_scene_0008_cfg_0003_104122062295#155", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001957/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 217 + ], + [ + 335, + 189 + ], + [ + 315, + 212 + ], + [ + 378, + 221 + ], + [ + 328, + 244 + ], + [ + 393, + 251 + ] + ], + [ + [ + 329, + 220 + ], + [ + 340, + 187 + ], + [ + 305, + 202 + ], + [ + 356, + 240 + ], + [ + 296, + 232 + ], + [ + 344, + 272 + ] + ], + [ + [ + 342, + 235 + ], + [ + 346, + 199 + ], + [ + 321, + 229 + ], + [ + 366, + 241 + ], + [ + 317, + 263 + ], + [ + 361, + 279 + ] + ], + [ + [ + 302, + 202 + ], + [ + 276, + 177 + ], + [ + 299, + 221 + ], + [ + 306, + 180 + ], + [ + 324, + 246 + ], + [ + 336, + 208 + ] + ] + ] + }, + { + "task_id": "0001959", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122061018#760", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001959/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 214, + 36 + ], + [ + 203, + 54 + ], + [ + 200, + 51 + ], + [ + 229, + 21 + ], + [ + 210, + 35 + ], + [ + 240, + 5 + ] + ], + [ + [ + 209, + 57 + ], + [ + 205, + 29 + ], + [ + 202, + 48 + ], + [ + 217, + 66 + ], + [ + 206, + 73 + ], + [ + 221, + 93 + ] + ], + [ + [ + 215, + 30 + ], + [ + 190, + 27 + ], + [ + 208, + 3 + ], + [ + 222, + 54 + ], + [ + 236, + 6 + ], + [ + 247, + 58 + ] + ], + [ + [ + 220, + 40 + ], + [ + 219, + 13 + ], + [ + 190, + 37 + ], + [ + 249, + 42 + ], + [ + 190, + 65 + ], + [ + 250, + 69 + ] + ] + ] + }, + { + "task_id": "0001963", + "id": "RH20T_cfg5_task_0119_user_0007_scene_0002_cfg_0005_104422070044#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn off the water tap'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001963/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 127 + ], + [ + 333, + 120 + ], + [ + 322, + 142 + ], + [ + 315, + 110 + ], + [ + 308, + 149 + ], + [ + 300, + 117 + ] + ], + [ + [ + 329, + 137 + ], + [ + 326, + 126 + ], + [ + 336, + 148 + ], + [ + 322, + 126 + ], + [ + 338, + 158 + ], + [ + 325, + 136 + ] + ], + [ + [ + 326, + 132 + ], + [ + 314, + 137 + ], + [ + 338, + 141 + ], + [ + 314, + 123 + ], + [ + 351, + 136 + ], + [ + 327, + 117 + ] + ], + [ + [ + 321, + 132 + ], + [ + 310, + 142 + ], + [ + 327, + 143 + ], + [ + 316, + 121 + ], + [ + 338, + 132 + ], + [ + 327, + 110 + ] + ] + ] + }, + { + "task_id": "0001965", + "id": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_104122061602#236", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001965/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 500, + 219 + ], + [ + 518, + 207 + ], + [ + 521, + 235 + ], + [ + 480, + 203 + ], + [ + 504, + 245 + ], + [ + 465, + 215 + ] + ], + [ + [ + 492, + 205 + ], + [ + 512, + 216 + ], + [ + 490, + 226 + ], + [ + 494, + 182 + ], + [ + 472, + 215 + ], + [ + 475, + 172 + ] + ], + [ + [ + 512, + 223 + ], + [ + 489, + 232 + ], + [ + 530, + 232 + ], + [ + 496, + 215 + ], + [ + 554, + 222 + ], + [ + 518, + 206 + ] + ], + [ + [ + 526, + 223 + ], + [ + 502, + 214 + ], + [ + 542, + 204 + ], + [ + 511, + 242 + ], + [ + 566, + 213 + ], + [ + 534, + 251 + ] + ] + ] + }, + { + "task_id": "0001966", + "id": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_104122061602#394", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001966/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 459, + 139 + ], + [ + 474, + 152 + ], + [ + 454, + 157 + ], + [ + 464, + 120 + ], + [ + 441, + 145 + ], + [ + 450, + 109 + ] + ], + [ + [ + 442, + 138 + ], + [ + 439, + 142 + ], + [ + 445, + 160 + ], + [ + 440, + 116 + ], + [ + 449, + 156 + ], + [ + 443, + 109 + ] + ], + [ + [ + 445, + 152 + ], + [ + 459, + 144 + ], + [ + 450, + 173 + ], + [ + 440, + 131 + ], + [ + 437, + 182 + ], + [ + 426, + 139 + ] + ], + [ + [ + 458, + 150 + ], + [ + 482, + 156 + ], + [ + 465, + 129 + ], + [ + 452, + 171 + ], + [ + 442, + 123 + ], + [ + 430, + 166 + ] + ] + ] + }, + { + "task_id": "0001967", + "id": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_104122061602#548", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001967/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 394, + 124 + ], + [ + 386, + 108 + ], + [ + 379, + 117 + ], + [ + 408, + 131 + ], + [ + 387, + 134 + ], + [ + 416, + 148 + ] + ], + [ + [ + 410, + 132 + ], + [ + 421, + 126 + ], + [ + 418, + 153 + ], + [ + 402, + 112 + ], + [ + 408, + 158 + ], + [ + 393, + 119 + ] + ], + [ + [ + 430, + 120 + ], + [ + 419, + 107 + ], + [ + 420, + 139 + ], + [ + 440, + 102 + ], + [ + 431, + 151 + ], + [ + 450, + 114 + ] + ], + [ + [ + 423, + 106 + ], + [ + 425, + 128 + ], + [ + 415, + 109 + ], + [ + 430, + 102 + ], + [ + 412, + 86 + ], + [ + 428, + 81 + ] + ] + ] + }, + { + "task_id": "0001968", + "id": "RH20T_cfg4_task_0058_user_0014_scene_0003_cfg_0004_036422060909#120", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001968/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 360, + 115 + ], + [ + 387, + 119 + ], + [ + 356, + 143 + ], + [ + 364, + 88 + ], + [ + 328, + 139 + ], + [ + 336, + 83 + ] + ], + [ + [ + 357, + 111 + ], + [ + 335, + 98 + ], + [ + 358, + 112 + ], + [ + 357, + 111 + ], + [ + 379, + 124 + ], + [ + 381, + 124 + ] + ], + [ + [ + 374, + 123 + ], + [ + 373, + 109 + ], + [ + 380, + 100 + ], + [ + 368, + 147 + ], + [ + 380, + 114 + ], + [ + 369, + 159 + ] + ], + [ + [ + 364, + 115 + ], + [ + 392, + 120 + ], + [ + 366, + 126 + ], + [ + 361, + 103 + ], + [ + 340, + 121 + ], + [ + 332, + 98 + ] + ] + ] + }, + { + "task_id": "0001971", + "id": "RH20T_cfg4_task_0058_user_0014_scene_0003_cfg_0004_036422060909#221", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001971/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 344, + 166 + ], + [ + 333, + 187 + ], + [ + 352, + 159 + ], + [ + 336, + 172 + ], + [ + 363, + 137 + ], + [ + 346, + 152 + ] + ], + [ + [ + 333, + 145 + ], + [ + 333, + 132 + ], + [ + 341, + 124 + ], + [ + 324, + 168 + ], + [ + 341, + 137 + ], + [ + 324, + 179 + ] + ], + [ + [ + 347, + 132 + ], + [ + 323, + 133 + ], + [ + 337, + 149 + ], + [ + 358, + 114 + ], + [ + 360, + 147 + ], + [ + 382, + 114 + ] + ], + [ + [ + 326, + 136 + ], + [ + 303, + 128 + ], + [ + 324, + 152 + ], + [ + 328, + 119 + ], + [ + 347, + 161 + ], + [ + 354, + 128 + ] + ] + ] + }, + { + "task_id": "0001973", + "id": "RH20T_cfg6_task_0106_user_0020_scene_0004_cfg_0006_104122064161#391", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the object down from the shelf'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001973/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 467, + 188 + ], + [ + 493, + 217 + ], + [ + 495, + 166 + ], + [ + 439, + 210 + ], + [ + 471, + 138 + ], + [ + 415, + 181 + ] + ], + [ + [ + 413, + 219 + ], + [ + 420, + 188 + ], + [ + 381, + 208 + ], + [ + 444, + 230 + ], + [ + 372, + 241 + ], + [ + 437, + 263 + ] + ], + [ + [ + 429, + 190 + ], + [ + 433, + 167 + ], + [ + 435, + 220 + ], + [ + 424, + 162 + ], + [ + 431, + 239 + ], + [ + 421, + 183 + ] + ], + [ + [ + 433, + 215 + ], + [ + 396, + 212 + ], + [ + 431, + 243 + ], + [ + 435, + 190 + ], + [ + 472, + 247 + ], + [ + 471, + 192 + ] + ] + ] + }, + { + "task_id": "0001975", + "id": "RH20T_cfg5_task_0085_user_0007_scene_0002_cfg_0005_104122063678#496", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the bubble ring from the assembled bubble ring and ball'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001975/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 330, + 224 + ], + [ + 349, + 213 + ], + [ + 325, + 205 + ], + [ + 335, + 242 + ], + [ + 305, + 217 + ], + [ + 316, + 254 + ] + ], + [ + [ + 330, + 183 + ], + [ + 334, + 195 + ], + [ + 316, + 200 + ], + [ + 344, + 167 + ], + [ + 311, + 187 + ], + [ + 340, + 154 + ] + ], + [ + [ + 342, + 198 + ], + [ + 347, + 184 + ], + [ + 321, + 203 + ], + [ + 362, + 194 + ], + [ + 317, + 217 + ], + [ + 356, + 207 + ] + ], + [ + [ + 348, + 190 + ], + [ + 357, + 209 + ], + [ + 359, + 182 + ], + [ + 338, + 197 + ], + [ + 350, + 162 + ], + [ + 331, + 178 + ] + ] + ] + }, + { + "task_id": "0001976", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0004_cfg_0004_104122062295#243", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001976/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 0, + 137 + ], + [ + 28, + 158 + ], + [ + 0, + 95 + ], + [ + 0, + 183 + ], + [ + 0, + 67 + ], + [ + 0, + 160 + ] + ], + [ + [ + 10, + 183 + ], + [ + 6, + 141 + ], + [ + 0, + 171 + ], + [ + 59, + 194 + ], + [ + 0, + 210 + ], + [ + 59, + 232 + ] + ], + [ + [ + 16, + 157 + ], + [ + 0, + 184 + ], + [ + 5, + 192 + ], + [ + 27, + 118 + ], + [ + 39, + 168 + ], + [ + 63, + 98 + ] + ], + [ + [ + 0, + 198 + ], + [ + 0, + 244 + ], + [ + 0, + 199 + ], + [ + 0, + 198 + ], + [ + 8, + 159 + ], + [ + 0, + 149 + ] + ] + ] + }, + { + "task_id": "0001977", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0004_cfg_0004_104122062295#297", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001977/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 0, + 171 + ], + [ + 0, + 214 + ], + [ + 10, + 186 + ], + [ + 0, + 157 + ], + [ + 0, + 135 + ], + [ + 0, + 109 + ] + ], + [ + [ + 52, + 210 + ], + [ + 16, + 251 + ], + [ + 91, + 221 + ], + [ + 5, + 196 + ], + [ + 122, + 184 + ], + [ + 45, + 150 + ] + ], + [ + [ + 49, + 207 + ], + [ + 55, + 160 + ], + [ + 10, + 186 + ], + [ + 90, + 229 + ], + [ + 6, + 229 + ], + [ + 83, + 272 + ] + ], + [ + [ + 106, + 210 + ], + [ + 86, + 248 + ], + [ + 88, + 174 + ], + [ + 121, + 241 + ], + [ + 108, + 142 + ], + [ + 137, + 208 + ] + ] + ] + }, + { + "task_id": "0001978", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0008_cfg_0006_104122061602#141", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001978/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 410, + 195 + ], + [ + 410, + 170 + ], + [ + 389, + 188 + ], + [ + 429, + 202 + ], + [ + 388, + 215 + ], + [ + 429, + 227 + ] + ], + [ + [ + 383, + 201 + ], + [ + 365, + 198 + ], + [ + 397, + 185 + ], + [ + 368, + 216 + ], + [ + 417, + 188 + ], + [ + 387, + 220 + ] + ], + [ + [ + 374, + 173 + ], + [ + 379, + 195 + ], + [ + 352, + 173 + ], + [ + 397, + 172 + ], + [ + 347, + 150 + ], + [ + 392, + 148 + ] + ], + [ + [ + 404, + 185 + ], + [ + 420, + 187 + ], + [ + 403, + 210 + ], + [ + 404, + 159 + ], + [ + 388, + 206 + ], + [ + 389, + 157 + ] + ] + ] + }, + { + "task_id": "0001979", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0008_cfg_0006_104122061602#317", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001979/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 407, + 180 + ], + [ + 397, + 192 + ], + [ + 424, + 168 + ], + [ + 391, + 191 + ], + [ + 433, + 157 + ], + [ + 401, + 180 + ] + ], + [ + [ + 414, + 165 + ], + [ + 395, + 168 + ], + [ + 401, + 174 + ], + [ + 429, + 156 + ], + [ + 419, + 172 + ], + [ + 449, + 153 + ] + ], + [ + [ + 422, + 169 + ], + [ + 439, + 173 + ], + [ + 420, + 194 + ], + [ + 423, + 144 + ], + [ + 405, + 189 + ], + [ + 407, + 142 + ] + ], + [ + [ + 436, + 178 + ], + [ + 412, + 169 + ], + [ + 444, + 162 + ], + [ + 428, + 195 + ], + [ + 469, + 170 + ], + [ + 455, + 205 + ] + ] + ] + }, + { + "task_id": "0001980", + "id": "RH20T_cfg1_task_0016_user_0014_scene_0001_cfg_0001_819112070044#221", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001980/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 242, + 233 + ], + [ + 221, + 223 + ], + [ + 252, + 232 + ], + [ + 232, + 234 + ], + [ + 272, + 241 + ], + [ + 252, + 243 + ] + ], + [ + [ + 239, + 273 + ], + [ + 263, + 279 + ], + [ + 242, + 262 + ], + [ + 236, + 284 + ], + [ + 219, + 256 + ], + [ + 211, + 277 + ] + ], + [ + [ + 245, + 260 + ], + [ + 246, + 249 + ], + [ + 243, + 240 + ], + [ + 248, + 283 + ], + [ + 242, + 251 + ], + [ + 248, + 293 + ] + ], + [ + [ + 262, + 273 + ], + [ + 252, + 294 + ], + [ + 257, + 262 + ], + [ + 268, + 286 + ], + [ + 267, + 241 + ], + [ + 279, + 264 + ] + ] + ] + }, + { + "task_id": "0001982", + "id": "RH20T_cfg7_task_0016_user_0020_scene_0004_cfg_0007_104122061602#152", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001982/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 443, + 140 + ], + [ + 416, + 139 + ], + [ + 444, + 116 + ], + [ + 442, + 165 + ], + [ + 471, + 116 + ], + [ + 469, + 165 + ] + ], + [ + [ + 442, + 146 + ], + [ + 456, + 160 + ], + [ + 443, + 164 + ], + [ + 440, + 127 + ], + [ + 431, + 151 + ], + [ + 427, + 115 + ] + ], + [ + [ + 438, + 134 + ], + [ + 416, + 125 + ], + [ + 425, + 149 + ], + [ + 450, + 120 + ], + [ + 448, + 160 + ], + [ + 473, + 129 + ] + ], + [ + [ + 469, + 141 + ], + [ + 460, + 152 + ], + [ + 473, + 165 + ], + [ + 466, + 119 + ], + [ + 483, + 154 + ], + [ + 476, + 107 + ] + ] + ] + }, + { + "task_id": "0001984", + "id": "RH20T_cfg1_task_0080_user_0015_scene_0005_cfg_0001_819112071037#135", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Flip over and spread out the paper that is laid flat on the table'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001984/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 551, + 256 + ], + [ + 534, + 316 + ], + [ + 623, + 279 + ], + [ + 487, + 235 + ], + [ + 638, + 219 + ], + [ + 505, + 182 + ] + ], + [ + [ + 540, + 296 + ], + [ + 503, + 259 + ], + [ + 494, + 263 + ], + [ + 582, + 326 + ], + [ + 539, + 311 + ], + [ + 635, + 360 + ] + ], + [ + [ + 579, + 222 + ], + [ + 606, + 162 + ], + [ + 519, + 218 + ], + [ + 635, + 226 + ], + [ + 497, + 274 + ], + [ + 609, + 278 + ] + ], + [ + [ + 640, + 249 + ], + [ + 640, + 228 + ], + [ + 605, + 198 + ], + [ + 640, + 314 + ], + [ + 541, + 217 + ], + [ + 624, + 337 + ] + ] + ] + }, + { + "task_id": "0001985", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0010_cfg_0001_039422060546#55", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001985/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 223, + 189 + ], + [ + 195, + 191 + ], + [ + 223, + 225 + ], + [ + 224, + 155 + ], + [ + 253, + 225 + ], + [ + 253, + 152 + ] + ], + [ + [ + 231, + 202 + ], + [ + 193, + 197 + ], + [ + 222, + 234 + ], + [ + 239, + 174 + ], + [ + 262, + 239 + ], + [ + 275, + 178 + ] + ], + [ + [ + 197, + 214 + ], + [ + 208, + 178 + ], + [ + 181, + 201 + ], + [ + 215, + 229 + ], + [ + 171, + 235 + ], + [ + 204, + 267 + ] + ], + [ + [ + 244, + 203 + ], + [ + 232, + 183 + ], + [ + 230, + 235 + ], + [ + 257, + 174 + ], + [ + 242, + 251 + ], + [ + 267, + 192 + ] + ] + ] + }, + { + "task_id": "0001987", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0003_cfg_0001_038522062547#141", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001987/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 376, + 205 + ], + [ + 378, + 179 + ], + [ + 362, + 204 + ], + [ + 388, + 206 + ], + [ + 359, + 231 + ], + [ + 386, + 230 + ] + ], + [ + [ + 401, + 177 + ], + [ + 404, + 167 + ], + [ + 428, + 180 + ], + [ + 374, + 174 + ], + [ + 424, + 189 + ], + [ + 373, + 183 + ] + ], + [ + [ + 377, + 177 + ], + [ + 360, + 192 + ], + [ + 361, + 177 + ], + [ + 395, + 176 + ], + [ + 376, + 163 + ], + [ + 411, + 161 + ] + ], + [ + [ + 411, + 182 + ], + [ + 412, + 172 + ], + [ + 401, + 203 + ], + [ + 422, + 159 + ], + [ + 399, + 214 + ], + [ + 422, + 169 + ] + ] + ] + }, + { + "task_id": "0001990", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0003_cfg_0001_038522062547#745", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001990/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 413, + 69 + ], + [ + 396, + 79 + ], + [ + 414, + 92 + ], + [ + 411, + 45 + ], + [ + 429, + 83 + ], + [ + 428, + 36 + ] + ], + [ + [ + 412, + 66 + ], + [ + 430, + 51 + ], + [ + 418, + 79 + ], + [ + 405, + 53 + ], + [ + 401, + 94 + ], + [ + 386, + 69 + ] + ], + [ + [ + 413, + 67 + ], + [ + 412, + 52 + ], + [ + 437, + 72 + ], + [ + 391, + 61 + ], + [ + 437, + 86 + ], + [ + 394, + 75 + ] + ], + [ + [ + 413, + 56 + ], + [ + 415, + 41 + ], + [ + 390, + 53 + ], + [ + 436, + 60 + ], + [ + 386, + 69 + ], + [ + 434, + 76 + ] + ] + ] + }, + { + "task_id": "0001991", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0003_cfg_0001_038522062547#1136", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001991/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 466, + 52 + ], + [ + 471, + 38 + ], + [ + 482, + 69 + ], + [ + 451, + 36 + ], + [ + 477, + 80 + ], + [ + 447, + 49 + ] + ], + [ + [ + 472, + 46 + ], + [ + 472, + 23 + ], + [ + 475, + 55 + ], + [ + 468, + 36 + ], + [ + 476, + 78 + ], + [ + 469, + 61 + ] + ], + [ + [ + 468, + 48 + ], + [ + 467, + 49 + ], + [ + 490, + 56 + ], + [ + 446, + 40 + ], + [ + 492, + 55 + ], + [ + 447, + 39 + ] + ], + [ + [ + 493, + 61 + ], + [ + 481, + 71 + ], + [ + 480, + 78 + ], + [ + 505, + 44 + ], + [ + 494, + 68 + ], + [ + 519, + 32 + ] + ] + ] + }, + { + "task_id": "0001993", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0003_cfg_0001_038522062547#1598", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001993/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 470, + 176 + ], + [ + 476, + 165 + ], + [ + 499, + 182 + ], + [ + 443, + 171 + ], + [ + 492, + 192 + ], + [ + 440, + 182 + ] + ], + [ + [ + 488, + 177 + ], + [ + 464, + 160 + ], + [ + 487, + 183 + ], + [ + 489, + 171 + ], + [ + 513, + 201 + ], + [ + 512, + 187 + ] + ], + [ + [ + 444, + 175 + ], + [ + 431, + 197 + ], + [ + 451, + 181 + ], + [ + 437, + 168 + ], + [ + 464, + 160 + ], + [ + 451, + 145 + ] + ], + [ + [ + 489, + 209 + ], + [ + 490, + 184 + ], + [ + 467, + 213 + ], + [ + 513, + 204 + ], + [ + 466, + 237 + ], + [ + 512, + 231 + ] + ] + ] + }, + { + "task_id": "0001996", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_038522062547#434", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001996/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 382, + 52 + ], + [ + 365, + 46 + ], + [ + 385, + 76 + ], + [ + 378, + 26 + ], + [ + 403, + 83 + ], + [ + 397, + 32 + ] + ], + [ + [ + 365, + 43 + ], + [ + 339, + 43 + ], + [ + 368, + 44 + ], + [ + 362, + 42 + ], + [ + 396, + 45 + ], + [ + 386, + 43 + ] + ], + [ + [ + 377, + 49 + ], + [ + 358, + 37 + ], + [ + 372, + 73 + ], + [ + 382, + 25 + ], + [ + 391, + 84 + ], + [ + 400, + 37 + ] + ], + [ + [ + 373, + 44 + ], + [ + 371, + 26 + ], + [ + 375, + 65 + ], + [ + 371, + 23 + ], + [ + 377, + 80 + ], + [ + 373, + 40 + ] + ] + ] + }, + { + "task_id": "0001998", + "id": "RH20T_cfg4_task_0091_user_0007_scene_0006_cfg_0004_104122062295#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001998/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 199, + 83 + ], + [ + 169, + 96 + ], + [ + 183, + 40 + ], + [ + 215, + 124 + ], + [ + 210, + 32 + ], + [ + 238, + 111 + ] + ], + [ + [ + 175, + 102 + ], + [ + 158, + 131 + ], + [ + 197, + 127 + ], + [ + 151, + 74 + ], + [ + 211, + 100 + ], + [ + 168, + 47 + ] + ], + [ + [ + 195, + 108 + ], + [ + 200, + 70 + ], + [ + 158, + 97 + ], + [ + 233, + 119 + ], + [ + 155, + 131 + ], + [ + 228, + 153 + ] + ], + [ + [ + 199, + 102 + ], + [ + 184, + 136 + ], + [ + 181, + 82 + ], + [ + 215, + 118 + ], + [ + 197, + 48 + ], + [ + 228, + 88 + ] + ] + ] + }, + { + "task_id": "0001999", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_035622060973#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0001999/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 156, + 139 + ], + [ + 149, + 127 + ], + [ + 145, + 126 + ], + [ + 168, + 152 + ], + [ + 152, + 138 + ], + [ + 174, + 163 + ] + ], + [ + [ + 158, + 126 + ], + [ + 167, + 138 + ], + [ + 142, + 121 + ], + [ + 174, + 131 + ], + [ + 131, + 109 + ], + [ + 165, + 118 + ] + ], + [ + [ + 144, + 150 + ], + [ + 154, + 148 + ], + [ + 163, + 144 + ], + [ + 124, + 157 + ], + [ + 154, + 146 + ], + [ + 115, + 158 + ] + ], + [ + [ + 143, + 151 + ], + [ + 133, + 157 + ], + [ + 123, + 147 + ], + [ + 162, + 155 + ], + [ + 134, + 141 + ], + [ + 173, + 150 + ] + ] + ] + }, + { + "task_id": "0002000", + "id": "RH20T_cfg2_task_0083_user_0007_scene_0009_cfg_0002_104122063678#455", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Swat with a flyswatter'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002000/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 446, + 219 + ], + [ + 443, + 245 + ], + [ + 445, + 214 + ], + [ + 446, + 224 + ], + [ + 448, + 189 + ], + [ + 449, + 196 + ] + ], + [ + [ + 445, + 211 + ], + [ + 453, + 196 + ], + [ + 426, + 222 + ], + [ + 462, + 201 + ], + [ + 419, + 235 + ], + [ + 454, + 215 + ] + ], + [ + [ + 447, + 235 + ], + [ + 457, + 259 + ], + [ + 465, + 234 + ], + [ + 428, + 236 + ], + [ + 456, + 212 + ], + [ + 419, + 213 + ] + ], + [ + [ + 438, + 238 + ], + [ + 436, + 230 + ], + [ + 460, + 252 + ], + [ + 417, + 224 + ], + [ + 461, + 258 + ], + [ + 419, + 231 + ] + ] + ] + }, + { + "task_id": "0002001", + "id": "RH20T_cfg1_task_0009_user_0014_scene_0004_cfg_0001_035622060973#216", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002001/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 243, + 213 + ], + [ + 253, + 198 + ], + [ + 260, + 227 + ], + [ + 225, + 198 + ], + [ + 249, + 243 + ], + [ + 213, + 213 + ] + ], + [ + [ + 228, + 211 + ], + [ + 227, + 200 + ], + [ + 214, + 197 + ], + [ + 243, + 226 + ], + [ + 215, + 208 + ], + [ + 243, + 235 + ] + ], + [ + [ + 213, + 217 + ], + [ + 213, + 230 + ], + [ + 210, + 236 + ], + [ + 215, + 199 + ], + [ + 209, + 222 + ], + [ + 214, + 185 + ] + ], + [ + [ + 219, + 217 + ], + [ + 232, + 202 + ], + [ + 204, + 202 + ], + [ + 234, + 232 + ], + [ + 189, + 216 + ], + [ + 220, + 247 + ] + ] + ] + }, + { + "task_id": "0002004", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122063633#601", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002004/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 237, + 260 + ], + [ + 260, + 245 + ], + [ + 231, + 249 + ], + [ + 245, + 273 + ], + [ + 209, + 264 + ], + [ + 221, + 289 + ] + ], + [ + [ + 238, + 247 + ], + [ + 237, + 273 + ], + [ + 222, + 241 + ], + [ + 252, + 253 + ], + [ + 223, + 216 + ], + [ + 253, + 230 + ] + ], + [ + [ + 226, + 234 + ], + [ + 254, + 234 + ], + [ + 229, + 252 + ], + [ + 224, + 213 + ], + [ + 202, + 252 + ], + [ + 195, + 213 + ] + ], + [ + [ + 213, + 241 + ], + [ + 214, + 218 + ], + [ + 216, + 228 + ], + [ + 210, + 257 + ], + [ + 215, + 249 + ], + [ + 209, + 280 + ] + ] + ] + }, + { + "task_id": "0002007", + "id": "RH20T_cfg6_task_0007_user_0014_scene_0007_cfg_0006_104122061602#218", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002007/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 496, + 253 + ], + [ + 489, + 243 + ], + [ + 524, + 238 + ], + [ + 469, + 267 + ], + [ + 529, + 248 + ], + [ + 476, + 275 + ] + ], + [ + [ + 505, + 244 + ], + [ + 489, + 259 + ], + [ + 533, + 254 + ], + [ + 480, + 236 + ], + [ + 548, + 240 + ], + [ + 495, + 223 + ] + ], + [ + [ + 520, + 263 + ], + [ + 497, + 267 + ], + [ + 511, + 291 + ], + [ + 529, + 234 + ], + [ + 532, + 286 + ], + [ + 550, + 231 + ] + ], + [ + [ + 523, + 253 + ], + [ + 544, + 249 + ], + [ + 539, + 282 + ], + [ + 508, + 226 + ], + [ + 520, + 284 + ], + [ + 491, + 231 + ] + ] + ] + }, + { + "task_id": "0002010", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_750612070853#213", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002010/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 317, + 178 + ], + [ + 336, + 144 + ], + [ + 290, + 171 + ], + [ + 347, + 185 + ], + [ + 270, + 205 + ], + [ + 326, + 223 + ] + ], + [ + [ + 327, + 181 + ], + [ + 325, + 145 + ], + [ + 294, + 170 + ], + [ + 363, + 193 + ], + [ + 297, + 203 + ], + [ + 363, + 227 + ] + ], + [ + [ + 323, + 151 + ], + [ + 288, + 137 + ], + [ + 342, + 128 + ], + [ + 305, + 171 + ], + [ + 376, + 142 + ], + [ + 336, + 183 + ] + ], + [ + [ + 362, + 176 + ], + [ + 381, + 158 + ], + [ + 330, + 152 + ], + [ + 394, + 200 + ], + [ + 308, + 170 + ], + [ + 375, + 223 + ] + ] + ] + }, + { + "task_id": "0002011", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_750612070853#400", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002011/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 507, + 307 + ], + [ + 520, + 255 + ], + [ + 489, + 305 + ], + [ + 530, + 310 + ], + [ + 477, + 353 + ], + [ + 515, + 360 + ] + ], + [ + [ + 485, + 260 + ], + [ + 498, + 218 + ], + [ + 432, + 253 + ], + [ + 539, + 268 + ], + [ + 423, + 291 + ], + [ + 524, + 306 + ] + ], + [ + [ + 466, + 248 + ], + [ + 474, + 279 + ], + [ + 492, + 278 + ], + [ + 437, + 217 + ], + [ + 483, + 250 + ], + [ + 432, + 193 + ] + ], + [ + [ + 517, + 315 + ], + [ + 524, + 357 + ], + [ + 569, + 344 + ], + [ + 467, + 287 + ], + [ + 559, + 307 + ], + [ + 466, + 257 + ] + ] + ] + }, + { + "task_id": "0002012", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_750612070853#641", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002012/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 280, + 289 + ], + [ + 234, + 250 + ], + [ + 262, + 276 + ], + [ + 302, + 305 + ], + [ + 300, + 308 + ], + [ + 348, + 345 + ] + ], + [ + [ + 296, + 359 + ], + [ + 297, + 321 + ], + [ + 260, + 318 + ], + [ + 339, + 360 + ], + [ + 262, + 351 + ], + [ + 335, + 360 + ] + ], + [ + [ + 308, + 328 + ], + [ + 259, + 353 + ], + [ + 286, + 282 + ], + [ + 333, + 360 + ], + [ + 331, + 261 + ], + [ + 381, + 352 + ] + ], + [ + [ + 260, + 360 + ], + [ + 309, + 360 + ], + [ + 221, + 360 + ], + [ + 299, + 345 + ], + [ + 170, + 360 + ], + [ + 248, + 309 + ] + ] + ] + }, + { + "task_id": "0002013", + "id": "RH20T_cfg3_task_0018_user_0010_scene_0003_cfg_0003_104122062823#191", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002013/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 228, + 134 + ], + [ + 240, + 102 + ], + [ + 218, + 135 + ], + [ + 236, + 134 + ], + [ + 206, + 170 + ], + [ + 225, + 164 + ] + ], + [ + [ + 251, + 117 + ], + [ + 234, + 125 + ], + [ + 238, + 143 + ], + [ + 262, + 92 + ], + [ + 257, + 135 + ], + [ + 281, + 82 + ] + ], + [ + [ + 224, + 169 + ], + [ + 205, + 167 + ], + [ + 222, + 138 + ], + [ + 226, + 201 + ], + [ + 239, + 142 + ], + [ + 242, + 201 + ] + ], + [ + [ + 216, + 137 + ], + [ + 215, + 109 + ], + [ + 243, + 148 + ], + [ + 189, + 126 + ], + [ + 242, + 173 + ], + [ + 190, + 151 + ] + ] + ] + }, + { + "task_id": "0002015", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0005_cfg_0006_104122061018#462", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002015/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 244, + 55 + ], + [ + 232, + 71 + ], + [ + 226, + 31 + ], + [ + 261, + 77 + ], + [ + 238, + 16 + ], + [ + 271, + 62 + ] + ], + [ + [ + 254, + 47 + ], + [ + 236, + 50 + ], + [ + 263, + 22 + ], + [ + 245, + 72 + ], + [ + 282, + 17 + ], + [ + 264, + 70 + ] + ], + [ + [ + 253, + 61 + ], + [ + 254, + 32 + ], + [ + 242, + 46 + ], + [ + 265, + 76 + ], + [ + 242, + 72 + ], + [ + 263, + 103 + ] + ], + [ + [ + 281, + 84 + ], + [ + 286, + 61 + ], + [ + 270, + 99 + ], + [ + 293, + 68 + ], + [ + 264, + 123 + ], + [ + 287, + 94 + ] + ] + ] + }, + { + "task_id": "0002017", + "id": "RH20T_cfg5_task_0072_user_0007_scene_0005_cfg_0005_104122063678#444", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drop coins into a piggy bank'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002017/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 213 + ], + [ + 391, + 240 + ], + [ + 393, + 217 + ], + [ + 394, + 209 + ], + [ + 395, + 192 + ], + [ + 397, + 181 + ] + ], + [ + [ + 410, + 221 + ], + [ + 423, + 207 + ], + [ + 395, + 227 + ], + [ + 427, + 214 + ], + [ + 382, + 242 + ], + [ + 414, + 230 + ] + ], + [ + [ + 397, + 206 + ], + [ + 403, + 190 + ], + [ + 404, + 228 + ], + [ + 390, + 186 + ], + [ + 397, + 243 + ], + [ + 384, + 201 + ] + ], + [ + [ + 406, + 218 + ], + [ + 389, + 218 + ], + [ + 395, + 193 + ], + [ + 417, + 242 + ], + [ + 413, + 191 + ], + [ + 436, + 243 + ] + ] + ] + }, + { + "task_id": "0002018", + "id": "RH20T_cfg1_task_0002_user_0014_scene_0004_cfg_0001_819112070044#136", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002018/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 381, + 202 + ], + [ + 393, + 220 + ], + [ + 363, + 214 + ], + [ + 399, + 191 + ], + [ + 352, + 196 + ], + [ + 387, + 173 + ] + ], + [ + [ + 392, + 201 + ], + [ + 375, + 195 + ], + [ + 397, + 213 + ], + [ + 387, + 188 + ], + [ + 415, + 220 + ], + [ + 405, + 194 + ] + ], + [ + [ + 390, + 181 + ], + [ + 394, + 166 + ], + [ + 404, + 174 + ], + [ + 375, + 190 + ], + [ + 400, + 188 + ], + [ + 371, + 204 + ] + ], + [ + [ + 393, + 204 + ], + [ + 394, + 189 + ], + [ + 414, + 208 + ], + [ + 372, + 200 + ], + [ + 413, + 223 + ], + [ + 369, + 216 + ] + ] + ] + }, + { + "task_id": "0002019", + "id": "RH20T_cfg7_task_0029_user_0020_scene_0010_cfg_0007_104122060811#170", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002019/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 342, + 136 + ], + [ + 343, + 124 + ], + [ + 332, + 143 + ], + [ + 352, + 129 + ], + [ + 332, + 155 + ], + [ + 351, + 140 + ] + ], + [ + [ + 326, + 134 + ], + [ + 324, + 141 + ], + [ + 336, + 125 + ], + [ + 316, + 141 + ], + [ + 337, + 119 + ], + [ + 318, + 135 + ] + ], + [ + [ + 325, + 135 + ], + [ + 339, + 132 + ], + [ + 324, + 144 + ], + [ + 327, + 126 + ], + [ + 310, + 146 + ], + [ + 313, + 128 + ] + ], + [ + [ + 321, + 140 + ], + [ + 331, + 147 + ], + [ + 318, + 152 + ], + [ + 324, + 128 + ], + [ + 307, + 144 + ], + [ + 313, + 121 + ] + ] + ] + }, + { + "task_id": "0002021", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_105422061350#272", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002021/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 99 + ], + [ + 262, + 86 + ], + [ + 274, + 122 + ], + [ + 296, + 78 + ], + [ + 298, + 134 + ], + [ + 319, + 90 + ] + ], + [ + [ + 294, + 119 + ], + [ + 309, + 136 + ], + [ + 315, + 107 + ], + [ + 273, + 130 + ], + [ + 299, + 87 + ], + [ + 257, + 112 + ] + ], + [ + [ + 287, + 104 + ], + [ + 292, + 86 + ], + [ + 297, + 86 + ], + [ + 277, + 123 + ], + [ + 292, + 103 + ], + [ + 272, + 139 + ] + ], + [ + [ + 296, + 90 + ], + [ + 317, + 74 + ], + [ + 282, + 92 + ], + [ + 309, + 87 + ], + [ + 262, + 107 + ], + [ + 290, + 101 + ] + ] + ] + }, + { + "task_id": "0002024", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_105422061350#966", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002024/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 265, + 138 + ], + [ + 267, + 120 + ], + [ + 279, + 118 + ], + [ + 249, + 158 + ], + [ + 276, + 135 + ], + [ + 248, + 173 + ] + ], + [ + [ + 251, + 119 + ], + [ + 247, + 139 + ], + [ + 276, + 114 + ], + [ + 223, + 126 + ], + [ + 281, + 93 + ], + [ + 226, + 104 + ] + ], + [ + [ + 247, + 145 + ], + [ + 223, + 134 + ], + [ + 242, + 170 + ], + [ + 252, + 117 + ], + [ + 267, + 181 + ], + [ + 278, + 128 + ] + ], + [ + [ + 251, + 147 + ], + [ + 277, + 160 + ], + [ + 248, + 142 + ], + [ + 255, + 151 + ], + [ + 221, + 129 + ], + [ + 231, + 139 + ] + ] + ] + }, + { + "task_id": "0002026", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122060811#395", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002026/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 122 + ], + [ + 372, + 133 + ], + [ + 368, + 116 + ], + [ + 358, + 128 + ], + [ + 359, + 105 + ], + [ + 349, + 117 + ] + ], + [ + [ + 368, + 108 + ], + [ + 380, + 113 + ], + [ + 359, + 115 + ], + [ + 376, + 102 + ], + [ + 347, + 110 + ], + [ + 364, + 97 + ] + ], + [ + [ + 375, + 118 + ], + [ + 388, + 115 + ], + [ + 372, + 126 + ], + [ + 379, + 109 + ], + [ + 360, + 129 + ], + [ + 366, + 112 + ] + ], + [ + [ + 372, + 104 + ], + [ + 371, + 116 + ], + [ + 368, + 111 + ], + [ + 377, + 97 + ], + [ + 370, + 99 + ], + [ + 378, + 85 + ] + ] + ] + }, + { + "task_id": "0002027", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122060811#686", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002027/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 399, + 136 + ], + [ + 406, + 147 + ], + [ + 412, + 132 + ], + [ + 387, + 140 + ], + [ + 405, + 121 + ], + [ + 380, + 128 + ] + ], + [ + [ + 383, + 142 + ], + [ + 394, + 139 + ], + [ + 376, + 150 + ], + [ + 389, + 134 + ], + [ + 365, + 153 + ], + [ + 378, + 137 + ] + ], + [ + [ + 391, + 152 + ], + [ + 395, + 139 + ], + [ + 382, + 155 + ], + [ + 400, + 149 + ], + [ + 379, + 168 + ], + [ + 397, + 161 + ] + ], + [ + [ + 397, + 148 + ], + [ + 403, + 148 + ], + [ + 411, + 147 + ], + [ + 384, + 149 + ], + [ + 406, + 147 + ], + [ + 379, + 149 + ] + ] + ] + }, + { + "task_id": "0002028", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122060811#1211", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002028/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 137 + ], + [ + 318, + 131 + ], + [ + 311, + 143 + ], + [ + 299, + 131 + ], + [ + 298, + 149 + ], + [ + 286, + 137 + ] + ], + [ + [ + 296, + 135 + ], + [ + 290, + 148 + ], + [ + 303, + 136 + ], + [ + 290, + 133 + ], + [ + 309, + 123 + ], + [ + 296, + 120 + ] + ], + [ + [ + 304, + 151 + ], + [ + 300, + 139 + ], + [ + 307, + 142 + ], + [ + 301, + 159 + ], + [ + 311, + 154 + ], + [ + 305, + 171 + ] + ], + [ + [ + 299, + 145 + ], + [ + 290, + 135 + ], + [ + 309, + 135 + ], + [ + 289, + 155 + ], + [ + 319, + 145 + ], + [ + 298, + 165 + ] + ] + ] + }, + { + "task_id": "0002029", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122060811#1608", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002029/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 162 + ], + [ + 289, + 153 + ], + [ + 298, + 160 + ], + [ + 271, + 163 + ], + [ + 293, + 168 + ], + [ + 267, + 171 + ] + ], + [ + [ + 281, + 167 + ], + [ + 280, + 167 + ], + [ + 277, + 182 + ], + [ + 285, + 153 + ], + [ + 278, + 182 + ], + [ + 286, + 154 + ] + ], + [ + [ + 290, + 143 + ], + [ + 285, + 131 + ], + [ + 294, + 149 + ], + [ + 285, + 137 + ], + [ + 299, + 162 + ], + [ + 290, + 150 + ] + ], + [ + [ + 288, + 172 + ], + [ + 285, + 159 + ], + [ + 302, + 171 + ], + [ + 274, + 174 + ], + [ + 305, + 184 + ], + [ + 277, + 187 + ] + ] + ] + }, + { + "task_id": "0002030", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122060811#2013", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002030/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 336, + 166 + ], + [ + 342, + 158 + ], + [ + 349, + 164 + ], + [ + 322, + 167 + ], + [ + 343, + 172 + ], + [ + 317, + 175 + ] + ], + [ + [ + 340, + 171 + ], + [ + 341, + 162 + ], + [ + 328, + 179 + ], + [ + 352, + 164 + ], + [ + 327, + 188 + ], + [ + 351, + 173 + ] + ], + [ + [ + 349, + 169 + ], + [ + 341, + 156 + ], + [ + 351, + 167 + ], + [ + 347, + 171 + ], + [ + 358, + 181 + ], + [ + 354, + 183 + ] + ], + [ + [ + 336, + 169 + ], + [ + 322, + 176 + ], + [ + 329, + 157 + ], + [ + 343, + 182 + ], + [ + 343, + 151 + ], + [ + 357, + 176 + ] + ] + ] + }, + { + "task_id": "0002031", + "id": "RH20T_cfg1_task_0053_user_0014_scene_0003_cfg_0001_819112071037#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a clamp'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002031/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 544, + 212 + ], + [ + 568, + 233 + ], + [ + 503, + 246 + ], + [ + 587, + 175 + ], + [ + 475, + 225 + ], + [ + 561, + 148 + ] + ], + [ + [ + 626, + 197 + ], + [ + 640, + 147 + ], + [ + 640, + 220 + ], + [ + 595, + 178 + ], + [ + 625, + 271 + ], + [ + 565, + 220 + ] + ], + [ + [ + 587, + 150 + ], + [ + 583, + 130 + ], + [ + 640, + 124 + ], + [ + 535, + 173 + ], + [ + 640, + 147 + ], + [ + 535, + 198 + ] + ], + [ + [ + 572, + 204 + ], + [ + 640, + 217 + ], + [ + 565, + 237 + ], + [ + 580, + 164 + ], + [ + 509, + 224 + ], + [ + 513, + 154 + ] + ] + ] + }, + { + "task_id": "0002033", + "id": "RH20T_cfg5_task_0091_user_0007_scene_0004_cfg_0005_104122061850#354", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002033/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 350, + 119 + ], + [ + 344, + 102 + ], + [ + 326, + 121 + ], + [ + 375, + 116 + ], + [ + 332, + 136 + ], + [ + 379, + 131 + ] + ], + [ + [ + 323, + 97 + ], + [ + 325, + 117 + ], + [ + 339, + 83 + ], + [ + 308, + 109 + ], + [ + 336, + 63 + ], + [ + 306, + 90 + ] + ], + [ + [ + 369, + 137 + ], + [ + 360, + 137 + ], + [ + 347, + 149 + ], + [ + 391, + 126 + ], + [ + 356, + 149 + ], + [ + 402, + 125 + ] + ], + [ + [ + 366, + 119 + ], + [ + 364, + 95 + ], + [ + 393, + 117 + ], + [ + 340, + 122 + ], + [ + 395, + 142 + ], + [ + 342, + 147 + ] + ] + ] + }, + { + "task_id": "0002035", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122064161#362", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002035/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 416, + 273 + ], + [ + 430, + 283 + ], + [ + 416, + 299 + ], + [ + 416, + 248 + ], + [ + 401, + 289 + ], + [ + 402, + 237 + ] + ], + [ + [ + 412, + 272 + ], + [ + 419, + 266 + ], + [ + 413, + 301 + ], + [ + 412, + 245 + ], + [ + 407, + 305 + ], + [ + 406, + 252 + ] + ], + [ + [ + 425, + 264 + ], + [ + 445, + 249 + ], + [ + 441, + 288 + ], + [ + 409, + 241 + ], + [ + 421, + 305 + ], + [ + 388, + 256 + ] + ], + [ + [ + 392, + 283 + ], + [ + 413, + 297 + ], + [ + 403, + 296 + ], + [ + 381, + 270 + ], + [ + 384, + 284 + ], + [ + 362, + 257 + ] + ] + ] + }, + { + "task_id": "0002038", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122064161#942", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002038/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 414, + 119 + ], + [ + 417, + 129 + ], + [ + 411, + 137 + ], + [ + 417, + 100 + ], + [ + 409, + 127 + ], + [ + 413, + 91 + ] + ], + [ + [ + 415, + 122 + ], + [ + 395, + 122 + ], + [ + 419, + 126 + ], + [ + 412, + 119 + ], + [ + 440, + 126 + ], + [ + 431, + 119 + ] + ], + [ + [ + 430, + 115 + ], + [ + 409, + 115 + ], + [ + 430, + 135 + ], + [ + 429, + 96 + ], + [ + 450, + 136 + ], + [ + 450, + 96 + ] + ], + [ + [ + 421, + 128 + ], + [ + 432, + 114 + ], + [ + 424, + 143 + ], + [ + 418, + 112 + ], + [ + 414, + 158 + ], + [ + 407, + 127 + ] + ] + ] + }, + { + "task_id": "0002040", + "id": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122064161#60", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the pen into the pen holder'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002040/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 312, + 168 + ], + [ + 323, + 151 + ], + [ + 307, + 151 + ], + [ + 317, + 186 + ], + [ + 297, + 167 + ], + [ + 307, + 202 + ] + ], + [ + [ + 322, + 175 + ], + [ + 300, + 170 + ], + [ + 330, + 154 + ], + [ + 315, + 196 + ], + [ + 352, + 161 + ], + [ + 337, + 201 + ] + ], + [ + [ + 327, + 175 + ], + [ + 305, + 171 + ], + [ + 333, + 155 + ], + [ + 321, + 193 + ], + [ + 355, + 159 + ], + [ + 342, + 196 + ] + ], + [ + [ + 313, + 182 + ], + [ + 318, + 169 + ], + [ + 312, + 202 + ], + [ + 313, + 163 + ], + [ + 308, + 214 + ], + [ + 309, + 176 + ] + ] + ] + }, + { + "task_id": "0002041", + "id": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122064161#129", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the pen into the pen holder'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002041/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 167 + ], + [ + 292, + 165 + ], + [ + 315, + 145 + ], + [ + 312, + 188 + ], + [ + 337, + 146 + ], + [ + 334, + 190 + ] + ], + [ + [ + 296, + 159 + ], + [ + 299, + 181 + ], + [ + 316, + 157 + ], + [ + 278, + 161 + ], + [ + 313, + 135 + ], + [ + 275, + 140 + ] + ], + [ + [ + 298, + 166 + ], + [ + 297, + 172 + ], + [ + 296, + 187 + ], + [ + 300, + 144 + ], + [ + 297, + 180 + ], + [ + 300, + 139 + ] + ], + [ + [ + 298, + 154 + ], + [ + 280, + 144 + ], + [ + 299, + 138 + ], + [ + 296, + 168 + ], + [ + 318, + 148 + ], + [ + 314, + 178 + ] + ] + ] + }, + { + "task_id": "0002042", + "id": "RH20T_cfg7_task_0046_user_0014_scene_0001_cfg_0007_104122061018#94", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002042/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 343, + 130 + ], + [ + 333, + 121 + ], + [ + 371, + 115 + ], + [ + 314, + 144 + ], + [ + 384, + 124 + ], + [ + 323, + 154 + ] + ], + [ + [ + 330, + 149 + ], + [ + 360, + 140 + ], + [ + 316, + 122 + ], + [ + 346, + 177 + ], + [ + 286, + 131 + ], + [ + 314, + 188 + ] + ], + [ + [ + 345, + 119 + ], + [ + 317, + 135 + ], + [ + 328, + 88 + ], + [ + 362, + 151 + ], + [ + 355, + 74 + ], + [ + 387, + 135 + ] + ], + [ + [ + 328, + 124 + ], + [ + 333, + 103 + ], + [ + 313, + 98 + ], + [ + 343, + 151 + ], + [ + 309, + 118 + ], + [ + 337, + 169 + ] + ] + ] + }, + { + "task_id": "0002043", + "id": "RH20T_cfg6_task_0015_user_0014_scene_0005_cfg_0006_104122064161#70", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002043/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 401, + 197 + ], + [ + 376, + 201 + ], + [ + 402, + 203 + ], + [ + 400, + 191 + ], + [ + 425, + 199 + ], + [ + 426, + 186 + ] + ], + [ + [ + 392, + 211 + ], + [ + 400, + 210 + ], + [ + 386, + 234 + ], + [ + 398, + 188 + ], + [ + 378, + 233 + ], + [ + 390, + 189 + ] + ], + [ + [ + 370, + 196 + ], + [ + 347, + 200 + ], + [ + 371, + 206 + ], + [ + 370, + 184 + ], + [ + 393, + 202 + ], + [ + 394, + 180 + ] + ], + [ + [ + 389, + 224 + ], + [ + 399, + 202 + ], + [ + 391, + 233 + ], + [ + 388, + 216 + ], + [ + 381, + 256 + ], + [ + 380, + 236 + ] + ] + ] + }, + { + "task_id": "0002044", + "id": "RH20T_cfg1_task_0001_user_0009_scene_0003_cfg_0001_038522063145#87", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002044/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 428, + 129 + ], + [ + 435, + 110 + ], + [ + 447, + 143 + ], + [ + 409, + 116 + ], + [ + 439, + 161 + ], + [ + 403, + 134 + ] + ], + [ + [ + 453, + 125 + ], + [ + 443, + 137 + ], + [ + 448, + 146 + ], + [ + 458, + 102 + ], + [ + 457, + 134 + ], + [ + 467, + 91 + ] + ], + [ + [ + 449, + 129 + ], + [ + 465, + 127 + ], + [ + 434, + 111 + ], + [ + 464, + 148 + ], + [ + 417, + 113 + ], + [ + 447, + 151 + ] + ], + [ + [ + 427, + 131 + ], + [ + 447, + 118 + ], + [ + 433, + 132 + ], + [ + 421, + 131 + ], + [ + 412, + 146 + ], + [ + 402, + 143 + ] + ] + ] + }, + { + "task_id": "0002045", + "id": "RH20T_cfg6_task_0038_user_0014_scene_0006_cfg_0006_037522061512#89", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002045/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 251, + 83 + ], + [ + 231, + 85 + ], + [ + 262, + 71 + ], + [ + 241, + 94 + ], + [ + 281, + 69 + ], + [ + 260, + 91 + ] + ], + [ + [ + 271, + 89 + ], + [ + 291, + 99 + ], + [ + 262, + 111 + ], + [ + 279, + 66 + ], + [ + 242, + 102 + ], + [ + 259, + 57 + ] + ], + [ + [ + 252, + 94 + ], + [ + 247, + 77 + ], + [ + 265, + 78 + ], + [ + 239, + 111 + ], + [ + 268, + 94 + ], + [ + 244, + 126 + ] + ], + [ + [ + 265, + 123 + ], + [ + 265, + 99 + ], + [ + 281, + 122 + ], + [ + 247, + 125 + ], + [ + 281, + 145 + ], + [ + 248, + 150 + ] + ] + ] + }, + { + "task_id": "0002047", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0009_cfg_0005_104122063678#468", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002047/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 279, + 176 + ], + [ + 282, + 159 + ], + [ + 263, + 188 + ], + [ + 294, + 164 + ], + [ + 261, + 204 + ], + [ + 290, + 179 + ] + ], + [ + [ + 280, + 152 + ], + [ + 274, + 164 + ], + [ + 258, + 142 + ], + [ + 302, + 162 + ], + [ + 264, + 132 + ], + [ + 307, + 151 + ] + ], + [ + [ + 289, + 183 + ], + [ + 274, + 198 + ], + [ + 273, + 186 + ], + [ + 306, + 180 + ], + [ + 286, + 172 + ], + [ + 319, + 166 + ] + ], + [ + [ + 301, + 157 + ], + [ + 278, + 150 + ], + [ + 304, + 157 + ], + [ + 299, + 157 + ], + [ + 329, + 165 + ], + [ + 321, + 164 + ] + ] + ] + }, + { + "task_id": "0002049", + "id": "RH20T_cfg5_task_0101_user_0007_scene_0007_cfg_0005_104122061850#216", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the table with a cloth'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002049/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 372, + 77 + ], + [ + 357, + 82 + ], + [ + 364, + 53 + ], + [ + 380, + 100 + ], + [ + 381, + 46 + ], + [ + 397, + 95 + ] + ], + [ + [ + 397, + 86 + ], + [ + 394, + 70 + ], + [ + 382, + 102 + ], + [ + 411, + 71 + ], + [ + 386, + 117 + ], + [ + 413, + 86 + ] + ], + [ + [ + 382, + 104 + ], + [ + 361, + 96 + ], + [ + 390, + 82 + ], + [ + 373, + 126 + ], + [ + 412, + 90 + ], + [ + 394, + 134 + ] + ], + [ + [ + 391, + 70 + ], + [ + 379, + 67 + ], + [ + 373, + 65 + ], + [ + 410, + 76 + ], + [ + 384, + 68 + ], + [ + 421, + 80 + ] + ] + ] + }, + { + "task_id": "0002052", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0003_cfg_0004_104122062295#400", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002052/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 195, + 87 + ], + [ + 202, + 31 + ], + [ + 153, + 65 + ], + [ + 240, + 110 + ], + [ + 149, + 114 + ], + [ + 231, + 159 + ] + ], + [ + [ + 186, + 89 + ], + [ + 155, + 133 + ], + [ + 225, + 102 + ], + [ + 150, + 77 + ], + [ + 254, + 59 + ], + [ + 178, + 39 + ] + ], + [ + [ + 193, + 69 + ], + [ + 143, + 65 + ], + [ + 213, + 120 + ], + [ + 169, + 6 + ], + [ + 265, + 127 + ], + [ + 229, + 7 + ] + ], + [ + [ + 209, + 133 + ], + [ + 213, + 180 + ], + [ + 152, + 155 + ], + [ + 266, + 112 + ], + [ + 153, + 114 + ], + [ + 258, + 74 + ] + ] + ] + }, + { + "task_id": "0002054", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_104422070044#99", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002054/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 421, + 170 + ], + [ + 410, + 184 + ], + [ + 407, + 168 + ], + [ + 436, + 172 + ], + [ + 417, + 155 + ], + [ + 447, + 158 + ] + ], + [ + [ + 432, + 178 + ], + [ + 433, + 163 + ], + [ + 449, + 186 + ], + [ + 415, + 170 + ], + [ + 448, + 200 + ], + [ + 415, + 183 + ] + ], + [ + [ + 423, + 167 + ], + [ + 426, + 154 + ], + [ + 410, + 156 + ], + [ + 436, + 178 + ], + [ + 406, + 170 + ], + [ + 434, + 192 + ] + ], + [ + [ + 443, + 191 + ], + [ + 441, + 175 + ], + [ + 463, + 193 + ], + [ + 424, + 190 + ], + [ + 466, + 210 + ], + [ + 426, + 207 + ] + ] + ] + }, + { + "task_id": "0002055", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_104422070044#149", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002055/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 400, + 151 + ], + [ + 382, + 155 + ], + [ + 407, + 156 + ], + [ + 392, + 146 + ], + [ + 425, + 152 + ], + [ + 410, + 142 + ] + ], + [ + [ + 397, + 163 + ], + [ + 397, + 149 + ], + [ + 413, + 171 + ], + [ + 381, + 156 + ], + [ + 412, + 184 + ], + [ + 382, + 169 + ] + ], + [ + [ + 394, + 187 + ], + [ + 398, + 170 + ], + [ + 400, + 196 + ], + [ + 387, + 177 + ], + [ + 396, + 212 + ], + [ + 383, + 195 + ] + ], + [ + [ + 404, + 172 + ], + [ + 408, + 192 + ], + [ + 423, + 168 + ], + [ + 386, + 177 + ], + [ + 418, + 148 + ], + [ + 382, + 158 + ] + ] + ] + }, + { + "task_id": "0002058", + "id": "RH20T_cfg1_task_0079_user_0015_scene_0003_cfg_0001_819112070044#187", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open the microwave door'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002058/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 353, + 198 + ], + [ + 339, + 204 + ], + [ + 371, + 185 + ], + [ + 333, + 211 + ], + [ + 387, + 178 + ], + [ + 348, + 205 + ] + ], + [ + [ + 384, + 186 + ], + [ + 384, + 189 + ], + [ + 400, + 164 + ], + [ + 368, + 208 + ], + [ + 401, + 160 + ], + [ + 367, + 207 + ] + ], + [ + [ + 372, + 176 + ], + [ + 373, + 152 + ], + [ + 392, + 167 + ], + [ + 350, + 186 + ], + [ + 390, + 189 + ], + [ + 349, + 209 + ] + ], + [ + [ + 381, + 179 + ], + [ + 406, + 191 + ], + [ + 383, + 174 + ], + [ + 378, + 185 + ], + [ + 360, + 162 + ], + [ + 353, + 172 + ] + ] + ] + }, + { + "task_id": "0002059", + "id": "RH20T_cfg1_task_0061_user_0015_scene_0001_cfg_0001_750612070851#527", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002059/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 441, + 215 + ], + [ + 459, + 239 + ], + [ + 413, + 236 + ], + [ + 469, + 193 + ], + [ + 398, + 214 + ], + [ + 451, + 172 + ] + ], + [ + [ + 431, + 185 + ], + [ + 409, + 173 + ], + [ + 439, + 152 + ], + [ + 424, + 217 + ], + [ + 465, + 163 + ], + [ + 448, + 232 + ] + ], + [ + [ + 422, + 195 + ], + [ + 444, + 169 + ], + [ + 448, + 212 + ], + [ + 399, + 180 + ], + [ + 424, + 240 + ], + [ + 378, + 205 + ] + ], + [ + [ + 423, + 188 + ], + [ + 428, + 167 + ], + [ + 392, + 202 + ], + [ + 452, + 174 + ], + [ + 389, + 220 + ], + [ + 445, + 193 + ] + ] + ] + }, + { + "task_id": "0002060", + "id": "RH20T_cfg1_task_0061_user_0015_scene_0001_cfg_0001_750612070851#723", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002060/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 461, + 133 + ], + [ + 428, + 136 + ], + [ + 458, + 146 + ], + [ + 464, + 121 + ], + [ + 496, + 142 + ], + [ + 497, + 118 + ] + ], + [ + [ + 449, + 192 + ], + [ + 447, + 169 + ], + [ + 416, + 197 + ], + [ + 482, + 186 + ], + [ + 420, + 217 + ], + [ + 483, + 207 + ] + ], + [ + [ + 457, + 170 + ], + [ + 463, + 149 + ], + [ + 434, + 189 + ], + [ + 479, + 152 + ], + [ + 430, + 207 + ], + [ + 473, + 172 + ] + ], + [ + [ + 429, + 167 + ], + [ + 435, + 164 + ], + [ + 441, + 137 + ], + [ + 417, + 195 + ], + [ + 435, + 138 + ], + [ + 410, + 199 + ] + ] + ] + }, + { + "task_id": "0002063", + "id": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122061330#129", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the pen into the pen holder'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002063/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 157 + ], + [ + 252, + 150 + ], + [ + 281, + 160 + ], + [ + 267, + 154 + ], + [ + 302, + 167 + ], + [ + 291, + 161 + ] + ], + [ + [ + 276, + 159 + ], + [ + 272, + 149 + ], + [ + 266, + 181 + ], + [ + 287, + 138 + ], + [ + 270, + 189 + ], + [ + 290, + 148 + ] + ], + [ + [ + 264, + 169 + ], + [ + 256, + 185 + ], + [ + 280, + 155 + ], + [ + 249, + 182 + ], + [ + 286, + 140 + ], + [ + 256, + 167 + ] + ], + [ + [ + 259, + 172 + ], + [ + 236, + 163 + ], + [ + 264, + 149 + ], + [ + 254, + 194 + ], + [ + 288, + 159 + ], + [ + 277, + 204 + ] + ] + ] + }, + { + "task_id": "0002064", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_037522062165#821", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002064/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 234, + 73 + ], + [ + 223, + 56 + ], + [ + 222, + 95 + ], + [ + 245, + 51 + ], + [ + 232, + 110 + ], + [ + 254, + 67 + ] + ], + [ + [ + 214, + 66 + ], + [ + 211, + 93 + ], + [ + 225, + 61 + ], + [ + 205, + 70 + ], + [ + 229, + 32 + ], + [ + 208, + 44 + ] + ], + [ + [ + 215, + 89 + ], + [ + 205, + 78 + ], + [ + 191, + 77 + ], + [ + 240, + 100 + ], + [ + 202, + 87 + ], + [ + 248, + 109 + ] + ], + [ + [ + 195, + 56 + ], + [ + 193, + 33 + ], + [ + 169, + 68 + ], + [ + 220, + 44 + ], + [ + 173, + 88 + ], + [ + 221, + 64 + ] + ] + ] + }, + { + "task_id": "0002065", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_037522062165#1379", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002065/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 177, + 251 + ], + [ + 146, + 269 + ], + [ + 184, + 274 + ], + [ + 168, + 226 + ], + [ + 210, + 257 + ], + [ + 198, + 209 + ] + ], + [ + [ + 187, + 295 + ], + [ + 183, + 287 + ], + [ + 206, + 268 + ], + [ + 167, + 324 + ], + [ + 208, + 277 + ], + [ + 171, + 330 + ] + ], + [ + [ + 210, + 301 + ], + [ + 236, + 319 + ], + [ + 190, + 331 + ], + [ + 229, + 273 + ], + [ + 161, + 313 + ], + [ + 202, + 255 + ] + ], + [ + [ + 202, + 291 + ], + [ + 174, + 293 + ], + [ + 197, + 258 + ], + [ + 208, + 324 + ], + [ + 226, + 255 + ], + [ + 237, + 323 + ] + ] + ] + }, + { + "task_id": "0002066", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_037522062165#1714", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002066/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 155, + 177 + ], + [ + 138, + 191 + ], + [ + 175, + 183 + ], + [ + 135, + 171 + ], + [ + 190, + 169 + ], + [ + 151, + 157 + ] + ], + [ + [ + 140, + 178 + ], + [ + 119, + 164 + ], + [ + 128, + 190 + ], + [ + 152, + 167 + ], + [ + 150, + 205 + ], + [ + 172, + 181 + ] + ], + [ + [ + 107, + 168 + ], + [ + 127, + 150 + ], + [ + 118, + 181 + ], + [ + 95, + 152 + ], + [ + 97, + 200 + ], + [ + 72, + 172 + ] + ], + [ + [ + 144, + 163 + ], + [ + 134, + 160 + ], + [ + 123, + 150 + ], + [ + 165, + 177 + ], + [ + 133, + 153 + ], + [ + 173, + 179 + ] + ] + ] + }, + { + "task_id": "0002068", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122063633#356", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002068/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 211, + 206 + ], + [ + 192, + 220 + ], + [ + 195, + 190 + ], + [ + 226, + 220 + ], + [ + 215, + 175 + ], + [ + 244, + 206 + ] + ], + [ + [ + 200, + 188 + ], + [ + 199, + 164 + ], + [ + 210, + 187 + ], + [ + 189, + 189 + ], + [ + 211, + 209 + ], + [ + 191, + 214 + ] + ], + [ + [ + 198, + 185 + ], + [ + 175, + 187 + ], + [ + 206, + 174 + ], + [ + 189, + 198 + ], + [ + 228, + 172 + ], + [ + 214, + 197 + ] + ], + [ + [ + 205, + 174 + ], + [ + 186, + 165 + ], + [ + 193, + 172 + ], + [ + 217, + 176 + ], + [ + 211, + 180 + ], + [ + 235, + 184 + ] + ] + ] + }, + { + "task_id": "0002070", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122063633#529", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002070/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 208, + 155 + ], + [ + 212, + 176 + ], + [ + 217, + 159 + ], + [ + 198, + 151 + ], + [ + 213, + 139 + ], + [ + 194, + 129 + ] + ], + [ + [ + 220, + 172 + ], + [ + 218, + 149 + ], + [ + 227, + 170 + ], + [ + 213, + 174 + ], + [ + 229, + 191 + ], + [ + 215, + 198 + ] + ], + [ + [ + 234, + 159 + ], + [ + 241, + 161 + ], + [ + 234, + 182 + ], + [ + 235, + 136 + ], + [ + 228, + 180 + ], + [ + 228, + 136 + ] + ], + [ + [ + 249, + 187 + ], + [ + 270, + 184 + ], + [ + 244, + 167 + ], + [ + 254, + 206 + ], + [ + 223, + 170 + ], + [ + 234, + 209 + ] + ] + ] + }, + { + "task_id": "0002071", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0008_cfg_0004_104122063550#63", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002071/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 301, + 227 + ], + [ + 316, + 199 + ], + [ + 323, + 241 + ], + [ + 280, + 215 + ], + [ + 308, + 270 + ], + [ + 266, + 242 + ] + ], + [ + [ + 307, + 229 + ], + [ + 309, + 216 + ], + [ + 326, + 256 + ], + [ + 289, + 203 + ], + [ + 322, + 266 + ], + [ + 287, + 216 + ] + ], + [ + [ + 303, + 212 + ], + [ + 322, + 201 + ], + [ + 316, + 240 + ], + [ + 290, + 184 + ], + [ + 297, + 249 + ], + [ + 273, + 195 + ] + ], + [ + [ + 319, + 226 + ], + [ + 345, + 210 + ], + [ + 299, + 203 + ], + [ + 340, + 251 + ], + [ + 272, + 220 + ], + [ + 313, + 270 + ] + ] + ] + }, + { + "task_id": "0002075", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0005_cfg_0005_104422070044#115", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002075/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 430, + 183 + ], + [ + 431, + 168 + ], + [ + 448, + 191 + ], + [ + 413, + 175 + ], + [ + 447, + 205 + ], + [ + 413, + 189 + ] + ], + [ + [ + 442, + 176 + ], + [ + 441, + 169 + ], + [ + 450, + 158 + ], + [ + 434, + 193 + ], + [ + 451, + 166 + ], + [ + 434, + 201 + ] + ], + [ + [ + 411, + 187 + ], + [ + 429, + 188 + ], + [ + 411, + 207 + ], + [ + 411, + 168 + ], + [ + 393, + 207 + ], + [ + 393, + 167 + ] + ], + [ + [ + 408, + 199 + ], + [ + 414, + 181 + ], + [ + 424, + 201 + ], + [ + 392, + 198 + ], + [ + 418, + 218 + ], + [ + 386, + 216 + ] + ] + ] + }, + { + "task_id": "0002077", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0005_cfg_0005_104422070044#183", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002077/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 143 + ], + [ + 364, + 145 + ], + [ + 350, + 158 + ], + [ + 346, + 129 + ], + [ + 334, + 156 + ], + [ + 330, + 128 + ] + ], + [ + [ + 340, + 152 + ], + [ + 350, + 140 + ], + [ + 328, + 141 + ], + [ + 352, + 164 + ], + [ + 318, + 154 + ], + [ + 343, + 177 + ] + ], + [ + [ + 367, + 144 + ], + [ + 361, + 157 + ], + [ + 370, + 157 + ], + [ + 364, + 130 + ], + [ + 376, + 144 + ], + [ + 371, + 118 + ] + ], + [ + [ + 354, + 144 + ], + [ + 354, + 129 + ], + [ + 369, + 150 + ], + [ + 340, + 138 + ], + [ + 369, + 164 + ], + [ + 341, + 152 + ] + ] + ] + }, + { + "task_id": "0002078", + "id": "RH20T_cfg7_task_0007_user_0014_scene_0005_cfg_0007_104122064161#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002078/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 365, + 217 + ], + [ + 389, + 218 + ], + [ + 361, + 217 + ], + [ + 369, + 217 + ], + [ + 338, + 216 + ], + [ + 343, + 216 + ] + ], + [ + [ + 388, + 237 + ], + [ + 380, + 261 + ], + [ + 375, + 228 + ], + [ + 403, + 247 + ], + [ + 383, + 205 + ], + [ + 411, + 222 + ] + ], + [ + [ + 381, + 220 + ], + [ + 385, + 204 + ], + [ + 382, + 241 + ], + [ + 380, + 200 + ], + [ + 378, + 254 + ], + [ + 376, + 215 + ] + ], + [ + [ + 357, + 239 + ], + [ + 376, + 234 + ], + [ + 363, + 217 + ], + [ + 352, + 259 + ], + [ + 343, + 222 + ], + [ + 333, + 265 + ] + ] + ] + }, + { + "task_id": "0002080", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_750612070851#225", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002080/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 196, + 296 + ], + [ + 192, + 275 + ], + [ + 153, + 298 + ], + [ + 239, + 294 + ], + [ + 160, + 316 + ], + [ + 240, + 312 + ] + ], + [ + [ + 185, + 279 + ], + [ + 173, + 283 + ], + [ + 148, + 269 + ], + [ + 222, + 290 + ], + [ + 161, + 266 + ], + [ + 229, + 286 + ] + ], + [ + [ + 232, + 299 + ], + [ + 234, + 337 + ], + [ + 269, + 293 + ], + [ + 194, + 304 + ], + [ + 266, + 257 + ], + [ + 193, + 268 + ] + ], + [ + [ + 222, + 284 + ], + [ + 210, + 328 + ], + [ + 180, + 273 + ], + [ + 261, + 294 + ], + [ + 193, + 230 + ], + [ + 271, + 253 + ] + ] + ] + }, + { + "task_id": "0002082", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_750612070851#508", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002082/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 283, + 292 + ], + [ + 284, + 274 + ], + [ + 242, + 289 + ], + [ + 323, + 295 + ], + [ + 244, + 305 + ], + [ + 319, + 311 + ] + ], + [ + [ + 305, + 323 + ], + [ + 284, + 303 + ], + [ + 323, + 289 + ], + [ + 287, + 357 + ], + [ + 348, + 310 + ], + [ + 310, + 360 + ] + ], + [ + [ + 245, + 310 + ], + [ + 273, + 332 + ], + [ + 215, + 333 + ], + [ + 277, + 287 + ], + [ + 185, + 311 + ], + [ + 248, + 263 + ] + ], + [ + [ + 279, + 274 + ], + [ + 261, + 299 + ], + [ + 267, + 296 + ], + [ + 293, + 250 + ], + [ + 283, + 274 + ], + [ + 308, + 229 + ] + ] + ] + }, + { + "task_id": "0002083", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_750612070851#740", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002083/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 294 + ], + [ + 314, + 332 + ], + [ + 309, + 286 + ], + [ + 301, + 301 + ], + [ + 300, + 248 + ], + [ + 294, + 268 + ] + ], + [ + [ + 319, + 319 + ], + [ + 296, + 286 + ], + [ + 348, + 299 + ], + [ + 287, + 342 + ], + [ + 371, + 331 + ], + [ + 311, + 360 + ] + ], + [ + [ + 297, + 344 + ], + [ + 316, + 311 + ], + [ + 307, + 314 + ], + [ + 284, + 360 + ], + [ + 290, + 343 + ], + [ + 266, + 360 + ] + ], + [ + [ + 319, + 313 + ], + [ + 324, + 296 + ], + [ + 278, + 318 + ], + [ + 358, + 308 + ], + [ + 276, + 333 + ], + [ + 351, + 323 + ] + ] + ] + }, + { + "task_id": "0002086", + "id": "RH20T_cfg4_task_0001_user_0010_scene_0004_cfg_0004_036422060909#339", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002086/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 335, + 93 + ], + [ + 339, + 77 + ], + [ + 313, + 82 + ], + [ + 357, + 104 + ], + [ + 311, + 97 + ], + [ + 353, + 118 + ] + ], + [ + [ + 349, + 103 + ], + [ + 348, + 127 + ], + [ + 362, + 94 + ], + [ + 336, + 111 + ], + [ + 362, + 69 + ], + [ + 337, + 89 + ] + ], + [ + [ + 324, + 109 + ], + [ + 301, + 104 + ], + [ + 331, + 91 + ], + [ + 316, + 128 + ], + [ + 354, + 95 + ], + [ + 341, + 133 + ] + ], + [ + [ + 331, + 81 + ], + [ + 327, + 62 + ], + [ + 337, + 65 + ], + [ + 325, + 99 + ], + [ + 340, + 84 + ], + [ + 330, + 117 + ] + ] + ] + }, + { + "task_id": "0002090", + "id": "RH20T_cfg6_task_0022_user_0014_scene_0006_cfg_0006_104122060811#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Wave the flag'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002090/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 372, + 145 + ], + [ + 379, + 135 + ], + [ + 363, + 147 + ], + [ + 380, + 143 + ], + [ + 356, + 157 + ], + [ + 373, + 153 + ] + ], + [ + [ + 366, + 141 + ], + [ + 375, + 131 + ], + [ + 375, + 142 + ], + [ + 356, + 139 + ], + [ + 366, + 150 + ], + [ + 347, + 148 + ] + ], + [ + [ + 360, + 138 + ], + [ + 372, + 144 + ], + [ + 358, + 141 + ], + [ + 362, + 135 + ], + [ + 345, + 135 + ], + [ + 350, + 129 + ] + ], + [ + [ + 354, + 150 + ], + [ + 360, + 137 + ], + [ + 362, + 156 + ], + [ + 345, + 144 + ], + [ + 356, + 169 + ], + [ + 339, + 157 + ] + ] + ] + }, + { + "task_id": "0002097", + "id": "RH20T_cfg4_task_0127_user_0010_scene_0001_cfg_0004_104122062823#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002097/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 258, + 28 + ], + [ + 257, + 0 + ], + [ + 240, + 30 + ], + [ + 278, + 25 + ], + [ + 241, + 59 + ], + [ + 279, + 57 + ] + ], + [ + [ + 258, + 28 + ], + [ + 276, + 52 + ], + [ + 240, + 36 + ], + [ + 275, + 20 + ], + [ + 221, + 11 + ], + [ + 258, + 0 + ] + ], + [ + [ + 256, + 73 + ], + [ + 276, + 61 + ], + [ + 234, + 72 + ], + [ + 276, + 73 + ], + [ + 216, + 84 + ], + [ + 258, + 83 + ] + ], + [ + [ + 265, + 61 + ], + [ + 277, + 43 + ], + [ + 281, + 50 + ], + [ + 248, + 74 + ], + [ + 270, + 67 + ], + [ + 238, + 91 + ] + ] + ] + }, + { + "task_id": "0002098", + "id": "RH20T_cfg4_task_0127_user_0010_scene_0001_cfg_0004_104122062823#287", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002098/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 110 + ], + [ + 326, + 106 + ], + [ + 313, + 141 + ], + [ + 284, + 78 + ], + [ + 288, + 143 + ], + [ + 260, + 84 + ] + ], + [ + [ + 286, + 93 + ], + [ + 316, + 107 + ], + [ + 304, + 60 + ], + [ + 268, + 125 + ], + [ + 271, + 43 + ], + [ + 235, + 111 + ] + ], + [ + [ + 279, + 78 + ], + [ + 289, + 51 + ], + [ + 267, + 98 + ], + [ + 290, + 59 + ], + [ + 258, + 123 + ], + [ + 281, + 83 + ] + ], + [ + [ + 315, + 110 + ], + [ + 309, + 77 + ], + [ + 331, + 96 + ], + [ + 296, + 125 + ], + [ + 336, + 126 + ], + [ + 303, + 157 + ] + ] + ] + }, + { + "task_id": "0002101", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0002_cfg_0005_105422061350#144", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002101/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 225, + 134 + ], + [ + 196, + 155 + ], + [ + 208, + 110 + ], + [ + 243, + 160 + ], + [ + 237, + 91 + ], + [ + 273, + 139 + ] + ], + [ + [ + 235, + 143 + ], + [ + 215, + 157 + ], + [ + 221, + 113 + ], + [ + 249, + 174 + ], + [ + 243, + 97 + ], + [ + 272, + 161 + ] + ], + [ + [ + 178, + 137 + ], + [ + 142, + 140 + ], + [ + 163, + 149 + ], + [ + 191, + 127 + ], + [ + 203, + 146 + ], + [ + 226, + 124 + ] + ], + [ + [ + 214, + 147 + ], + [ + 211, + 125 + ], + [ + 243, + 131 + ], + [ + 183, + 164 + ], + [ + 244, + 151 + ], + [ + 188, + 183 + ] + ] + ] + }, + { + "task_id": "0002103", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0002_cfg_0005_105422061350#327", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002103/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 193, + 134 + ], + [ + 186, + 113 + ], + [ + 221, + 117 + ], + [ + 163, + 151 + ], + [ + 225, + 136 + ], + [ + 171, + 169 + ] + ], + [ + [ + 174, + 129 + ], + [ + 189, + 118 + ], + [ + 149, + 105 + ], + [ + 198, + 152 + ], + [ + 130, + 115 + ], + [ + 183, + 166 + ] + ], + [ + [ + 206, + 171 + ], + [ + 229, + 171 + ], + [ + 227, + 185 + ], + [ + 186, + 157 + ], + [ + 201, + 186 + ], + [ + 159, + 156 + ] + ], + [ + [ + 206, + 136 + ], + [ + 209, + 124 + ], + [ + 235, + 117 + ], + [ + 178, + 155 + ], + [ + 234, + 130 + ], + [ + 173, + 170 + ] + ] + ] + }, + { + "task_id": "0002104", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0002_cfg_0005_105422061350#425", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002104/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 155, + 177 + ], + [ + 186, + 167 + ], + [ + 179, + 191 + ], + [ + 128, + 160 + ], + [ + 151, + 200 + ], + [ + 97, + 170 + ] + ], + [ + [ + 183, + 184 + ], + [ + 174, + 164 + ], + [ + 215, + 167 + ], + [ + 148, + 201 + ], + [ + 220, + 185 + ], + [ + 158, + 218 + ] + ], + [ + [ + 154, + 169 + ], + [ + 119, + 183 + ], + [ + 145, + 173 + ], + [ + 161, + 165 + ], + [ + 182, + 158 + ], + [ + 193, + 151 + ] + ], + [ + [ + 186, + 184 + ], + [ + 202, + 188 + ], + [ + 177, + 217 + ], + [ + 196, + 152 + ], + [ + 157, + 214 + ], + [ + 179, + 144 + ] + ] + ] + }, + { + "task_id": "0002105", + "id": "RH20T_cfg5_task_0123_user_0007_scene_0007_cfg_0005_104122061850#217", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the knife on the knife rack'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002105/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 415, + 123 + ], + [ + 413, + 106 + ], + [ + 398, + 141 + ], + [ + 430, + 106 + ], + [ + 401, + 155 + ], + [ + 431, + 122 + ] + ], + [ + [ + 411, + 102 + ], + [ + 411, + 114 + ], + [ + 391, + 87 + ], + [ + 430, + 117 + ], + [ + 392, + 77 + ], + [ + 429, + 106 + ] + ], + [ + [ + 416, + 126 + ], + [ + 439, + 128 + ], + [ + 428, + 104 + ], + [ + 404, + 148 + ], + [ + 407, + 103 + ], + [ + 384, + 144 + ] + ], + [ + [ + 430, + 116 + ], + [ + 453, + 123 + ], + [ + 429, + 142 + ], + [ + 431, + 89 + ], + [ + 409, + 135 + ], + [ + 410, + 84 + ] + ] + ] + }, + { + "task_id": "0002108", + "id": "RH20T_cfg3_task_0043_user_0010_scene_0006_cfg_0003_104122062295#200", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002108/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 211, + 178 + ], + [ + 202, + 137 + ], + [ + 162, + 172 + ], + [ + 261, + 184 + ], + [ + 172, + 206 + ], + [ + 265, + 219 + ] + ], + [ + [ + 160, + 139 + ], + [ + 172, + 136 + ], + [ + 137, + 86 + ], + [ + 183, + 190 + ], + [ + 119, + 82 + ], + [ + 171, + 199 + ] + ], + [ + [ + 239, + 136 + ], + [ + 189, + 144 + ], + [ + 247, + 120 + ], + [ + 230, + 157 + ], + [ + 294, + 112 + ], + [ + 287, + 149 + ] + ], + [ + [ + 181, + 206 + ], + [ + 130, + 239 + ], + [ + 213, + 252 + ], + [ + 149, + 160 + ], + [ + 256, + 221 + ], + [ + 196, + 134 + ] + ] + ] + }, + { + "task_id": "0002109", + "id": "RH20T_cfg4_task_0066_user_0015_scene_0001_cfg_0004_104122063550#73", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002109/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 181 + ], + [ + 292, + 183 + ], + [ + 298, + 168 + ], + [ + 329, + 193 + ], + [ + 322, + 165 + ], + [ + 352, + 191 + ] + ], + [ + [ + 320, + 119 + ], + [ + 316, + 102 + ], + [ + 298, + 103 + ], + [ + 343, + 135 + ], + [ + 303, + 119 + ], + [ + 346, + 150 + ] + ], + [ + [ + 295, + 151 + ], + [ + 295, + 138 + ], + [ + 301, + 178 + ], + [ + 289, + 125 + ], + [ + 301, + 187 + ], + [ + 290, + 137 + ] + ], + [ + [ + 289, + 172 + ], + [ + 299, + 200 + ], + [ + 264, + 177 + ], + [ + 316, + 166 + ], + [ + 253, + 149 + ], + [ + 306, + 135 + ] + ] + ] + }, + { + "task_id": "0002111", + "id": "RH20T_cfg1_task_0059_user_0001_scene_0001_cfg_0001_819112071037#202", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002111/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 232, + 74 + ], + [ + 271, + 55 + ], + [ + 267, + 108 + ], + [ + 191, + 34 + ], + [ + 233, + 122 + ], + [ + 156, + 54 + ] + ], + [ + [ + 188, + 82 + ], + [ + 193, + 21 + ], + [ + 161, + 90 + ], + [ + 209, + 76 + ], + [ + 157, + 153 + ], + [ + 204, + 127 + ] + ], + [ + [ + 159, + 65 + ], + [ + 170, + 106 + ], + [ + 200, + 98 + ], + [ + 109, + 24 + ], + [ + 189, + 64 + ], + [ + 103, + 0 + ] + ], + [ + [ + 143, + 68 + ], + [ + 190, + 99 + ], + [ + 96, + 99 + ], + [ + 193, + 33 + ], + [ + 33, + 65 + ], + [ + 137, + 0 + ] + ] + ] + }, + { + "task_id": "0002112", + "id": "RH20T_cfg1_task_0065_user_0014_scene_0004_cfg_0001_750612070851#91", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002112/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 221, + 294 + ], + [ + 217, + 260 + ], + [ + 259, + 281 + ], + [ + 183, + 307 + ], + [ + 266, + 318 + ], + [ + 185, + 345 + ] + ], + [ + [ + 205, + 284 + ], + [ + 200, + 267 + ], + [ + 165, + 289 + ], + [ + 243, + 280 + ], + [ + 172, + 303 + ], + [ + 245, + 295 + ] + ], + [ + [ + 240, + 294 + ], + [ + 238, + 268 + ], + [ + 230, + 262 + ], + [ + 249, + 323 + ], + [ + 232, + 290 + ], + [ + 251, + 354 + ] + ], + [ + [ + 217, + 282 + ], + [ + 245, + 284 + ], + [ + 218, + 246 + ], + [ + 216, + 318 + ], + [ + 192, + 245 + ], + [ + 190, + 314 + ] + ] + ] + }, + { + "task_id": "0002113", + "id": "RH20T_cfg5_task_0085_user_0007_scene_0006_cfg_0005_104422070044#226", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the bubble ring from the assembled bubble ring and ball'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002113/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 350, + 167 + ], + [ + 355, + 180 + ], + [ + 345, + 157 + ], + [ + 354, + 177 + ], + [ + 340, + 145 + ], + [ + 350, + 165 + ] + ], + [ + [ + 357, + 169 + ], + [ + 354, + 158 + ], + [ + 348, + 181 + ], + [ + 365, + 158 + ], + [ + 351, + 192 + ], + [ + 367, + 169 + ] + ], + [ + [ + 351, + 175 + ], + [ + 339, + 164 + ], + [ + 342, + 187 + ], + [ + 361, + 163 + ], + [ + 354, + 198 + ], + [ + 373, + 174 + ] + ], + [ + [ + 351, + 174 + ], + [ + 359, + 159 + ], + [ + 346, + 168 + ], + [ + 356, + 180 + ], + [ + 338, + 183 + ], + [ + 348, + 194 + ] + ] + ] + }, + { + "task_id": "0002117", + "id": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_035622060973#564", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002117/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 267, + 227 + ], + [ + 265, + 211 + ], + [ + 247, + 236 + ], + [ + 287, + 219 + ], + [ + 250, + 251 + ], + [ + 289, + 234 + ] + ], + [ + [ + 264, + 188 + ], + [ + 266, + 206 + ], + [ + 286, + 185 + ], + [ + 243, + 191 + ], + [ + 283, + 169 + ], + [ + 241, + 174 + ] + ], + [ + [ + 267, + 209 + ], + [ + 266, + 196 + ], + [ + 247, + 202 + ], + [ + 287, + 216 + ], + [ + 249, + 215 + ], + [ + 288, + 228 + ] + ], + [ + [ + 271, + 199 + ], + [ + 285, + 184 + ], + [ + 280, + 211 + ], + [ + 261, + 187 + ], + [ + 267, + 226 + ], + [ + 246, + 203 + ] + ] + ] + }, + { + "task_id": "0002119", + "id": "RH20T_cfg6_task_0209_user_0014_scene_0004_cfg_0006_104122061602#74", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slice the carrots'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002119/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 364, + 87 + ], + [ + 383, + 101 + ], + [ + 361, + 92 + ], + [ + 367, + 81 + ], + [ + 343, + 79 + ], + [ + 347, + 67 + ] + ], + [ + [ + 367, + 100 + ], + [ + 370, + 121 + ], + [ + 345, + 103 + ], + [ + 388, + 97 + ], + [ + 342, + 81 + ], + [ + 384, + 75 + ] + ], + [ + [ + 381, + 88 + ], + [ + 388, + 67 + ], + [ + 389, + 100 + ], + [ + 372, + 76 + ], + [ + 382, + 119 + ], + [ + 366, + 95 + ] + ], + [ + [ + 366, + 75 + ], + [ + 353, + 57 + ], + [ + 347, + 86 + ], + [ + 387, + 64 + ], + [ + 359, + 103 + ], + [ + 399, + 82 + ] + ] + ] + }, + { + "task_id": "0002120", + "id": "RH20T_cfg6_task_0209_user_0014_scene_0004_cfg_0006_104122061602#110", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slice the carrots'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002120/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 383, + 87 + ], + [ + 405, + 85 + ], + [ + 380, + 95 + ], + [ + 387, + 78 + ], + [ + 358, + 97 + ], + [ + 363, + 80 + ] + ], + [ + [ + 386, + 92 + ], + [ + 365, + 101 + ], + [ + 387, + 110 + ], + [ + 384, + 72 + ], + [ + 406, + 101 + ], + [ + 405, + 63 + ] + ], + [ + [ + 377, + 100 + ], + [ + 382, + 80 + ], + [ + 386, + 111 + ], + [ + 369, + 89 + ], + [ + 381, + 131 + ], + [ + 365, + 108 + ] + ], + [ + [ + 367, + 120 + ], + [ + 369, + 140 + ], + [ + 359, + 111 + ], + [ + 374, + 129 + ], + [ + 358, + 91 + ], + [ + 373, + 110 + ] + ] + ] + }, + { + "task_id": "0002121", + "id": "RH20T_cfg2_task_0004_user_0006_scene_0003_cfg_0002_104122063678#288", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002121/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 353 + ], + [ + 323, + 339 + ], + [ + 309, + 360 + ], + [ + 341, + 328 + ], + [ + 313, + 360 + ], + [ + 343, + 342 + ] + ], + [ + [ + 306, + 360 + ], + [ + 330, + 360 + ], + [ + 324, + 342 + ], + [ + 287, + 360 + ], + [ + 302, + 330 + ], + [ + 264, + 360 + ] + ], + [ + [ + 322, + 359 + ], + [ + 348, + 342 + ], + [ + 325, + 360 + ], + [ + 321, + 352 + ], + [ + 297, + 360 + ], + [ + 296, + 360 + ] + ], + [ + [ + 306, + 360 + ], + [ + 292, + 360 + ], + [ + 304, + 351 + ], + [ + 308, + 360 + ], + [ + 317, + 327 + ], + [ + 320, + 360 + ] + ] + ] + }, + { + "task_id": "0002122", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061018#287", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002122/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 294, + 131 + ], + [ + 284, + 102 + ], + [ + 287, + 127 + ], + [ + 301, + 135 + ], + [ + 296, + 152 + ], + [ + 311, + 164 + ] + ], + [ + [ + 257, + 112 + ], + [ + 276, + 118 + ], + [ + 279, + 106 + ], + [ + 233, + 119 + ], + [ + 261, + 102 + ], + [ + 215, + 114 + ] + ], + [ + [ + 314, + 133 + ], + [ + 338, + 145 + ], + [ + 303, + 126 + ], + [ + 327, + 142 + ], + [ + 279, + 114 + ], + [ + 301, + 129 + ] + ], + [ + [ + 314, + 137 + ], + [ + 320, + 127 + ], + [ + 320, + 164 + ], + [ + 308, + 108 + ], + [ + 314, + 176 + ], + [ + 302, + 118 + ] + ] + ] + }, + { + "task_id": "0002123", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061018#367", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002123/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 132 + ], + [ + 295, + 103 + ], + [ + 301, + 127 + ], + [ + 310, + 137 + ], + [ + 310, + 153 + ], + [ + 320, + 166 + ] + ], + [ + [ + 333, + 147 + ], + [ + 356, + 140 + ], + [ + 325, + 162 + ], + [ + 343, + 130 + ], + [ + 301, + 169 + ], + [ + 317, + 137 + ] + ], + [ + [ + 277, + 123 + ], + [ + 249, + 133 + ], + [ + 272, + 110 + ], + [ + 280, + 135 + ], + [ + 301, + 99 + ], + [ + 306, + 125 + ] + ], + [ + [ + 291, + 154 + ], + [ + 285, + 136 + ], + [ + 321, + 148 + ], + [ + 259, + 160 + ], + [ + 329, + 168 + ], + [ + 264, + 181 + ] + ] + ] + }, + { + "task_id": "0002124", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061018#442", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002124/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 317, + 88 + ], + [ + 337, + 65 + ], + [ + 302, + 72 + ], + [ + 331, + 102 + ], + [ + 281, + 97 + ], + [ + 311, + 124 + ] + ], + [ + [ + 309, + 84 + ], + [ + 294, + 67 + ], + [ + 333, + 69 + ], + [ + 285, + 98 + ], + [ + 345, + 84 + ], + [ + 298, + 112 + ] + ], + [ + [ + 306, + 104 + ], + [ + 296, + 76 + ], + [ + 299, + 101 + ], + [ + 313, + 107 + ], + [ + 308, + 126 + ], + [ + 323, + 135 + ] + ], + [ + [ + 285, + 116 + ], + [ + 313, + 109 + ], + [ + 292, + 143 + ], + [ + 279, + 88 + ], + [ + 264, + 150 + ], + [ + 251, + 94 + ] + ] + ] + }, + { + "task_id": "0002125", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061018#553", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002125/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 312, + 90 + ], + [ + 302, + 116 + ], + [ + 334, + 96 + ], + [ + 288, + 82 + ], + [ + 345, + 70 + ], + [ + 299, + 53 + ] + ], + [ + [ + 288, + 105 + ], + [ + 272, + 121 + ], + [ + 276, + 118 + ], + [ + 299, + 92 + ], + [ + 294, + 100 + ], + [ + 316, + 75 + ] + ], + [ + [ + 308, + 96 + ], + [ + 337, + 96 + ], + [ + 305, + 123 + ], + [ + 311, + 69 + ], + [ + 277, + 122 + ], + [ + 284, + 69 + ] + ], + [ + [ + 310, + 105 + ], + [ + 300, + 78 + ], + [ + 303, + 101 + ], + [ + 317, + 110 + ], + [ + 311, + 125 + ], + [ + 327, + 138 + ] + ] + ] + }, + { + "task_id": "0002126", + "id": "RH20T_cfg5_task_0010_user_0010_scene_0007_cfg_0005_105422061350#307", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002126/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 94 + ], + [ + 335, + 79 + ], + [ + 338, + 80 + ], + [ + 313, + 110 + ], + [ + 330, + 94 + ], + [ + 306, + 124 + ] + ], + [ + [ + 317, + 111 + ], + [ + 337, + 128 + ], + [ + 308, + 123 + ], + [ + 328, + 99 + ], + [ + 288, + 106 + ], + [ + 307, + 81 + ] + ], + [ + [ + 340, + 103 + ], + [ + 318, + 99 + ], + [ + 352, + 89 + ], + [ + 327, + 118 + ], + [ + 376, + 92 + ], + [ + 352, + 122 + ] + ], + [ + [ + 323, + 93 + ], + [ + 341, + 80 + ], + [ + 332, + 114 + ], + [ + 314, + 73 + ], + [ + 315, + 126 + ], + [ + 298, + 86 + ] + ] + ] + }, + { + "task_id": "0002128", + "id": "RH20T_cfg7_task_0020_user_0014_scene_0010_cfg_0007_104122061018#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002128/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 308, + 154 + ], + [ + 335, + 156 + ], + [ + 309, + 126 + ], + [ + 306, + 183 + ], + [ + 283, + 125 + ], + [ + 279, + 181 + ] + ], + [ + [ + 328, + 160 + ], + [ + 350, + 174 + ], + [ + 346, + 147 + ], + [ + 308, + 174 + ], + [ + 326, + 135 + ], + [ + 287, + 159 + ] + ], + [ + [ + 312, + 166 + ], + [ + 301, + 140 + ], + [ + 306, + 178 + ], + [ + 319, + 153 + ], + [ + 317, + 203 + ], + [ + 331, + 181 + ] + ], + [ + [ + 309, + 166 + ], + [ + 303, + 166 + ], + [ + 312, + 137 + ], + [ + 305, + 195 + ], + [ + 317, + 139 + ], + [ + 310, + 194 + ] + ] + ] + }, + { + "task_id": "0002130", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122061018#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002130/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 291, + 137 + ], + [ + 277, + 148 + ], + [ + 267, + 139 + ], + [ + 314, + 135 + ], + [ + 282, + 126 + ], + [ + 330, + 124 + ] + ], + [ + [ + 286, + 106 + ], + [ + 284, + 78 + ], + [ + 274, + 96 + ], + [ + 299, + 116 + ], + [ + 276, + 121 + ], + [ + 301, + 143 + ] + ], + [ + [ + 280, + 98 + ], + [ + 257, + 101 + ], + [ + 271, + 68 + ], + [ + 288, + 127 + ], + [ + 296, + 62 + ], + [ + 312, + 124 + ] + ], + [ + [ + 309, + 118 + ], + [ + 296, + 94 + ], + [ + 333, + 104 + ], + [ + 285, + 132 + ], + [ + 346, + 129 + ], + [ + 298, + 156 + ] + ] + ] + }, + { + "task_id": "0002131", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122061018#351", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002131/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 298, + 100 + ], + [ + 272, + 99 + ], + [ + 307, + 111 + ], + [ + 290, + 89 + ], + [ + 333, + 112 + ], + [ + 314, + 90 + ] + ], + [ + [ + 292, + 80 + ], + [ + 290, + 52 + ], + [ + 281, + 74 + ], + [ + 304, + 86 + ], + [ + 283, + 100 + ], + [ + 306, + 114 + ] + ], + [ + [ + 275, + 67 + ], + [ + 288, + 91 + ], + [ + 260, + 80 + ], + [ + 289, + 56 + ], + [ + 246, + 55 + ], + [ + 276, + 32 + ] + ], + [ + [ + 279, + 102 + ], + [ + 301, + 98 + ], + [ + 291, + 95 + ], + [ + 268, + 108 + ], + [ + 267, + 98 + ], + [ + 246, + 112 + ] + ] + ] + }, + { + "task_id": "0002133", + "id": "RH20T_cfg5_task_0001_user_0010_scene_0003_cfg_0005_104122063678#282", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002133/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 198 + ], + [ + 365, + 180 + ], + [ + 387, + 184 + ], + [ + 354, + 213 + ], + [ + 393, + 201 + ], + [ + 360, + 231 + ] + ], + [ + [ + 369, + 200 + ], + [ + 361, + 204 + ], + [ + 349, + 188 + ], + [ + 391, + 214 + ], + [ + 357, + 185 + ], + [ + 398, + 210 + ] + ], + [ + [ + 369, + 202 + ], + [ + 378, + 186 + ], + [ + 376, + 223 + ], + [ + 361, + 181 + ], + [ + 367, + 238 + ], + [ + 353, + 196 + ] + ], + [ + [ + 374, + 197 + ], + [ + 348, + 198 + ], + [ + 377, + 172 + ], + [ + 371, + 221 + ], + [ + 404, + 172 + ], + [ + 397, + 221 + ] + ] + ] + }, + { + "task_id": "0002135", + "id": "RH20T_cfg3_task_0070_user_0010_scene_0004_cfg_0003_104422070011#121", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002135/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 333, + 191 + ], + [ + 346, + 212 + ], + [ + 355, + 180 + ], + [ + 311, + 201 + ], + [ + 342, + 159 + ], + [ + 299, + 180 + ] + ], + [ + [ + 340, + 188 + ], + [ + 353, + 189 + ], + [ + 317, + 193 + ], + [ + 362, + 183 + ], + [ + 307, + 192 + ], + [ + 351, + 182 + ] + ], + [ + [ + 331, + 186 + ], + [ + 347, + 202 + ], + [ + 312, + 199 + ], + [ + 351, + 173 + ], + [ + 295, + 182 + ], + [ + 334, + 155 + ] + ], + [ + [ + 360, + 200 + ], + [ + 353, + 187 + ], + [ + 355, + 220 + ], + [ + 365, + 179 + ], + [ + 362, + 235 + ], + [ + 373, + 192 + ] + ] + ] + }, + { + "task_id": "0002140", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_104122063678#537", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002140/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 220 + ], + [ + 321, + 239 + ], + [ + 340, + 215 + ], + [ + 290, + 224 + ], + [ + 333, + 193 + ], + [ + 283, + 204 + ] + ], + [ + [ + 309, + 226 + ], + [ + 296, + 244 + ], + [ + 310, + 239 + ], + [ + 308, + 212 + ], + [ + 321, + 223 + ], + [ + 321, + 195 + ] + ], + [ + [ + 310, + 180 + ], + [ + 286, + 183 + ], + [ + 314, + 190 + ], + [ + 305, + 169 + ], + [ + 338, + 187 + ], + [ + 331, + 166 + ] + ], + [ + [ + 302, + 209 + ], + [ + 307, + 193 + ], + [ + 277, + 208 + ], + [ + 326, + 210 + ], + [ + 273, + 223 + ], + [ + 321, + 225 + ] + ] + ] + }, + { + "task_id": "0002143", + "id": "RH20T_cfg5_task_0048_user_0010_scene_0010_cfg_0005_104422070044#85", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002143/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 425, + 119 + ], + [ + 440, + 108 + ], + [ + 437, + 137 + ], + [ + 413, + 102 + ], + [ + 421, + 148 + ], + [ + 397, + 113 + ] + ], + [ + [ + 403, + 144 + ], + [ + 396, + 147 + ], + [ + 390, + 160 + ], + [ + 415, + 129 + ], + [ + 397, + 158 + ], + [ + 423, + 125 + ] + ], + [ + [ + 437, + 127 + ], + [ + 442, + 148 + ], + [ + 420, + 132 + ], + [ + 455, + 123 + ], + [ + 415, + 112 + ], + [ + 449, + 101 + ] + ], + [ + [ + 426, + 146 + ], + [ + 421, + 135 + ], + [ + 445, + 151 + ], + [ + 408, + 140 + ], + [ + 450, + 161 + ], + [ + 413, + 150 + ] + ] + ] + }, + { + "task_id": "0002144", + "id": "RH20T_cfg4_task_0026_user_0016_scene_0001_cfg_0004_104122062823#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push and close the drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002144/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 225, + 34 + ], + [ + 191, + 43 + ], + [ + 211, + 1 + ], + [ + 239, + 66 + ], + [ + 244, + 0 + ], + [ + 270, + 56 + ] + ], + [ + [ + 232, + 36 + ], + [ + 233, + 53 + ], + [ + 196, + 31 + ], + [ + 268, + 41 + ], + [ + 198, + 17 + ], + [ + 266, + 27 + ] + ], + [ + [ + 239, + 0 + ], + [ + 235, + 0 + ], + [ + 256, + 17 + ], + [ + 224, + 0 + ], + [ + 259, + 48 + ], + [ + 228, + 13 + ] + ], + [ + [ + 229, + 0 + ], + [ + 228, + 0 + ], + [ + 195, + 0 + ], + [ + 260, + 0 + ], + [ + 194, + 0 + ], + [ + 264, + 0 + ] + ] + ] + }, + { + "task_id": "0002146", + "id": "RH20T_cfg1_task_0051_user_0015_scene_0002_cfg_0001_819112071037#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the dish off the dish rack'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002146/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 259, + 39 + ], + [ + 230, + 0 + ], + [ + 203, + 80 + ], + [ + 304, + 5 + ], + [ + 234, + 149 + ], + [ + 321, + 70 + ] + ], + [ + [ + 246, + 9 + ], + [ + 217, + 81 + ], + [ + 170, + 0 + ], + [ + 312, + 43 + ], + [ + 201, + 0 + ], + [ + 341, + 0 + ] + ], + [ + [ + 294, + 66 + ], + [ + 364, + 41 + ], + [ + 350, + 112 + ], + [ + 222, + 7 + ], + [ + 296, + 124 + ], + [ + 172, + 36 + ] + ], + [ + [ + 296, + 34 + ], + [ + 273, + 0 + ], + [ + 267, + 81 + ], + [ + 332, + 0 + ], + [ + 289, + 133 + ], + [ + 363, + 32 + ] + ] + ] + }, + { + "task_id": "0002151", + "id": "RH20T_cfg1_task_0015_user_0008_scene_0008_cfg_0001_819112070044#201", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002151/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 170 + ], + [ + 347, + 154 + ], + [ + 331, + 175 + ], + [ + 367, + 163 + ], + [ + 332, + 191 + ], + [ + 369, + 179 + ] + ], + [ + [ + 345, + 164 + ], + [ + 344, + 178 + ], + [ + 340, + 151 + ], + [ + 351, + 176 + ], + [ + 341, + 138 + ], + [ + 352, + 163 + ] + ], + [ + [ + 340, + 177 + ], + [ + 324, + 170 + ], + [ + 343, + 162 + ], + [ + 337, + 190 + ], + [ + 360, + 169 + ], + [ + 353, + 197 + ] + ], + [ + [ + 354, + 173 + ], + [ + 359, + 159 + ], + [ + 336, + 171 + ], + [ + 372, + 175 + ], + [ + 332, + 184 + ], + [ + 367, + 188 + ] + ] + ] + }, + { + "task_id": "0002152", + "id": "RH20T_cfg1_task_0090_user_0007_scene_0001_cfg_0001_038522063145#164", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch an object with obstacles'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002152/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 484, + 189 + ], + [ + 507, + 185 + ], + [ + 494, + 180 + ], + [ + 475, + 198 + ], + [ + 469, + 185 + ], + [ + 452, + 203 + ] + ], + [ + [ + 489, + 224 + ], + [ + 478, + 236 + ], + [ + 512, + 241 + ], + [ + 467, + 207 + ], + [ + 522, + 229 + ], + [ + 478, + 197 + ] + ], + [ + [ + 500, + 206 + ], + [ + 510, + 188 + ], + [ + 523, + 222 + ], + [ + 477, + 191 + ], + [ + 513, + 238 + ], + [ + 469, + 208 + ] + ], + [ + [ + 462, + 211 + ], + [ + 464, + 189 + ], + [ + 437, + 205 + ], + [ + 487, + 218 + ], + [ + 435, + 228 + ], + [ + 486, + 241 + ] + ] + ] + }, + { + "task_id": "0002154", + "id": "RH20T_cfg5_task_0001_user_0020_scene_0001_cfg_0005_104422070044#188", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002154/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 410, + 168 + ], + [ + 419, + 155 + ], + [ + 425, + 181 + ], + [ + 395, + 156 + ], + [ + 415, + 195 + ], + [ + 385, + 169 + ] + ], + [ + [ + 422, + 187 + ], + [ + 436, + 186 + ], + [ + 423, + 207 + ], + [ + 421, + 168 + ], + [ + 411, + 208 + ], + [ + 409, + 170 + ] + ], + [ + [ + 416, + 162 + ], + [ + 423, + 177 + ], + [ + 429, + 150 + ], + [ + 405, + 174 + ], + [ + 422, + 136 + ], + [ + 399, + 160 + ] + ], + [ + [ + 413, + 164 + ], + [ + 412, + 151 + ], + [ + 430, + 171 + ], + [ + 396, + 158 + ], + [ + 430, + 184 + ], + [ + 397, + 170 + ] + ] + ] + }, + { + "task_id": "0002155", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_104122061330#204", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002155/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 302, + 203 + ], + [ + 290, + 214 + ], + [ + 310, + 230 + ], + [ + 294, + 175 + ], + [ + 320, + 219 + ], + [ + 305, + 167 + ] + ], + [ + [ + 310, + 214 + ], + [ + 302, + 203 + ], + [ + 283, + 212 + ], + [ + 337, + 215 + ], + [ + 291, + 222 + ], + [ + 343, + 224 + ] + ], + [ + [ + 316, + 197 + ], + [ + 293, + 182 + ], + [ + 333, + 171 + ], + [ + 300, + 222 + ], + [ + 354, + 186 + ], + [ + 322, + 235 + ] + ], + [ + [ + 315, + 229 + ], + [ + 322, + 214 + ], + [ + 310, + 254 + ], + [ + 321, + 207 + ], + [ + 303, + 268 + ], + [ + 314, + 221 + ] + ] + ] + }, + { + "task_id": "0002156", + "id": "RH20T_cfg4_task_0071_user_0007_scene_0010_cfg_0004_104122062823#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002156/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 225, + 152 + ], + [ + 252, + 146 + ], + [ + 213, + 179 + ], + [ + 236, + 127 + ], + [ + 188, + 184 + ], + [ + 212, + 133 + ] + ], + [ + [ + 237, + 133 + ], + [ + 234, + 163 + ], + [ + 264, + 138 + ], + [ + 211, + 127 + ], + [ + 267, + 106 + ], + [ + 213, + 97 + ] + ], + [ + [ + 208, + 110 + ], + [ + 189, + 105 + ], + [ + 195, + 143 + ], + [ + 220, + 77 + ], + [ + 215, + 150 + ], + [ + 242, + 80 + ] + ], + [ + [ + 228, + 132 + ], + [ + 217, + 105 + ], + [ + 256, + 115 + ], + [ + 198, + 150 + ], + [ + 265, + 140 + ], + [ + 209, + 174 + ] + ] + ] + }, + { + "task_id": "0002157", + "id": "RH20T_cfg4_task_0071_user_0007_scene_0010_cfg_0004_104122062823#213", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002157/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 238, + 138 + ], + [ + 216, + 133 + ], + [ + 258, + 114 + ], + [ + 217, + 163 + ], + [ + 282, + 118 + ], + [ + 240, + 170 + ] + ], + [ + [ + 220, + 128 + ], + [ + 201, + 108 + ], + [ + 193, + 148 + ], + [ + 246, + 109 + ], + [ + 212, + 170 + ], + [ + 266, + 129 + ] + ], + [ + [ + 227, + 119 + ], + [ + 247, + 127 + ], + [ + 238, + 86 + ], + [ + 216, + 150 + ], + [ + 219, + 80 + ], + [ + 199, + 141 + ] + ], + [ + [ + 218, + 138 + ], + [ + 207, + 110 + ], + [ + 249, + 122 + ], + [ + 186, + 155 + ], + [ + 257, + 147 + ], + [ + 198, + 180 + ] + ] + ] + }, + { + "task_id": "0002158", + "id": "RH20T_cfg3_task_0005_user_0016_scene_0003_cfg_0003_104122062823#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002158/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 296, + 99 + ], + [ + 314, + 85 + ], + [ + 274, + 90 + ], + [ + 319, + 110 + ], + [ + 254, + 104 + ], + [ + 300, + 126 + ] + ], + [ + [ + 282, + 39 + ], + [ + 288, + 70 + ], + [ + 261, + 39 + ], + [ + 305, + 39 + ], + [ + 253, + 8 + ], + [ + 298, + 5 + ] + ], + [ + [ + 289, + 82 + ], + [ + 286, + 55 + ], + [ + 318, + 88 + ], + [ + 261, + 77 + ], + [ + 320, + 113 + ], + [ + 266, + 101 + ] + ], + [ + [ + 292, + 96 + ], + [ + 269, + 102 + ], + [ + 278, + 117 + ], + [ + 306, + 74 + ], + [ + 299, + 111 + ], + [ + 327, + 69 + ] + ] + ] + }, + { + "task_id": "0002165", + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_104122063678#903", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002165/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 453, + 335 + ], + [ + 452, + 321 + ], + [ + 448, + 359 + ], + [ + 458, + 310 + ], + [ + 450, + 360 + ], + [ + 460, + 324 + ] + ], + [ + [ + 406, + 337 + ], + [ + 398, + 354 + ], + [ + 398, + 357 + ], + [ + 414, + 315 + ], + [ + 406, + 341 + ], + [ + 422, + 301 + ] + ], + [ + [ + 421, + 329 + ], + [ + 426, + 351 + ], + [ + 401, + 343 + ], + [ + 441, + 315 + ], + [ + 397, + 323 + ], + [ + 435, + 296 + ] + ], + [ + [ + 428, + 319 + ], + [ + 432, + 304 + ], + [ + 401, + 321 + ], + [ + 455, + 318 + ], + [ + 398, + 335 + ], + [ + 449, + 332 + ] + ] + ] + }, + { + "task_id": "0002166", + "id": "RH20T_cfg5_task_0047_user_0007_scene_0009_cfg_0005_104122061850#281", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002166/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 412, + 74 + ], + [ + 424, + 98 + ], + [ + 389, + 85 + ], + [ + 434, + 63 + ], + [ + 377, + 61 + ], + [ + 422, + 39 + ] + ], + [ + [ + 455, + 71 + ], + [ + 428, + 69 + ], + [ + 458, + 60 + ], + [ + 452, + 83 + ], + [ + 485, + 63 + ], + [ + 481, + 86 + ] + ], + [ + [ + 460, + 58 + ], + [ + 476, + 79 + ], + [ + 477, + 45 + ], + [ + 446, + 70 + ], + [ + 461, + 22 + ], + [ + 431, + 49 + ] + ], + [ + [ + 429, + 53 + ], + [ + 429, + 36 + ], + [ + 409, + 70 + ], + [ + 449, + 37 + ], + [ + 410, + 85 + ], + [ + 448, + 54 + ] + ] + ] + }, + { + "task_id": "0002169", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_038522062547#606", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002169/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 459, + 82 + ], + [ + 477, + 65 + ], + [ + 442, + 65 + ], + [ + 476, + 100 + ], + [ + 424, + 83 + ], + [ + 457, + 117 + ] + ], + [ + [ + 430, + 109 + ], + [ + 449, + 126 + ], + [ + 444, + 95 + ], + [ + 417, + 123 + ], + [ + 425, + 77 + ], + [ + 400, + 107 + ] + ], + [ + [ + 453, + 95 + ], + [ + 459, + 81 + ], + [ + 455, + 117 + ], + [ + 450, + 73 + ], + [ + 449, + 129 + ], + [ + 445, + 86 + ] + ], + [ + [ + 468, + 87 + ], + [ + 482, + 63 + ], + [ + 471, + 90 + ], + [ + 466, + 82 + ], + [ + 458, + 112 + ], + [ + 452, + 106 + ] + ] + ] + }, + { + "task_id": "0002172", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122061602#315", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002172/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 447, + 158 + ], + [ + 471, + 162 + ], + [ + 446, + 179 + ], + [ + 447, + 138 + ], + [ + 422, + 175 + ], + [ + 424, + 134 + ] + ], + [ + [ + 439, + 158 + ], + [ + 455, + 170 + ], + [ + 437, + 177 + ], + [ + 443, + 137 + ], + [ + 423, + 165 + ], + [ + 428, + 126 + ] + ], + [ + [ + 436, + 156 + ], + [ + 452, + 140 + ], + [ + 420, + 143 + ], + [ + 451, + 169 + ], + [ + 404, + 159 + ], + [ + 435, + 185 + ] + ], + [ + [ + 444, + 144 + ], + [ + 462, + 135 + ], + [ + 454, + 165 + ], + [ + 434, + 123 + ], + [ + 436, + 173 + ], + [ + 418, + 133 + ] + ] + ] + }, + { + "task_id": "0002173", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122061602#468", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002173/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 469, + 168 + ], + [ + 465, + 146 + ], + [ + 488, + 174 + ], + [ + 451, + 162 + ], + [ + 491, + 196 + ], + [ + 455, + 182 + ] + ], + [ + [ + 498, + 177 + ], + [ + 485, + 197 + ], + [ + 482, + 178 + ], + [ + 512, + 177 + ], + [ + 496, + 157 + ], + [ + 525, + 157 + ] + ], + [ + [ + 472, + 186 + ], + [ + 494, + 175 + ], + [ + 464, + 192 + ], + [ + 481, + 180 + ], + [ + 444, + 203 + ], + [ + 458, + 192 + ] + ], + [ + [ + 480, + 154 + ], + [ + 489, + 131 + ], + [ + 504, + 163 + ], + [ + 456, + 146 + ], + [ + 495, + 186 + ], + [ + 448, + 168 + ] + ] + ] + }, + { + "task_id": "0002174", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122061602#569", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002174/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 414, + 166 + ], + [ + 435, + 169 + ], + [ + 407, + 187 + ], + [ + 423, + 145 + ], + [ + 385, + 184 + ], + [ + 401, + 142 + ] + ], + [ + [ + 417, + 174 + ], + [ + 419, + 194 + ], + [ + 411, + 163 + ], + [ + 423, + 187 + ], + [ + 409, + 143 + ], + [ + 421, + 165 + ] + ], + [ + [ + 445, + 177 + ], + [ + 457, + 170 + ], + [ + 456, + 200 + ], + [ + 433, + 155 + ], + [ + 445, + 206 + ], + [ + 423, + 163 + ] + ], + [ + [ + 435, + 165 + ], + [ + 433, + 147 + ], + [ + 416, + 176 + ], + [ + 454, + 153 + ], + [ + 418, + 192 + ], + [ + 455, + 170 + ] + ] + ] + }, + { + "task_id": "0002176", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0010_cfg_0005_037522062165#147", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002176/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 148, + 265 + ], + [ + 173, + 239 + ], + [ + 135, + 249 + ], + [ + 162, + 284 + ], + [ + 110, + 274 + ], + [ + 135, + 312 + ] + ], + [ + [ + 180, + 268 + ], + [ + 171, + 266 + ], + [ + 210, + 275 + ], + [ + 148, + 260 + ], + [ + 221, + 278 + ], + [ + 156, + 261 + ] + ], + [ + [ + 131, + 226 + ], + [ + 158, + 207 + ], + [ + 119, + 231 + ], + [ + 140, + 221 + ], + [ + 90, + 251 + ], + [ + 114, + 239 + ] + ], + [ + [ + 153, + 254 + ], + [ + 145, + 238 + ], + [ + 129, + 239 + ], + [ + 178, + 269 + ], + [ + 137, + 254 + ], + [ + 184, + 282 + ] + ] + ] + }, + { + "task_id": "0002177", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0010_cfg_0005_037522062165#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002177/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 217, + 291 + ], + [ + 199, + 316 + ], + [ + 237, + 284 + ], + [ + 197, + 297 + ], + [ + 253, + 262 + ], + [ + 214, + 275 + ] + ], + [ + [ + 211, + 250 + ], + [ + 185, + 268 + ], + [ + 229, + 275 + ], + [ + 193, + 224 + ], + [ + 254, + 256 + ], + [ + 219, + 206 + ] + ], + [ + [ + 181, + 280 + ], + [ + 192, + 263 + ], + [ + 179, + 310 + ], + [ + 184, + 249 + ], + [ + 166, + 332 + ], + [ + 170, + 267 + ] + ], + [ + [ + 201, + 273 + ], + [ + 197, + 260 + ], + [ + 184, + 250 + ], + [ + 219, + 299 + ], + [ + 189, + 263 + ], + [ + 222, + 310 + ] + ] + ] + }, + { + "task_id": "0002179", + "id": "RH20T_cfg6_task_0104_user_0020_scene_0004_cfg_0006_104122061330#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate back after holding it down'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002179/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 376, + 107 + ], + [ + 397, + 114 + ], + [ + 379, + 109 + ], + [ + 372, + 104 + ], + [ + 359, + 102 + ], + [ + 349, + 96 + ] + ], + [ + [ + 386, + 105 + ], + [ + 380, + 125 + ], + [ + 406, + 119 + ], + [ + 368, + 92 + ], + [ + 414, + 97 + ], + [ + 374, + 70 + ] + ], + [ + [ + 371, + 106 + ], + [ + 360, + 118 + ], + [ + 367, + 125 + ], + [ + 376, + 86 + ], + [ + 377, + 114 + ], + [ + 386, + 75 + ] + ], + [ + [ + 364, + 88 + ], + [ + 358, + 77 + ], + [ + 381, + 101 + ], + [ + 346, + 74 + ], + [ + 388, + 114 + ], + [ + 351, + 87 + ] + ] + ] + }, + { + "task_id": "0002180", + "id": "RH20T_cfg1_task_0007_user_0015_scene_0002_cfg_0001_819112070044#462", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002180/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 369, + 189 + ], + [ + 361, + 177 + ], + [ + 386, + 182 + ], + [ + 352, + 196 + ], + [ + 394, + 193 + ], + [ + 360, + 207 + ] + ], + [ + [ + 387, + 189 + ], + [ + 377, + 173 + ], + [ + 390, + 192 + ], + [ + 385, + 186 + ], + [ + 400, + 208 + ], + [ + 395, + 203 + ] + ], + [ + [ + 380, + 180 + ], + [ + 386, + 165 + ], + [ + 397, + 179 + ], + [ + 362, + 181 + ], + [ + 392, + 192 + ], + [ + 357, + 195 + ] + ], + [ + [ + 383, + 183 + ], + [ + 393, + 166 + ], + [ + 365, + 174 + ], + [ + 401, + 192 + ], + [ + 355, + 191 + ], + [ + 391, + 208 + ] + ] + ] + }, + { + "task_id": "0002184", + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_750612070851#143", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002184/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 343, + 143 + ], + [ + 341, + 121 + ], + [ + 309, + 158 + ], + [ + 377, + 129 + ], + [ + 313, + 176 + ], + [ + 377, + 149 + ] + ], + [ + [ + 348, + 165 + ], + [ + 330, + 155 + ], + [ + 375, + 140 + ], + [ + 321, + 191 + ], + [ + 398, + 150 + ], + [ + 339, + 205 + ] + ], + [ + [ + 322, + 167 + ], + [ + 321, + 193 + ], + [ + 347, + 188 + ], + [ + 299, + 147 + ], + [ + 350, + 158 + ], + [ + 299, + 117 + ] + ], + [ + [ + 303, + 150 + ], + [ + 321, + 149 + ], + [ + 315, + 182 + ], + [ + 291, + 118 + ], + [ + 299, + 182 + ], + [ + 275, + 121 + ] + ] + ] + }, + { + "task_id": "0002187", + "id": "RH20T_cfg7_task_0004_user_0020_scene_0005_cfg_0007_104122061330#144", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002187/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 304, + 256 + ], + [ + 301, + 230 + ], + [ + 328, + 260 + ], + [ + 277, + 253 + ], + [ + 332, + 287 + ], + [ + 279, + 281 + ] + ], + [ + [ + 290, + 271 + ], + [ + 275, + 271 + ], + [ + 264, + 270 + ], + [ + 313, + 272 + ], + [ + 279, + 270 + ], + [ + 329, + 273 + ] + ], + [ + [ + 293, + 219 + ], + [ + 269, + 234 + ], + [ + 288, + 219 + ], + [ + 298, + 218 + ], + [ + 313, + 203 + ], + [ + 321, + 204 + ] + ], + [ + [ + 289, + 245 + ], + [ + 276, + 243 + ], + [ + 278, + 273 + ], + [ + 299, + 218 + ], + [ + 290, + 273 + ], + [ + 310, + 221 + ] + ] + ] + }, + { + "task_id": "0002188", + "id": "RH20T_cfg6_task_0046_user_0020_scene_0004_cfg_0006_104122060811#376", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002188/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 365, + 118 + ], + [ + 376, + 123 + ], + [ + 365, + 132 + ], + [ + 365, + 104 + ], + [ + 354, + 127 + ], + [ + 353, + 99 + ] + ], + [ + [ + 363, + 108 + ], + [ + 368, + 120 + ], + [ + 349, + 111 + ], + [ + 378, + 106 + ], + [ + 344, + 99 + ], + [ + 374, + 93 + ] + ], + [ + [ + 373, + 122 + ], + [ + 384, + 124 + ], + [ + 364, + 133 + ], + [ + 382, + 110 + ], + [ + 352, + 130 + ], + [ + 370, + 107 + ] + ], + [ + [ + 350, + 124 + ], + [ + 336, + 119 + ], + [ + 344, + 134 + ], + [ + 357, + 113 + ], + [ + 357, + 138 + ], + [ + 371, + 118 + ] + ] + ] + }, + { + "task_id": "0002189", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122060811#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002189/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 124 + ], + [ + 358, + 116 + ], + [ + 358, + 136 + ], + [ + 367, + 112 + ], + [ + 363, + 145 + ], + [ + 372, + 121 + ] + ], + [ + [ + 365, + 128 + ], + [ + 369, + 140 + ], + [ + 368, + 134 + ], + [ + 362, + 121 + ], + [ + 364, + 123 + ], + [ + 358, + 109 + ] + ], + [ + [ + 372, + 139 + ], + [ + 382, + 129 + ], + [ + 381, + 143 + ], + [ + 362, + 135 + ], + [ + 371, + 152 + ], + [ + 352, + 145 + ] + ], + [ + [ + 372, + 129 + ], + [ + 385, + 135 + ], + [ + 365, + 139 + ], + [ + 380, + 118 + ], + [ + 352, + 133 + ], + [ + 367, + 112 + ] + ] + ] + }, + { + "task_id": "0002191", + "id": "RH20T_cfg6_task_0045_user_0020_scene_0010_cfg_0006_104122061330#258", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the brush on the pen rack'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002191/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 297, + 224 + ], + [ + 310, + 237 + ], + [ + 296, + 250 + ], + [ + 298, + 196 + ], + [ + 284, + 237 + ], + [ + 286, + 185 + ] + ], + [ + [ + 276, + 227 + ], + [ + 260, + 235 + ], + [ + 298, + 234 + ], + [ + 251, + 221 + ], + [ + 316, + 226 + ], + [ + 268, + 211 + ] + ], + [ + [ + 310, + 225 + ], + [ + 289, + 240 + ], + [ + 317, + 248 + ], + [ + 302, + 201 + ], + [ + 335, + 233 + ], + [ + 322, + 187 + ] + ], + [ + [ + 285, + 242 + ], + [ + 309, + 230 + ], + [ + 272, + 219 + ], + [ + 298, + 264 + ], + [ + 248, + 232 + ], + [ + 274, + 276 + ] + ] + ] + }, + { + "task_id": "0002193", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122063633#213", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002193/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 193, + 169 + ], + [ + 177, + 155 + ], + [ + 178, + 175 + ], + [ + 207, + 164 + ], + [ + 194, + 190 + ], + [ + 222, + 178 + ] + ], + [ + [ + 178, + 147 + ], + [ + 176, + 141 + ], + [ + 176, + 167 + ], + [ + 181, + 126 + ], + [ + 179, + 174 + ], + [ + 184, + 131 + ] + ], + [ + [ + 204, + 173 + ], + [ + 206, + 163 + ], + [ + 225, + 179 + ], + [ + 181, + 167 + ], + [ + 224, + 189 + ], + [ + 178, + 178 + ] + ], + [ + [ + 198, + 161 + ], + [ + 193, + 150 + ], + [ + 200, + 143 + ], + [ + 196, + 180 + ], + [ + 205, + 154 + ], + [ + 202, + 191 + ] + ] + ] + }, + { + "task_id": "0002194", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122063633#277", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002194/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 187, + 151 + ], + [ + 177, + 145 + ], + [ + 190, + 169 + ], + [ + 184, + 132 + ], + [ + 200, + 176 + ], + [ + 194, + 138 + ] + ], + [ + [ + 189, + 173 + ], + [ + 183, + 164 + ], + [ + 168, + 173 + ], + [ + 210, + 173 + ], + [ + 175, + 182 + ], + [ + 216, + 182 + ] + ], + [ + [ + 207, + 158 + ], + [ + 216, + 143 + ], + [ + 189, + 147 + ], + [ + 226, + 169 + ], + [ + 179, + 162 + ], + [ + 217, + 184 + ] + ], + [ + [ + 189, + 162 + ], + [ + 185, + 149 + ], + [ + 192, + 144 + ], + [ + 185, + 180 + ], + [ + 195, + 156 + ], + [ + 189, + 192 + ] + ] + ] + }, + { + "task_id": "0002197", + "id": "RH20T_cfg6_task_0046_user_0020_scene_0010_cfg_0006_104122063633#323", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002197/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 144, + 146 + ], + [ + 142, + 121 + ], + [ + 125, + 151 + ], + [ + 163, + 141 + ], + [ + 126, + 175 + ], + [ + 164, + 166 + ] + ], + [ + [ + 161, + 132 + ], + [ + 173, + 140 + ], + [ + 140, + 145 + ], + [ + 180, + 121 + ], + [ + 128, + 137 + ], + [ + 168, + 114 + ] + ], + [ + [ + 168, + 158 + ], + [ + 176, + 149 + ], + [ + 172, + 180 + ], + [ + 165, + 135 + ], + [ + 164, + 190 + ], + [ + 156, + 143 + ] + ], + [ + [ + 164, + 139 + ], + [ + 151, + 156 + ], + [ + 185, + 148 + ], + [ + 144, + 131 + ], + [ + 197, + 131 + ], + [ + 157, + 115 + ] + ] + ] + }, + { + "task_id": "0002198", + "id": "RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_104122064161#126", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Cover the box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002198/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 372, + 225 + ], + [ + 374, + 204 + ], + [ + 373, + 241 + ], + [ + 371, + 210 + ], + [ + 372, + 261 + ], + [ + 370, + 229 + ] + ], + [ + [ + 353, + 235 + ], + [ + 338, + 238 + ], + [ + 368, + 254 + ], + [ + 340, + 216 + ], + [ + 381, + 250 + ], + [ + 354, + 214 + ] + ], + [ + [ + 368, + 220 + ], + [ + 366, + 221 + ], + [ + 352, + 240 + ], + [ + 383, + 201 + ], + [ + 353, + 240 + ], + [ + 385, + 199 + ] + ], + [ + [ + 389, + 206 + ], + [ + 377, + 211 + ], + [ + 368, + 215 + ], + [ + 409, + 197 + ], + [ + 381, + 209 + ], + [ + 424, + 191 + ] + ] + ] + }, + { + "task_id": "0002200", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0002_cfg_0006_104122061602#118", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002200/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 431, + 160 + ], + [ + 447, + 179 + ], + [ + 411, + 174 + ], + [ + 453, + 146 + ], + [ + 396, + 156 + ], + [ + 437, + 127 + ] + ], + [ + [ + 469, + 194 + ], + [ + 473, + 220 + ], + [ + 497, + 192 + ], + [ + 443, + 196 + ], + [ + 493, + 165 + ], + [ + 439, + 171 + ] + ], + [ + [ + 460, + 177 + ], + [ + 462, + 154 + ], + [ + 480, + 188 + ], + [ + 441, + 167 + ], + [ + 477, + 209 + ], + [ + 439, + 187 + ] + ], + [ + [ + 452, + 203 + ], + [ + 430, + 202 + ], + [ + 443, + 183 + ], + [ + 461, + 225 + ], + [ + 463, + 185 + ], + [ + 482, + 226 + ] + ] + ] + }, + { + "task_id": "0002201", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0002_cfg_0006_104122061602#265", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002201/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 388, + 161 + ], + [ + 399, + 143 + ], + [ + 404, + 183 + ], + [ + 373, + 141 + ], + [ + 393, + 199 + ], + [ + 364, + 158 + ] + ], + [ + [ + 415, + 168 + ], + [ + 419, + 180 + ], + [ + 421, + 143 + ], + [ + 409, + 192 + ], + [ + 417, + 133 + ], + [ + 406, + 180 + ] + ], + [ + [ + 387, + 157 + ], + [ + 396, + 180 + ], + [ + 373, + 144 + ], + [ + 399, + 169 + ], + [ + 365, + 123 + ], + [ + 391, + 149 + ] + ], + [ + [ + 376, + 150 + ], + [ + 400, + 139 + ], + [ + 369, + 127 + ], + [ + 383, + 172 + ], + [ + 345, + 138 + ], + [ + 360, + 183 + ] + ] + ] + }, + { + "task_id": "0002202", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0002_cfg_0006_104122061602#418", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002202/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 411, + 183 + ], + [ + 430, + 168 + ], + [ + 402, + 177 + ], + [ + 422, + 189 + ], + [ + 384, + 192 + ], + [ + 402, + 205 + ] + ], + [ + [ + 409, + 167 + ], + [ + 419, + 151 + ], + [ + 422, + 187 + ], + [ + 397, + 148 + ], + [ + 412, + 201 + ], + [ + 389, + 163 + ] + ], + [ + [ + 397, + 203 + ], + [ + 376, + 209 + ], + [ + 386, + 182 + ], + [ + 409, + 224 + ], + [ + 406, + 177 + ], + [ + 429, + 218 + ] + ], + [ + [ + 406, + 152 + ], + [ + 422, + 156 + ], + [ + 421, + 135 + ], + [ + 392, + 169 + ], + [ + 403, + 130 + ], + [ + 375, + 165 + ] + ] + ] + }, + { + "task_id": "0002203", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0002_cfg_0006_104122061602#599", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002203/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 394, + 117 + ], + [ + 390, + 137 + ], + [ + 379, + 109 + ], + [ + 409, + 124 + ], + [ + 383, + 88 + ], + [ + 412, + 105 + ] + ], + [ + [ + 406, + 127 + ], + [ + 416, + 147 + ], + [ + 402, + 122 + ], + [ + 410, + 133 + ], + [ + 392, + 102 + ], + [ + 401, + 114 + ] + ], + [ + [ + 385, + 118 + ], + [ + 396, + 118 + ], + [ + 390, + 140 + ], + [ + 381, + 96 + ], + [ + 380, + 139 + ], + [ + 371, + 98 + ] + ], + [ + [ + 372, + 120 + ], + [ + 369, + 136 + ], + [ + 353, + 110 + ], + [ + 390, + 128 + ], + [ + 357, + 95 + ], + [ + 393, + 113 + ] + ] + ] + }, + { + "task_id": "0002204", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0002_cfg_0006_104122061602#735", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002204/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 412, + 141 + ], + [ + 416, + 126 + ], + [ + 426, + 158 + ], + [ + 398, + 125 + ], + [ + 421, + 171 + ], + [ + 394, + 139 + ] + ], + [ + [ + 425, + 139 + ], + [ + 403, + 141 + ], + [ + 418, + 129 + ], + [ + 433, + 150 + ], + [ + 439, + 126 + ], + [ + 456, + 147 + ] + ], + [ + [ + 414, + 158 + ], + [ + 421, + 142 + ], + [ + 421, + 141 + ], + [ + 408, + 173 + ], + [ + 414, + 158 + ], + [ + 401, + 190 + ] + ], + [ + [ + 393, + 139 + ], + [ + 415, + 130 + ], + [ + 385, + 118 + ], + [ + 402, + 159 + ], + [ + 364, + 127 + ], + [ + 381, + 168 + ] + ] + ] + }, + { + "task_id": "0002206", + "id": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_035622060973#245", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002206/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 250, + 165 + ], + [ + 247, + 149 + ], + [ + 239, + 154 + ], + [ + 261, + 177 + ], + [ + 242, + 169 + ], + [ + 263, + 192 + ] + ], + [ + [ + 241, + 176 + ], + [ + 260, + 168 + ], + [ + 246, + 196 + ], + [ + 236, + 156 + ], + [ + 226, + 204 + ], + [ + 215, + 164 + ] + ], + [ + [ + 225, + 145 + ], + [ + 207, + 142 + ], + [ + 211, + 143 + ], + [ + 239, + 147 + ], + [ + 227, + 146 + ], + [ + 255, + 150 + ] + ], + [ + [ + 252, + 158 + ], + [ + 235, + 163 + ], + [ + 247, + 138 + ], + [ + 257, + 177 + ], + [ + 263, + 134 + ], + [ + 273, + 172 + ] + ] + ] + }, + { + "task_id": "0002207", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0008_cfg_0004_038522062288#135", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002207/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 210, + 173 + ], + [ + 184, + 198 + ], + [ + 184, + 148 + ], + [ + 234, + 198 + ], + [ + 210, + 124 + ], + [ + 260, + 174 + ] + ], + [ + [ + 183, + 138 + ], + [ + 218, + 137 + ], + [ + 184, + 167 + ], + [ + 182, + 107 + ], + [ + 149, + 170 + ], + [ + 143, + 107 + ] + ], + [ + [ + 195, + 135 + ], + [ + 187, + 114 + ], + [ + 164, + 126 + ], + [ + 226, + 144 + ], + [ + 173, + 145 + ], + [ + 232, + 162 + ] + ], + [ + [ + 190, + 169 + ], + [ + 166, + 183 + ], + [ + 213, + 178 + ], + [ + 164, + 158 + ], + [ + 237, + 164 + ], + [ + 190, + 143 + ] + ] + ] + }, + { + "task_id": "0002208", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0008_cfg_0004_038522062288#231", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002208/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 185, + 149 + ], + [ + 177, + 124 + ], + [ + 157, + 137 + ], + [ + 215, + 161 + ], + [ + 165, + 160 + ], + [ + 220, + 183 + ] + ], + [ + [ + 153, + 137 + ], + [ + 145, + 148 + ], + [ + 143, + 169 + ], + [ + 162, + 107 + ], + [ + 152, + 159 + ], + [ + 172, + 93 + ] + ], + [ + [ + 213, + 141 + ], + [ + 238, + 159 + ], + [ + 235, + 120 + ], + [ + 189, + 163 + ], + [ + 210, + 102 + ], + [ + 163, + 144 + ] + ], + [ + [ + 181, + 183 + ], + [ + 194, + 155 + ], + [ + 208, + 180 + ], + [ + 152, + 186 + ], + [ + 196, + 205 + ], + [ + 141, + 214 + ] + ] + ] + }, + { + "task_id": "0002209", + "id": "RH20T_cfg2_task_0057_user_0016_scene_0001_cfg_0002_037522062165#356", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002209/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 138, + 182 + ], + [ + 129, + 180 + ], + [ + 130, + 158 + ], + [ + 146, + 205 + ], + [ + 139, + 160 + ], + [ + 154, + 205 + ] + ], + [ + [ + 115, + 195 + ], + [ + 106, + 219 + ], + [ + 142, + 201 + ], + [ + 89, + 189 + ], + [ + 150, + 178 + ], + [ + 98, + 166 + ] + ], + [ + [ + 152, + 190 + ], + [ + 149, + 174 + ], + [ + 132, + 179 + ], + [ + 173, + 202 + ], + [ + 135, + 195 + ], + [ + 175, + 216 + ] + ], + [ + [ + 149, + 172 + ], + [ + 127, + 176 + ], + [ + 151, + 150 + ], + [ + 146, + 194 + ], + [ + 173, + 145 + ], + [ + 169, + 189 + ] + ] + ] + }, + { + "task_id": "0002210", + "id": "RH20T_cfg2_task_0057_user_0016_scene_0001_cfg_0002_037522062165#424", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002210/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 147, + 191 + ], + [ + 125, + 204 + ], + [ + 160, + 212 + ], + [ + 133, + 171 + ], + [ + 180, + 199 + ], + [ + 154, + 159 + ] + ], + [ + [ + 146, + 209 + ], + [ + 143, + 192 + ], + [ + 126, + 197 + ], + [ + 167, + 221 + ], + [ + 129, + 212 + ], + [ + 169, + 235 + ] + ], + [ + [ + 158, + 214 + ], + [ + 172, + 231 + ], + [ + 139, + 234 + ], + [ + 176, + 195 + ], + [ + 124, + 216 + ], + [ + 163, + 176 + ] + ], + [ + [ + 174, + 213 + ], + [ + 194, + 198 + ], + [ + 191, + 232 + ], + [ + 156, + 194 + ], + [ + 172, + 246 + ], + [ + 136, + 208 + ] + ] + ] + }, + { + "task_id": "0002211", + "id": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_038522063145#150", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002211/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 435, + 97 + ], + [ + 414, + 93 + ], + [ + 438, + 78 + ], + [ + 433, + 116 + ], + [ + 460, + 82 + ], + [ + 454, + 121 + ] + ], + [ + [ + 421, + 137 + ], + [ + 423, + 127 + ], + [ + 400, + 142 + ], + [ + 443, + 133 + ], + [ + 399, + 151 + ], + [ + 440, + 143 + ] + ], + [ + [ + 442, + 124 + ], + [ + 426, + 112 + ], + [ + 456, + 106 + ], + [ + 428, + 141 + ], + [ + 473, + 119 + ], + [ + 444, + 153 + ] + ], + [ + [ + 436, + 121 + ], + [ + 446, + 107 + ], + [ + 454, + 135 + ], + [ + 419, + 108 + ], + [ + 444, + 147 + ], + [ + 411, + 122 + ] + ] + ] + }, + { + "task_id": "0002212", + "id": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_038522063145#217", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002212/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 428, + 130 + ], + [ + 436, + 113 + ], + [ + 449, + 140 + ], + [ + 408, + 119 + ], + [ + 441, + 156 + ], + [ + 402, + 135 + ] + ], + [ + [ + 416, + 123 + ], + [ + 396, + 120 + ], + [ + 417, + 103 + ], + [ + 415, + 144 + ], + [ + 436, + 106 + ], + [ + 435, + 147 + ] + ], + [ + [ + 422, + 124 + ], + [ + 437, + 137 + ], + [ + 438, + 108 + ], + [ + 407, + 140 + ], + [ + 423, + 95 + ], + [ + 393, + 127 + ] + ], + [ + [ + 418, + 114 + ], + [ + 433, + 96 + ], + [ + 406, + 100 + ], + [ + 428, + 127 + ], + [ + 391, + 119 + ], + [ + 414, + 145 + ] + ] + ] + }, + { + "task_id": "0002215", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122064161#286", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002215/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 341, + 133 + ], + [ + 343, + 124 + ], + [ + 341, + 153 + ], + [ + 342, + 114 + ], + [ + 339, + 161 + ], + [ + 341, + 123 + ] + ], + [ + [ + 326, + 139 + ], + [ + 317, + 158 + ], + [ + 334, + 149 + ], + [ + 318, + 129 + ], + [ + 344, + 132 + ], + [ + 328, + 110 + ] + ], + [ + [ + 335, + 135 + ], + [ + 356, + 139 + ], + [ + 338, + 139 + ], + [ + 332, + 131 + ], + [ + 315, + 135 + ], + [ + 312, + 127 + ] + ], + [ + [ + 349, + 132 + ], + [ + 344, + 143 + ], + [ + 353, + 150 + ], + [ + 346, + 114 + ], + [ + 357, + 140 + ], + [ + 351, + 104 + ] + ] + ] + }, + { + "task_id": "0002219", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_819112071037#100", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002219/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 11, + 199 + ], + [ + 17, + 253 + ], + [ + 70, + 229 + ], + [ + 0, + 162 + ], + [ + 61, + 187 + ], + [ + 0, + 116 + ] + ], + [ + [ + 0, + 257 + ], + [ + 0, + 214 + ], + [ + 0, + 344 + ], + [ + 0, + 186 + ], + [ + 0, + 360 + ], + [ + 0, + 223 + ] + ], + [ + [ + 14, + 124 + ], + [ + 0, + 47 + ], + [ + 0, + 221 + ], + [ + 60, + 46 + ], + [ + 29, + 259 + ], + [ + 105, + 104 + ] + ], + [ + [ + 12, + 190 + ], + [ + 0, + 102 + ], + [ + 0, + 185 + ], + [ + 104, + 194 + ], + [ + 0, + 263 + ], + [ + 112, + 269 + ] + ] + ] + }, + { + "task_id": "0002220", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_819112071037#253", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002220/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 130, + 200 + ], + [ + 126, + 114 + ], + [ + 39, + 195 + ], + [ + 219, + 205 + ], + [ + 48, + 273 + ], + [ + 218, + 280 + ] + ], + [ + [ + 156, + 289 + ], + [ + 159, + 212 + ], + [ + 185, + 245 + ], + [ + 132, + 323 + ], + [ + 185, + 350 + ], + [ + 126, + 360 + ] + ], + [ + [ + 81, + 199 + ], + [ + 99, + 109 + ], + [ + 162, + 181 + ], + [ + 0, + 225 + ], + [ + 144, + 247 + ], + [ + 0, + 314 + ] + ], + [ + [ + 75, + 267 + ], + [ + 27, + 329 + ], + [ + 133, + 226 + ], + [ + 23, + 305 + ], + [ + 161, + 185 + ], + [ + 64, + 257 + ] + ] + ] + }, + { + "task_id": "0002223", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_819112071037#808", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002223/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 412, + 257 + ], + [ + 367, + 335 + ], + [ + 422, + 267 + ], + [ + 406, + 251 + ], + [ + 477, + 173 + ], + [ + 444, + 186 + ] + ], + [ + [ + 431, + 167 + ], + [ + 421, + 148 + ], + [ + 462, + 250 + ], + [ + 402, + 91 + ], + [ + 465, + 251 + ], + [ + 413, + 115 + ] + ], + [ + [ + 389, + 244 + ], + [ + 403, + 158 + ], + [ + 298, + 243 + ], + [ + 474, + 245 + ], + [ + 292, + 320 + ], + [ + 457, + 318 + ] + ], + [ + [ + 329, + 360 + ], + [ + 407, + 346 + ], + [ + 311, + 360 + ], + [ + 350, + 278 + ], + [ + 217, + 360 + ], + [ + 245, + 296 + ] + ] + ] + }, + { + "task_id": "0002224", + "id": "RH20T_cfg2_task_0018_user_0003_scene_0010_cfg_0002_104422070044#98", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002224/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 389, + 218 + ], + [ + 383, + 206 + ], + [ + 371, + 222 + ], + [ + 406, + 214 + ], + [ + 377, + 234 + ], + [ + 413, + 226 + ] + ], + [ + [ + 388, + 224 + ], + [ + 387, + 214 + ], + [ + 391, + 241 + ], + [ + 384, + 208 + ], + [ + 391, + 250 + ], + [ + 385, + 218 + ] + ], + [ + [ + 397, + 209 + ], + [ + 415, + 217 + ], + [ + 400, + 195 + ], + [ + 394, + 223 + ], + [ + 382, + 187 + ], + [ + 378, + 216 + ] + ], + [ + [ + 385, + 237 + ], + [ + 398, + 250 + ], + [ + 400, + 225 + ], + [ + 371, + 248 + ], + [ + 388, + 212 + ], + [ + 358, + 234 + ] + ] + ] + }, + { + "task_id": "0002226", + "id": "RH20T_cfg3_task_0006_user_0016_scene_0004_cfg_0003_104122063550#176", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002226/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 297 + ], + [ + 361, + 308 + ], + [ + 380, + 325 + ], + [ + 363, + 272 + ], + [ + 391, + 313 + ], + [ + 373, + 259 + ] + ], + [ + [ + 358, + 266 + ], + [ + 334, + 281 + ], + [ + 373, + 290 + ], + [ + 344, + 242 + ], + [ + 397, + 274 + ], + [ + 367, + 227 + ] + ], + [ + [ + 327, + 280 + ], + [ + 302, + 282 + ], + [ + 336, + 302 + ], + [ + 319, + 256 + ], + [ + 362, + 300 + ], + [ + 346, + 253 + ] + ], + [ + [ + 354, + 272 + ], + [ + 357, + 265 + ], + [ + 365, + 299 + ], + [ + 343, + 246 + ], + [ + 361, + 304 + ], + [ + 341, + 254 + ] + ] + ] + }, + { + "task_id": "0002227", + "id": "RH20T_cfg5_task_0004_user_0010_scene_0001_cfg_0005_036422060215#429", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002227/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 354, + 154 + ], + [ + 352, + 136 + ], + [ + 364, + 159 + ], + [ + 345, + 149 + ], + [ + 365, + 177 + ], + [ + 347, + 166 + ] + ], + [ + [ + 349, + 151 + ], + [ + 354, + 136 + ], + [ + 362, + 146 + ], + [ + 334, + 156 + ], + [ + 357, + 160 + ], + [ + 330, + 170 + ] + ], + [ + [ + 349, + 136 + ], + [ + 360, + 152 + ], + [ + 349, + 133 + ], + [ + 349, + 139 + ], + [ + 339, + 118 + ], + [ + 337, + 122 + ] + ], + [ + [ + 347, + 149 + ], + [ + 350, + 154 + ], + [ + 343, + 131 + ], + [ + 350, + 166 + ], + [ + 340, + 127 + ], + [ + 346, + 161 + ] + ] + ] + }, + { + "task_id": "0002228", + "id": "RH20T_cfg3_task_0032_user_0011_scene_0001_cfg_0003_104122062295#145", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002228/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 240, + 206 + ], + [ + 210, + 216 + ], + [ + 214, + 193 + ], + [ + 268, + 221 + ], + [ + 240, + 185 + ], + [ + 293, + 211 + ] + ], + [ + [ + 199, + 162 + ], + [ + 191, + 177 + ], + [ + 230, + 186 + ], + [ + 167, + 137 + ], + [ + 234, + 172 + ], + [ + 177, + 126 + ] + ], + [ + [ + 230, + 191 + ], + [ + 228, + 156 + ], + [ + 218, + 172 + ], + [ + 244, + 213 + ], + [ + 219, + 202 + ], + [ + 244, + 245 + ] + ], + [ + [ + 212, + 181 + ], + [ + 204, + 159 + ], + [ + 219, + 143 + ], + [ + 205, + 216 + ], + [ + 229, + 167 + ], + [ + 213, + 244 + ] + ] + ] + }, + { + "task_id": "0002230", + "id": "RH20T_cfg3_task_0032_user_0011_scene_0001_cfg_0003_104122062295#371", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002230/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 249, + 210 + ], + [ + 206, + 195 + ], + [ + 268, + 177 + ], + [ + 231, + 240 + ], + [ + 309, + 192 + ], + [ + 270, + 253 + ] + ], + [ + [ + 292, + 286 + ], + [ + 331, + 273 + ], + [ + 276, + 247 + ], + [ + 310, + 328 + ], + [ + 235, + 258 + ], + [ + 267, + 345 + ] + ], + [ + [ + 280, + 228 + ], + [ + 276, + 190 + ], + [ + 260, + 210 + ], + [ + 302, + 248 + ], + [ + 264, + 244 + ], + [ + 305, + 285 + ] + ], + [ + [ + 306, + 256 + ], + [ + 265, + 266 + ], + [ + 316, + 298 + ], + [ + 296, + 217 + ], + [ + 355, + 288 + ], + [ + 334, + 209 + ] + ] + ] + }, + { + "task_id": "0002231", + "id": "RH20T_cfg4_task_0016_user_0016_scene_0005_cfg_0004_038522062288#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002231/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 211, + 274 + ], + [ + 205, + 255 + ], + [ + 181, + 259 + ], + [ + 241, + 290 + ], + [ + 188, + 275 + ], + [ + 245, + 305 + ] + ], + [ + [ + 204, + 294 + ], + [ + 226, + 268 + ], + [ + 182, + 285 + ], + [ + 228, + 303 + ], + [ + 159, + 312 + ], + [ + 204, + 332 + ] + ], + [ + [ + 226, + 275 + ], + [ + 213, + 281 + ], + [ + 247, + 243 + ], + [ + 205, + 307 + ], + [ + 256, + 240 + ], + [ + 218, + 300 + ] + ], + [ + [ + 192, + 266 + ], + [ + 170, + 285 + ], + [ + 216, + 257 + ], + [ + 169, + 274 + ], + [ + 235, + 241 + ], + [ + 189, + 258 + ] + ] + ] + }, + { + "task_id": "0002233", + "id": "RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_037522061512#226", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002233/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 203, + 129 + ], + [ + 217, + 146 + ], + [ + 186, + 141 + ], + [ + 218, + 118 + ], + [ + 171, + 123 + ], + [ + 204, + 101 + ] + ], + [ + [ + 238, + 92 + ], + [ + 215, + 93 + ], + [ + 239, + 115 + ], + [ + 237, + 69 + ], + [ + 262, + 114 + ], + [ + 260, + 68 + ] + ], + [ + [ + 219, + 87 + ], + [ + 197, + 87 + ], + [ + 215, + 71 + ], + [ + 223, + 101 + ], + [ + 239, + 70 + ], + [ + 245, + 101 + ] + ], + [ + [ + 228, + 109 + ], + [ + 223, + 91 + ], + [ + 239, + 96 + ], + [ + 215, + 123 + ], + [ + 243, + 113 + ], + [ + 220, + 140 + ] + ] + ] + }, + { + "task_id": "0002235", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0002_cfg_0004_104422070011#316", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002235/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 427, + 178 + ], + [ + 410, + 182 + ], + [ + 442, + 158 + ], + [ + 414, + 198 + ], + [ + 458, + 155 + ], + [ + 430, + 194 + ] + ], + [ + [ + 423, + 169 + ], + [ + 441, + 157 + ], + [ + 435, + 193 + ], + [ + 412, + 145 + ], + [ + 419, + 203 + ], + [ + 397, + 157 + ] + ], + [ + [ + 416, + 170 + ], + [ + 432, + 153 + ], + [ + 436, + 185 + ], + [ + 396, + 155 + ], + [ + 420, + 203 + ], + [ + 380, + 172 + ] + ], + [ + [ + 406, + 155 + ], + [ + 405, + 135 + ], + [ + 428, + 144 + ], + [ + 385, + 165 + ], + [ + 429, + 166 + ], + [ + 385, + 187 + ] + ] + ] + }, + { + "task_id": "0002236", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0002_cfg_0004_104422070011#546", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002236/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 310, + 156 + ], + [ + 284, + 152 + ], + [ + 324, + 151 + ], + [ + 298, + 160 + ], + [ + 350, + 154 + ], + [ + 322, + 163 + ] + ], + [ + [ + 273, + 145 + ], + [ + 244, + 141 + ], + [ + 283, + 157 + ], + [ + 264, + 133 + ], + [ + 311, + 161 + ], + [ + 290, + 137 + ] + ], + [ + [ + 301, + 144 + ], + [ + 316, + 128 + ], + [ + 306, + 171 + ], + [ + 297, + 119 + ], + [ + 292, + 185 + ], + [ + 285, + 134 + ] + ], + [ + [ + 317, + 172 + ], + [ + 347, + 165 + ], + [ + 314, + 151 + ], + [ + 320, + 192 + ], + [ + 283, + 158 + ], + [ + 292, + 199 + ] + ] + ] + }, + { + "task_id": "0002237", + "id": "RH20T_cfg6_task_0038_user_0020_scene_0001_cfg_0006_104122063633#549", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002237/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 161, + 210 + ], + [ + 161, + 222 + ], + [ + 168, + 233 + ], + [ + 155, + 189 + ], + [ + 168, + 220 + ], + [ + 155, + 175 + ] + ], + [ + [ + 198, + 208 + ], + [ + 219, + 196 + ], + [ + 213, + 224 + ], + [ + 185, + 193 + ], + [ + 191, + 238 + ], + [ + 163, + 205 + ] + ], + [ + [ + 179, + 189 + ], + [ + 184, + 168 + ], + [ + 180, + 202 + ], + [ + 177, + 174 + ], + [ + 175, + 223 + ], + [ + 171, + 196 + ] + ], + [ + [ + 200, + 202 + ], + [ + 172, + 205 + ], + [ + 202, + 217 + ], + [ + 198, + 188 + ], + [ + 230, + 213 + ], + [ + 223, + 184 + ] + ] + ] + }, + { + "task_id": "0002239", + "id": "RH20T_cfg5_task_0058_user_0007_scene_0002_cfg_0005_037522062165#240", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002239/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 259, + 201 + ], + [ + 278, + 210 + ], + [ + 259, + 228 + ], + [ + 259, + 177 + ], + [ + 237, + 219 + ], + [ + 239, + 166 + ] + ], + [ + [ + 248, + 228 + ], + [ + 238, + 212 + ], + [ + 228, + 220 + ], + [ + 270, + 236 + ], + [ + 239, + 234 + ], + [ + 279, + 251 + ] + ], + [ + [ + 251, + 214 + ], + [ + 266, + 194 + ], + [ + 263, + 230 + ], + [ + 238, + 197 + ], + [ + 248, + 251 + ], + [ + 221, + 219 + ] + ], + [ + [ + 277, + 217 + ], + [ + 289, + 207 + ], + [ + 295, + 202 + ], + [ + 260, + 231 + ], + [ + 282, + 211 + ], + [ + 247, + 242 + ] + ] + ] + }, + { + "task_id": "0002241", + "id": "RH20T_cfg5_task_0042_user_0007_scene_0008_cfg_0005_036422060215#303", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002241/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 268, + 145 + ], + [ + 270, + 161 + ], + [ + 287, + 150 + ], + [ + 248, + 141 + ], + [ + 285, + 136 + ], + [ + 246, + 127 + ] + ], + [ + [ + 293, + 168 + ], + [ + 279, + 167 + ], + [ + 295, + 148 + ], + [ + 290, + 188 + ], + [ + 310, + 148 + ], + [ + 305, + 189 + ] + ], + [ + [ + 261, + 175 + ], + [ + 265, + 157 + ], + [ + 280, + 179 + ], + [ + 241, + 172 + ], + [ + 276, + 196 + ], + [ + 237, + 190 + ] + ], + [ + [ + 273, + 150 + ], + [ + 277, + 135 + ], + [ + 292, + 152 + ], + [ + 253, + 148 + ], + [ + 288, + 165 + ], + [ + 250, + 162 + ] + ] + ] + }, + { + "task_id": "0002242", + "id": "RH20T_cfg1_task_0045_user_0015_scene_0002_cfg_0001_039422060546#111", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the brush on the pen rack'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002242/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 137 + ], + [ + 369, + 153 + ], + [ + 341, + 153 + ], + [ + 354, + 122 + ], + [ + 318, + 136 + ], + [ + 333, + 106 + ] + ], + [ + [ + 351, + 136 + ], + [ + 351, + 115 + ], + [ + 368, + 154 + ], + [ + 334, + 118 + ], + [ + 367, + 172 + ], + [ + 334, + 137 + ] + ], + [ + [ + 350, + 112 + ], + [ + 341, + 92 + ], + [ + 369, + 124 + ], + [ + 333, + 101 + ], + [ + 377, + 142 + ], + [ + 343, + 119 + ] + ], + [ + [ + 351, + 128 + ], + [ + 332, + 119 + ], + [ + 368, + 110 + ], + [ + 334, + 147 + ], + [ + 389, + 119 + ], + [ + 354, + 157 + ] + ] + ] + }, + { + "task_id": "0002244", + "id": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_750612070853#71", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002244/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 599, + 195 + ], + [ + 620, + 134 + ], + [ + 595, + 244 + ], + [ + 602, + 154 + ], + [ + 575, + 294 + ], + [ + 586, + 203 + ] + ], + [ + [ + 640, + 174 + ], + [ + 640, + 86 + ], + [ + 591, + 187 + ], + [ + 640, + 161 + ], + [ + 597, + 274 + ], + [ + 640, + 243 + ] + ], + [ + [ + 532, + 200 + ], + [ + 462, + 168 + ], + [ + 510, + 238 + ], + [ + 560, + 152 + ], + [ + 578, + 272 + ], + [ + 640, + 191 + ] + ], + [ + [ + 523, + 261 + ], + [ + 455, + 231 + ], + [ + 509, + 310 + ], + [ + 534, + 221 + ], + [ + 592, + 349 + ], + [ + 601, + 248 + ] + ] + ] + }, + { + "task_id": "0002246", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0006_cfg_0004_104122063550#111", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002246/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 306 + ], + [ + 319, + 333 + ], + [ + 366, + 307 + ], + [ + 313, + 305 + ], + [ + 382, + 281 + ], + [ + 330, + 282 + ] + ], + [ + [ + 320, + 265 + ], + [ + 326, + 229 + ], + [ + 359, + 271 + ], + [ + 281, + 259 + ], + [ + 353, + 311 + ], + [ + 273, + 298 + ] + ], + [ + [ + 331, + 283 + ], + [ + 337, + 259 + ], + [ + 325, + 307 + ], + [ + 338, + 256 + ], + [ + 319, + 335 + ], + [ + 333, + 282 + ] + ], + [ + [ + 319, + 276 + ], + [ + 322, + 269 + ], + [ + 342, + 249 + ], + [ + 294, + 305 + ], + [ + 338, + 257 + ], + [ + 292, + 309 + ] + ] + ] + }, + { + "task_id": "0002249", + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104122062295#236", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002249/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 104, + 51 + ], + [ + 97, + 2 + ], + [ + 74, + 31 + ], + [ + 137, + 71 + ], + [ + 82, + 73 + ], + [ + 141, + 113 + ] + ], + [ + [ + 136, + 86 + ], + [ + 127, + 100 + ], + [ + 161, + 128 + ], + [ + 111, + 42 + ], + [ + 175, + 116 + ], + [ + 120, + 21 + ] + ], + [ + [ + 82, + 52 + ], + [ + 100, + 96 + ], + [ + 117, + 45 + ], + [ + 50, + 58 + ], + [ + 99, + 0 + ], + [ + 31, + 12 + ] + ], + [ + [ + 123, + 43 + ], + [ + 159, + 38 + ], + [ + 95, + 16 + ], + [ + 152, + 71 + ], + [ + 52, + 19 + ], + [ + 112, + 78 + ] + ] + ] + }, + { + "task_id": "0002250", + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104122062295#294", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002250/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 80, + 11 + ], + [ + 58, + 52 + ], + [ + 126, + 47 + ], + [ + 26, + 0 + ], + [ + 142, + 12 + ], + [ + 50, + 0 + ] + ], + [ + [ + 67, + 34 + ], + [ + 79, + 0 + ], + [ + 109, + 73 + ], + [ + 22, + 0 + ], + [ + 96, + 97 + ], + [ + 15, + 25 + ] + ], + [ + [ + 101, + 37 + ], + [ + 93, + 0 + ], + [ + 69, + 19 + ], + [ + 136, + 57 + ], + [ + 78, + 64 + ], + [ + 141, + 102 + ] + ], + [ + [ + 132, + 18 + ], + [ + 163, + 0 + ], + [ + 92, + 15 + ], + [ + 176, + 20 + ], + [ + 56, + 46 + ], + [ + 141, + 54 + ] + ] + ] + }, + { + "task_id": "0002252", + "id": "RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_104122060811#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002252/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 383, + 158 + ], + [ + 384, + 147 + ], + [ + 395, + 154 + ], + [ + 372, + 162 + ], + [ + 395, + 166 + ], + [ + 370, + 174 + ] + ], + [ + [ + 384, + 153 + ], + [ + 376, + 165 + ], + [ + 387, + 159 + ], + [ + 381, + 147 + ], + [ + 395, + 147 + ], + [ + 389, + 135 + ] + ], + [ + [ + 394, + 159 + ], + [ + 397, + 151 + ], + [ + 407, + 155 + ], + [ + 381, + 162 + ], + [ + 404, + 162 + ], + [ + 378, + 169 + ] + ], + [ + [ + 405, + 155 + ], + [ + 404, + 141 + ], + [ + 391, + 157 + ], + [ + 418, + 154 + ], + [ + 393, + 171 + ], + [ + 419, + 168 + ] + ] + ] + }, + { + "task_id": "0002256", + "id": "RH20T_cfg1_task_0034_user_0001_scene_0006_cfg_0001_819112070044#782", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002256/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 272, + 172 + ], + [ + 275, + 157 + ], + [ + 292, + 178 + ], + [ + 252, + 166 + ], + [ + 289, + 195 + ], + [ + 247, + 182 + ] + ], + [ + [ + 290, + 195 + ], + [ + 291, + 180 + ], + [ + 307, + 188 + ], + [ + 272, + 202 + ], + [ + 306, + 201 + ], + [ + 271, + 216 + ] + ], + [ + [ + 272, + 188 + ], + [ + 268, + 195 + ], + [ + 253, + 196 + ], + [ + 291, + 181 + ], + [ + 257, + 188 + ], + [ + 296, + 173 + ] + ], + [ + [ + 280, + 190 + ], + [ + 290, + 204 + ], + [ + 289, + 175 + ], + [ + 271, + 204 + ], + [ + 280, + 161 + ], + [ + 263, + 190 + ] + ] + ] + }, + { + "task_id": "0002261", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0009_cfg_0004_104122062823#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002261/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 383, + 0 + ], + [ + 387, + 6 + ], + [ + 379, + 25 + ], + [ + 387, + 0 + ], + [ + 376, + 10 + ], + [ + 383, + 0 + ] + ], + [ + [ + 378, + 0 + ], + [ + 376, + 0 + ], + [ + 413, + 0 + ], + [ + 343, + 0 + ], + [ + 413, + 13 + ], + [ + 346, + 19 + ] + ], + [ + [ + 395, + 0 + ], + [ + 370, + 0 + ], + [ + 380, + 0 + ], + [ + 413, + 0 + ], + [ + 405, + 0 + ], + [ + 442, + 0 + ] + ], + [ + [ + 392, + 0 + ], + [ + 405, + 0 + ], + [ + 378, + 0 + ], + [ + 408, + 0 + ], + [ + 367, + 0 + ], + [ + 395, + 0 + ] + ] + ] + }, + { + "task_id": "0002263", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0009_cfg_0004_104122062823#232", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002263/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 0 + ], + [ + 317, + 0 + ], + [ + 314, + 0 + ], + [ + 314, + 0 + ], + [ + 312, + 14 + ], + [ + 311, + 0 + ] + ], + [ + [ + 346, + 0 + ], + [ + 344, + 0 + ], + [ + 367, + 0 + ], + [ + 327, + 23 + ], + [ + 371, + 0 + ], + [ + 328, + 30 + ] + ], + [ + [ + 286, + 17 + ], + [ + 293, + 3 + ], + [ + 319, + 4 + ], + [ + 255, + 31 + ], + [ + 312, + 19 + ], + [ + 245, + 47 + ] + ], + [ + [ + 325, + 0 + ], + [ + 327, + 0 + ], + [ + 362, + 0 + ], + [ + 288, + 0 + ], + [ + 358, + 36 + ], + [ + 288, + 28 + ] + ] + ] + }, + { + "task_id": "0002265", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0009_cfg_0004_104122062823#294", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002265/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 0 + ], + [ + 312, + 0 + ], + [ + 341, + 27 + ], + [ + 326, + 0 + ], + [ + 363, + 44 + ], + [ + 351, + 0 + ] + ], + [ + [ + 330, + 39 + ], + [ + 320, + 10 + ], + [ + 368, + 24 + ], + [ + 293, + 54 + ], + [ + 374, + 51 + ], + [ + 304, + 78 + ] + ], + [ + [ + 326, + 9 + ], + [ + 327, + 0 + ], + [ + 361, + 2 + ], + [ + 289, + 15 + ], + [ + 359, + 39 + ], + [ + 291, + 53 + ] + ], + [ + [ + 302, + 0 + ], + [ + 341, + 1 + ], + [ + 296, + 14 + ], + [ + 308, + 0 + ], + [ + 254, + 9 + ], + [ + 271, + 0 + ] + ] + ] + }, + { + "task_id": "0002271", + "id": "RH20T_cfg7_task_0050_user_0020_scene_0008_cfg_0007_104122064161#130", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the dish on the dish rack'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002271/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 380, + 189 + ], + [ + 380, + 169 + ], + [ + 391, + 202 + ], + [ + 370, + 177 + ], + [ + 390, + 221 + ], + [ + 370, + 196 + ] + ], + [ + [ + 356, + 191 + ], + [ + 376, + 192 + ], + [ + 363, + 173 + ], + [ + 349, + 209 + ], + [ + 342, + 173 + ], + [ + 329, + 209 + ] + ], + [ + [ + 398, + 180 + ], + [ + 420, + 187 + ], + [ + 399, + 196 + ], + [ + 397, + 162 + ], + [ + 379, + 189 + ], + [ + 375, + 154 + ] + ], + [ + [ + 357, + 190 + ], + [ + 369, + 196 + ], + [ + 357, + 213 + ], + [ + 357, + 167 + ], + [ + 346, + 208 + ], + [ + 346, + 163 + ] + ] + ] + }, + { + "task_id": "0002272", + "id": "RH20T_cfg5_task_0039_user_0007_scene_0003_cfg_0005_105422061350#263", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002272/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 228, + 119 + ], + [ + 202, + 131 + ], + [ + 248, + 125 + ], + [ + 206, + 111 + ], + [ + 274, + 113 + ], + [ + 235, + 97 + ] + ], + [ + [ + 245, + 137 + ], + [ + 279, + 138 + ], + [ + 246, + 156 + ], + [ + 245, + 115 + ], + [ + 214, + 155 + ], + [ + 208, + 113 + ] + ], + [ + [ + 208, + 129 + ], + [ + 195, + 98 + ], + [ + 225, + 133 + ], + [ + 188, + 123 + ], + [ + 237, + 162 + ], + [ + 202, + 157 + ] + ], + [ + [ + 218, + 133 + ], + [ + 220, + 113 + ], + [ + 236, + 111 + ], + [ + 200, + 155 + ], + [ + 234, + 129 + ], + [ + 200, + 172 + ] + ] + ] + }, + { + "task_id": "0002274", + "id": "RH20T_cfg1_task_0065_user_0015_scene_0001_cfg_0001_038522062547#489", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002274/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 373, + 120 + ], + [ + 384, + 101 + ], + [ + 356, + 104 + ], + [ + 390, + 135 + ], + [ + 345, + 124 + ], + [ + 380, + 154 + ] + ], + [ + [ + 391, + 144 + ], + [ + 393, + 122 + ], + [ + 413, + 150 + ], + [ + 369, + 138 + ], + [ + 412, + 173 + ], + [ + 367, + 161 + ] + ], + [ + [ + 362, + 111 + ], + [ + 355, + 132 + ], + [ + 352, + 101 + ], + [ + 373, + 123 + ], + [ + 358, + 79 + ], + [ + 381, + 101 + ] + ], + [ + [ + 381, + 126 + ], + [ + 383, + 114 + ], + [ + 406, + 128 + ], + [ + 356, + 124 + ], + [ + 403, + 139 + ], + [ + 355, + 135 + ] + ] + ] + }, + { + "task_id": "0002275", + "id": "RH20T_cfg1_task_0001_user_0001_scene_0002_cfg_0001_819112070044#252", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002275/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 374, + 153 + ], + [ + 365, + 137 + ], + [ + 375, + 169 + ], + [ + 373, + 136 + ], + [ + 384, + 185 + ], + [ + 383, + 153 + ] + ], + [ + [ + 376, + 146 + ], + [ + 385, + 129 + ], + [ + 395, + 144 + ], + [ + 355, + 147 + ], + [ + 386, + 159 + ], + [ + 347, + 163 + ] + ], + [ + [ + 370, + 142 + ], + [ + 371, + 119 + ], + [ + 355, + 137 + ], + [ + 385, + 147 + ], + [ + 354, + 158 + ], + [ + 384, + 170 + ] + ], + [ + [ + 384, + 146 + ], + [ + 385, + 122 + ], + [ + 381, + 146 + ], + [ + 387, + 146 + ], + [ + 380, + 168 + ], + [ + 386, + 171 + ] + ] + ] + }, + { + "task_id": "0002276", + "id": "RH20T_cfg5_task_0118_user_0007_scene_0009_cfg_0005_104122061850#206", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the water tap'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002276/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 332, + 251 + ], + [ + 306, + 256 + ], + [ + 343, + 274 + ], + [ + 321, + 228 + ], + [ + 370, + 270 + ], + [ + 348, + 222 + ] + ], + [ + [ + 385, + 244 + ], + [ + 415, + 232 + ], + [ + 397, + 276 + ], + [ + 373, + 214 + ], + [ + 366, + 287 + ], + [ + 343, + 226 + ] + ], + [ + [ + 361, + 245 + ], + [ + 353, + 231 + ], + [ + 337, + 265 + ], + [ + 384, + 225 + ], + [ + 345, + 276 + ], + [ + 390, + 238 + ] + ], + [ + [ + 382, + 269 + ], + [ + 380, + 301 + ], + [ + 356, + 268 + ], + [ + 406, + 270 + ], + [ + 359, + 235 + ], + [ + 408, + 239 + ] + ] + ] + }, + { + "task_id": "0002277", + "id": "RH20T_cfg7_task_0036_user_0020_scene_0001_cfg_0007_104122061330#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Shake the test tube'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002277/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 313, + 167 + ], + [ + 302, + 194 + ], + [ + 321, + 171 + ], + [ + 305, + 162 + ], + [ + 331, + 147 + ], + [ + 317, + 134 + ] + ], + [ + [ + 330, + 147 + ], + [ + 349, + 145 + ], + [ + 311, + 138 + ], + [ + 350, + 156 + ], + [ + 290, + 140 + ], + [ + 329, + 159 + ] + ], + [ + [ + 316, + 195 + ], + [ + 332, + 217 + ], + [ + 298, + 212 + ], + [ + 334, + 179 + ], + [ + 281, + 189 + ], + [ + 319, + 158 + ] + ], + [ + [ + 328, + 166 + ], + [ + 316, + 192 + ], + [ + 353, + 177 + ], + [ + 304, + 155 + ], + [ + 366, + 151 + ], + [ + 316, + 130 + ] + ] + ] + }, + { + "task_id": "0002278", + "id": "RH20T_cfg6_task_0225_user_0014_scene_0006_cfg_0006_104122061330#176", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002278/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 189 + ], + [ + 261, + 176 + ], + [ + 261, + 219 + ], + [ + 286, + 159 + ], + [ + 274, + 228 + ], + [ + 297, + 172 + ] + ], + [ + [ + 277, + 198 + ], + [ + 248, + 189 + ], + [ + 271, + 178 + ], + [ + 282, + 217 + ], + [ + 303, + 187 + ], + [ + 311, + 226 + ] + ], + [ + [ + 277, + 186 + ], + [ + 277, + 171 + ], + [ + 306, + 172 + ], + [ + 248, + 201 + ], + [ + 304, + 187 + ], + [ + 250, + 214 + ] + ], + [ + [ + 291, + 155 + ], + [ + 320, + 167 + ], + [ + 308, + 150 + ], + [ + 271, + 161 + ], + [ + 282, + 140 + ], + [ + 243, + 150 + ] + ] + ] + }, + { + "task_id": "0002279", + "id": "RH20T_cfg6_task_0225_user_0014_scene_0006_cfg_0006_104122061330#257", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002279/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 323, + 146 + ], + [ + 353, + 142 + ], + [ + 329, + 182 + ], + [ + 318, + 109 + ], + [ + 297, + 188 + ], + [ + 285, + 111 + ] + ], + [ + [ + 274, + 150 + ], + [ + 247, + 138 + ], + [ + 259, + 184 + ], + [ + 289, + 115 + ], + [ + 285, + 193 + ], + [ + 313, + 127 + ] + ], + [ + [ + 272, + 167 + ], + [ + 300, + 176 + ], + [ + 257, + 195 + ], + [ + 288, + 138 + ], + [ + 226, + 187 + ], + [ + 256, + 127 + ] + ], + [ + [ + 269, + 164 + ], + [ + 285, + 158 + ], + [ + 297, + 181 + ], + [ + 240, + 146 + ], + [ + 281, + 185 + ], + [ + 227, + 153 + ] + ] + ] + }, + { + "task_id": "0002280", + "id": "RH20T_cfg6_task_0225_user_0014_scene_0006_cfg_0006_104122061330#462", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002280/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 134 + ], + [ + 250, + 139 + ], + [ + 283, + 176 + ], + [ + 286, + 89 + ], + [ + 313, + 169 + ], + [ + 317, + 88 + ] + ], + [ + [ + 311, + 170 + ], + [ + 297, + 182 + ], + [ + 275, + 152 + ], + [ + 346, + 186 + ], + [ + 289, + 137 + ], + [ + 366, + 174 + ] + ], + [ + [ + 288, + 121 + ], + [ + 253, + 93 + ], + [ + 276, + 153 + ], + [ + 298, + 94 + ], + [ + 311, + 180 + ], + [ + 328, + 119 + ] + ], + [ + [ + 285, + 123 + ], + [ + 276, + 71 + ], + [ + 235, + 132 + ], + [ + 335, + 114 + ], + [ + 245, + 181 + ], + [ + 343, + 163 + ] + ] + ] + }, + { + "task_id": "0002281", + "id": "RH20T_cfg4_task_0082_user_0015_scene_0010_cfg_0004_104122063550#51", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002281/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 374, + 222 + ], + [ + 375, + 206 + ], + [ + 388, + 202 + ], + [ + 360, + 241 + ], + [ + 388, + 219 + ], + [ + 358, + 260 + ] + ], + [ + [ + 378, + 236 + ], + [ + 392, + 230 + ], + [ + 402, + 255 + ], + [ + 354, + 216 + ], + [ + 388, + 259 + ], + [ + 343, + 223 + ] + ], + [ + [ + 344, + 265 + ], + [ + 345, + 292 + ], + [ + 330, + 279 + ], + [ + 356, + 252 + ], + [ + 328, + 249 + ], + [ + 355, + 225 + ] + ], + [ + [ + 349, + 237 + ], + [ + 351, + 232 + ], + [ + 377, + 250 + ], + [ + 322, + 225 + ], + [ + 374, + 254 + ], + [ + 322, + 231 + ] + ] + ] + }, + { + "task_id": "0002282", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0006_cfg_0004_036422060909#188", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002282/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 282, + 159 + ], + [ + 281, + 133 + ], + [ + 294, + 162 + ], + [ + 272, + 156 + ], + [ + 295, + 188 + ], + [ + 273, + 180 + ] + ], + [ + [ + 292, + 174 + ], + [ + 314, + 187 + ], + [ + 292, + 163 + ], + [ + 293, + 184 + ], + [ + 270, + 149 + ], + [ + 273, + 172 + ] + ], + [ + [ + 275, + 156 + ], + [ + 296, + 164 + ], + [ + 273, + 139 + ], + [ + 278, + 175 + ], + [ + 252, + 130 + ], + [ + 256, + 167 + ] + ], + [ + [ + 294, + 147 + ], + [ + 295, + 128 + ], + [ + 302, + 131 + ], + [ + 285, + 165 + ], + [ + 300, + 148 + ], + [ + 284, + 183 + ] + ] + ] + }, + { + "task_id": "0002284", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0006_cfg_0004_036422060909#307", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002284/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 279, + 170 + ], + [ + 281, + 153 + ], + [ + 282, + 149 + ], + [ + 275, + 192 + ], + [ + 280, + 165 + ], + [ + 273, + 207 + ] + ], + [ + [ + 294, + 189 + ], + [ + 289, + 186 + ], + [ + 271, + 200 + ], + [ + 319, + 179 + ], + [ + 277, + 202 + ], + [ + 322, + 183 + ] + ], + [ + [ + 302, + 144 + ], + [ + 281, + 145 + ], + [ + 310, + 170 + ], + [ + 294, + 119 + ], + [ + 329, + 168 + ], + [ + 313, + 120 + ] + ], + [ + [ + 280, + 159 + ], + [ + 292, + 162 + ], + [ + 257, + 153 + ], + [ + 304, + 164 + ], + [ + 242, + 149 + ], + [ + 290, + 161 + ] + ] + ] + }, + { + "task_id": "0002285", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0006_cfg_0004_036422060909#376", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002285/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 245, + 179 + ], + [ + 247, + 162 + ], + [ + 255, + 159 + ], + [ + 235, + 200 + ], + [ + 253, + 174 + ], + [ + 234, + 215 + ] + ], + [ + [ + 251, + 198 + ], + [ + 225, + 184 + ], + [ + 245, + 204 + ], + [ + 255, + 193 + ], + [ + 273, + 218 + ], + [ + 280, + 205 + ] + ], + [ + [ + 268, + 181 + ], + [ + 294, + 192 + ], + [ + 276, + 159 + ], + [ + 261, + 202 + ], + [ + 250, + 147 + ], + [ + 236, + 191 + ] + ], + [ + [ + 263, + 160 + ], + [ + 278, + 152 + ], + [ + 273, + 139 + ], + [ + 253, + 180 + ], + [ + 255, + 147 + ], + [ + 236, + 190 + ] + ] + ] + }, + { + "task_id": "0002286", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0006_cfg_0004_036422060909#521", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002286/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 229, + 142 + ], + [ + 252, + 156 + ], + [ + 215, + 165 + ], + [ + 243, + 119 + ], + [ + 191, + 150 + ], + [ + 220, + 104 + ] + ], + [ + [ + 211, + 129 + ], + [ + 232, + 117 + ], + [ + 224, + 152 + ], + [ + 196, + 105 + ], + [ + 203, + 166 + ], + [ + 173, + 118 + ] + ], + [ + [ + 217, + 159 + ], + [ + 215, + 186 + ], + [ + 206, + 156 + ], + [ + 229, + 163 + ], + [ + 208, + 130 + ], + [ + 231, + 135 + ] + ], + [ + [ + 228, + 150 + ], + [ + 228, + 134 + ], + [ + 226, + 129 + ], + [ + 230, + 172 + ], + [ + 226, + 145 + ], + [ + 230, + 187 + ] + ] + ] + }, + { + "task_id": "0002287", + "id": "RH20T_cfg7_task_0015_user_0020_scene_0007_cfg_0007_104122061330#224", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002287/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 380, + 125 + ], + [ + 382, + 150 + ], + [ + 383, + 127 + ], + [ + 376, + 123 + ], + [ + 381, + 100 + ], + [ + 374, + 98 + ] + ], + [ + [ + 356, + 119 + ], + [ + 341, + 121 + ], + [ + 377, + 125 + ], + [ + 334, + 113 + ], + [ + 394, + 124 + ], + [ + 350, + 112 + ] + ], + [ + [ + 368, + 88 + ], + [ + 348, + 86 + ], + [ + 352, + 102 + ], + [ + 383, + 76 + ], + [ + 374, + 105 + ], + [ + 405, + 77 + ] + ], + [ + [ + 353, + 109 + ], + [ + 351, + 133 + ], + [ + 366, + 105 + ], + [ + 338, + 113 + ], + [ + 369, + 80 + ], + [ + 341, + 86 + ] + ] + ] + }, + { + "task_id": "0002288", + "id": "RH20T_cfg6_task_0013_user_0014_scene_0007_cfg_0006_104122063633#124", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002288/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 286, + 198 + ], + [ + 263, + 203 + ], + [ + 282, + 191 + ], + [ + 290, + 206 + ], + [ + 303, + 187 + ], + [ + 313, + 201 + ] + ], + [ + [ + 272, + 203 + ], + [ + 285, + 201 + ], + [ + 260, + 221 + ], + [ + 285, + 185 + ], + [ + 246, + 225 + ], + [ + 271, + 187 + ] + ], + [ + [ + 270, + 188 + ], + [ + 269, + 168 + ], + [ + 271, + 177 + ], + [ + 269, + 201 + ], + [ + 272, + 196 + ], + [ + 269, + 221 + ] + ], + [ + [ + 262, + 200 + ], + [ + 284, + 200 + ], + [ + 255, + 195 + ], + [ + 270, + 206 + ], + [ + 233, + 194 + ], + [ + 246, + 205 + ] + ] + ] + }, + { + "task_id": "0002290", + "id": "RH20T_cfg1_task_0016_user_0001_scene_0004_cfg_0001_035622060973#188", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002290/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 280, + 79 + ], + [ + 280, + 67 + ], + [ + 274, + 54 + ], + [ + 286, + 102 + ], + [ + 273, + 67 + ], + [ + 286, + 116 + ] + ], + [ + [ + 272, + 100 + ], + [ + 263, + 118 + ], + [ + 289, + 116 + ], + [ + 253, + 83 + ], + [ + 296, + 99 + ], + [ + 262, + 66 + ] + ], + [ + [ + 261, + 99 + ], + [ + 268, + 103 + ], + [ + 238, + 109 + ], + [ + 285, + 89 + ], + [ + 230, + 105 + ], + [ + 279, + 84 + ] + ], + [ + [ + 282, + 96 + ], + [ + 289, + 81 + ], + [ + 257, + 98 + ], + [ + 305, + 93 + ], + [ + 251, + 112 + ], + [ + 297, + 106 + ] + ] + ] + }, + { + "task_id": "0002291", + "id": "RH20T_cfg4_task_0012_user_0010_scene_0001_cfg_0004_104122062295#102", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push the soccer ball into the goal'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002291/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 214, + 243 + ], + [ + 236, + 201 + ], + [ + 181, + 214 + ], + [ + 244, + 269 + ], + [ + 152, + 264 + ], + [ + 223, + 315 + ] + ], + [ + [ + 230, + 262 + ], + [ + 229, + 227 + ], + [ + 180, + 267 + ], + [ + 277, + 258 + ], + [ + 183, + 298 + ], + [ + 275, + 289 + ] + ], + [ + [ + 253, + 266 + ], + [ + 240, + 264 + ], + [ + 298, + 262 + ], + [ + 203, + 270 + ], + [ + 318, + 263 + ], + [ + 213, + 273 + ] + ], + [ + [ + 245, + 271 + ], + [ + 242, + 241 + ], + [ + 206, + 259 + ], + [ + 281, + 283 + ], + [ + 206, + 294 + ], + [ + 287, + 316 + ] + ] + ] + }, + { + "task_id": "0002293", + "id": "RH20T_cfg7_task_0044_user_0020_scene_0001_cfg_0007_104122060811#126", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002293/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 386, + 136 + ], + [ + 391, + 148 + ], + [ + 400, + 129 + ], + [ + 373, + 143 + ], + [ + 395, + 117 + ], + [ + 369, + 131 + ] + ], + [ + [ + 395, + 151 + ], + [ + 407, + 157 + ], + [ + 388, + 163 + ], + [ + 402, + 138 + ], + [ + 376, + 157 + ], + [ + 390, + 131 + ] + ], + [ + [ + 391, + 145 + ], + [ + 387, + 131 + ], + [ + 377, + 149 + ], + [ + 405, + 141 + ], + [ + 381, + 163 + ], + [ + 409, + 155 + ] + ], + [ + [ + 390, + 142 + ], + [ + 396, + 135 + ], + [ + 403, + 139 + ], + [ + 378, + 145 + ], + [ + 398, + 145 + ], + [ + 373, + 151 + ] + ] + ] + }, + { + "task_id": "0002294", + "id": "RH20T_cfg5_task_0010_user_0010_scene_0002_cfg_0005_105422061350#330", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002294/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 96 + ], + [ + 333, + 83 + ], + [ + 361, + 100 + ], + [ + 315, + 91 + ], + [ + 367, + 114 + ], + [ + 319, + 105 + ] + ], + [ + [ + 329, + 80 + ], + [ + 306, + 74 + ], + [ + 328, + 100 + ], + [ + 331, + 59 + ], + [ + 351, + 106 + ], + [ + 356, + 65 + ] + ], + [ + [ + 319, + 93 + ], + [ + 327, + 77 + ], + [ + 321, + 75 + ], + [ + 316, + 114 + ], + [ + 313, + 90 + ], + [ + 309, + 128 + ] + ], + [ + [ + 315, + 78 + ], + [ + 316, + 97 + ], + [ + 327, + 93 + ], + [ + 303, + 65 + ], + [ + 325, + 73 + ], + [ + 301, + 45 + ] + ] + ] + }, + { + "task_id": "0002296", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_036422060909#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002296/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 109 + ], + [ + 316, + 86 + ], + [ + 335, + 93 + ], + [ + 303, + 126 + ], + [ + 337, + 114 + ], + [ + 306, + 148 + ] + ], + [ + [ + 297, + 137 + ], + [ + 282, + 120 + ], + [ + 320, + 122 + ], + [ + 275, + 152 + ], + [ + 336, + 139 + ], + [ + 290, + 171 + ] + ], + [ + [ + 346, + 82 + ], + [ + 351, + 54 + ], + [ + 328, + 71 + ], + [ + 362, + 93 + ], + [ + 323, + 101 + ], + [ + 358, + 120 + ] + ], + [ + [ + 325, + 130 + ], + [ + 297, + 133 + ], + [ + 324, + 126 + ], + [ + 326, + 134 + ], + [ + 354, + 122 + ], + [ + 353, + 131 + ] + ] + ] + }, + { + "task_id": "0002300", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0007_cfg_0004_104422070011#479", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002300/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 426, + 145 + ], + [ + 439, + 132 + ], + [ + 439, + 166 + ], + [ + 413, + 124 + ], + [ + 426, + 177 + ], + [ + 402, + 136 + ] + ], + [ + [ + 418, + 131 + ], + [ + 400, + 138 + ], + [ + 412, + 108 + ], + [ + 424, + 154 + ], + [ + 432, + 100 + ], + [ + 445, + 148 + ] + ], + [ + [ + 404, + 154 + ], + [ + 399, + 133 + ], + [ + 398, + 145 + ], + [ + 411, + 165 + ], + [ + 403, + 164 + ], + [ + 416, + 186 + ] + ], + [ + [ + 422, + 142 + ], + [ + 418, + 127 + ], + [ + 437, + 126 + ], + [ + 407, + 158 + ], + [ + 440, + 141 + ], + [ + 411, + 172 + ] + ] + ] + }, + { + "task_id": "0002301", + "id": "RH20T_cfg1_task_0029_user_0001_scene_0004_cfg_0001_035622060973#209", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002301/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 234, + 230 + ], + [ + 226, + 224 + ], + [ + 224, + 212 + ], + [ + 245, + 249 + ], + [ + 232, + 219 + ], + [ + 252, + 254 + ] + ], + [ + [ + 229, + 207 + ], + [ + 244, + 195 + ], + [ + 235, + 228 + ], + [ + 224, + 186 + ], + [ + 221, + 238 + ], + [ + 210, + 197 + ] + ], + [ + [ + 236, + 224 + ], + [ + 252, + 224 + ], + [ + 230, + 202 + ], + [ + 242, + 246 + ], + [ + 212, + 201 + ], + [ + 224, + 247 + ] + ], + [ + [ + 236, + 230 + ], + [ + 238, + 238 + ], + [ + 233, + 251 + ], + [ + 238, + 209 + ], + [ + 231, + 243 + ], + [ + 237, + 203 + ] + ] + ] + }, + { + "task_id": "0002302", + "id": "RH20T_cfg7_task_0013_user_0020_scene_0004_cfg_0007_104122061018#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002302/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 322, + 177 + ], + [ + 295, + 188 + ], + [ + 313, + 147 + ], + [ + 331, + 206 + ], + [ + 340, + 137 + ], + [ + 357, + 195 + ] + ], + [ + [ + 323, + 183 + ], + [ + 326, + 159 + ], + [ + 310, + 162 + ], + [ + 337, + 205 + ], + [ + 308, + 184 + ], + [ + 334, + 226 + ] + ], + [ + [ + 357, + 213 + ], + [ + 370, + 239 + ], + [ + 343, + 205 + ], + [ + 370, + 219 + ], + [ + 331, + 179 + ], + [ + 359, + 195 + ] + ], + [ + [ + 333, + 176 + ], + [ + 324, + 189 + ], + [ + 361, + 178 + ], + [ + 304, + 174 + ], + [ + 372, + 164 + ], + [ + 312, + 159 + ] + ] + ] + }, + { + "task_id": "0002303", + "id": "RH20T_cfg1_task_0001_user_0014_scene_0006_cfg_0001_035622060973#213", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002303/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 289, + 225 + ], + [ + 290, + 214 + ], + [ + 275, + 209 + ], + [ + 303, + 242 + ], + [ + 275, + 220 + ], + [ + 302, + 252 + ] + ], + [ + [ + 294, + 220 + ], + [ + 283, + 223 + ], + [ + 274, + 223 + ], + [ + 313, + 218 + ], + [ + 286, + 220 + ], + [ + 326, + 215 + ] + ], + [ + [ + 295, + 213 + ], + [ + 295, + 191 + ], + [ + 314, + 215 + ], + [ + 276, + 211 + ], + [ + 314, + 237 + ], + [ + 276, + 232 + ] + ], + [ + [ + 287, + 229 + ], + [ + 275, + 232 + ], + [ + 269, + 235 + ], + [ + 306, + 224 + ], + [ + 281, + 232 + ], + [ + 317, + 221 + ] + ] + ] + }, + { + "task_id": "0002306", + "id": "RH20T_cfg3_task_0091_user_0010_scene_0006_cfg_0003_036422060909#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002306/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 453, + 172 + ], + [ + 477, + 175 + ], + [ + 459, + 145 + ], + [ + 446, + 197 + ], + [ + 434, + 141 + ], + [ + 422, + 194 + ] + ], + [ + [ + 433, + 185 + ], + [ + 443, + 208 + ], + [ + 408, + 188 + ], + [ + 457, + 182 + ], + [ + 399, + 167 + ], + [ + 446, + 161 + ] + ], + [ + [ + 443, + 124 + ], + [ + 470, + 134 + ], + [ + 454, + 98 + ], + [ + 433, + 149 + ], + [ + 427, + 89 + ], + [ + 407, + 140 + ] + ], + [ + [ + 454, + 164 + ], + [ + 459, + 146 + ], + [ + 462, + 145 + ], + [ + 445, + 184 + ], + [ + 457, + 162 + ], + [ + 441, + 200 + ] + ] + ] + }, + { + "task_id": "0002307", + "id": "RH20T_cfg3_task_0091_user_0010_scene_0006_cfg_0003_036422060909#244", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002307/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 474, + 165 + ], + [ + 474, + 146 + ], + [ + 483, + 148 + ], + [ + 465, + 183 + ], + [ + 483, + 167 + ], + [ + 466, + 201 + ] + ], + [ + [ + 491, + 160 + ], + [ + 514, + 172 + ], + [ + 477, + 181 + ], + [ + 505, + 137 + ], + [ + 455, + 169 + ], + [ + 481, + 125 + ] + ], + [ + [ + 481, + 178 + ], + [ + 455, + 178 + ], + [ + 482, + 181 + ], + [ + 480, + 176 + ], + [ + 512, + 182 + ], + [ + 507, + 177 + ] + ], + [ + [ + 469, + 171 + ], + [ + 457, + 149 + ], + [ + 494, + 160 + ], + [ + 445, + 182 + ], + [ + 506, + 184 + ], + [ + 457, + 205 + ] + ] + ] + }, + { + "task_id": "0002308", + "id": "RH20T_cfg4_task_0083_user_0015_scene_0004_cfg_0004_104122063550#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Swat with a flyswatter'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002308/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 341, + 220 + ], + [ + 356, + 200 + ], + [ + 362, + 235 + ], + [ + 320, + 206 + ], + [ + 347, + 256 + ], + [ + 305, + 226 + ] + ], + [ + [ + 349, + 185 + ], + [ + 337, + 165 + ], + [ + 360, + 165 + ], + [ + 337, + 206 + ], + [ + 371, + 183 + ], + [ + 349, + 223 + ] + ], + [ + [ + 385, + 225 + ], + [ + 400, + 205 + ], + [ + 382, + 211 + ], + [ + 388, + 237 + ], + [ + 365, + 232 + ], + [ + 373, + 257 + ] + ], + [ + [ + 363, + 203 + ], + [ + 365, + 193 + ], + [ + 371, + 228 + ], + [ + 354, + 180 + ], + [ + 369, + 236 + ], + [ + 353, + 190 + ] + ] + ] + }, + { + "task_id": "0002310", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0007_cfg_0004_104122063550#94", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002310/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 353, + 129 + ], + [ + 344, + 153 + ], + [ + 368, + 127 + ], + [ + 339, + 130 + ], + [ + 378, + 103 + ], + [ + 348, + 108 + ] + ], + [ + [ + 354, + 131 + ], + [ + 348, + 156 + ], + [ + 349, + 121 + ], + [ + 359, + 142 + ], + [ + 354, + 96 + ], + [ + 365, + 115 + ] + ], + [ + [ + 369, + 159 + ], + [ + 350, + 174 + ], + [ + 379, + 178 + ], + [ + 359, + 139 + ], + [ + 398, + 164 + ], + [ + 378, + 124 + ] + ], + [ + [ + 357, + 157 + ], + [ + 361, + 143 + ], + [ + 378, + 146 + ], + [ + 335, + 169 + ], + [ + 373, + 158 + ], + [ + 332, + 182 + ] + ] + ] + }, + { + "task_id": "0002311", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0007_cfg_0004_104122063550#206", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002311/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 286, + 93 + ], + [ + 286, + 78 + ], + [ + 312, + 87 + ], + [ + 261, + 98 + ], + [ + 311, + 100 + ], + [ + 263, + 112 + ] + ], + [ + [ + 268, + 99 + ], + [ + 273, + 126 + ], + [ + 295, + 94 + ], + [ + 240, + 104 + ], + [ + 290, + 66 + ], + [ + 235, + 76 + ] + ], + [ + [ + 294, + 99 + ], + [ + 288, + 84 + ], + [ + 295, + 78 + ], + [ + 293, + 118 + ], + [ + 301, + 94 + ], + [ + 298, + 134 + ] + ], + [ + [ + 272, + 110 + ], + [ + 299, + 112 + ], + [ + 270, + 137 + ], + [ + 274, + 82 + ], + [ + 242, + 135 + ], + [ + 246, + 79 + ] + ] + ] + }, + { + "task_id": "0002312", + "id": "RH20T_cfg2_task_0042_user_0007_scene_0002_cfg_0002_105422061350#243", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002312/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 342, + 135 + ], + [ + 345, + 123 + ], + [ + 365, + 118 + ], + [ + 321, + 151 + ], + [ + 363, + 131 + ], + [ + 317, + 165 + ] + ], + [ + [ + 288, + 139 + ], + [ + 301, + 115 + ], + [ + 267, + 134 + ], + [ + 308, + 143 + ], + [ + 254, + 158 + ], + [ + 296, + 165 + ] + ], + [ + [ + 320, + 141 + ], + [ + 316, + 133 + ], + [ + 340, + 122 + ], + [ + 300, + 160 + ], + [ + 343, + 131 + ], + [ + 305, + 167 + ] + ], + [ + [ + 356, + 139 + ], + [ + 363, + 128 + ], + [ + 335, + 121 + ], + [ + 377, + 157 + ], + [ + 331, + 131 + ], + [ + 370, + 165 + ] + ] + ] + }, + { + "task_id": "0002313", + "id": "RH20T_cfg7_task_0025_user_0020_scene_0008_cfg_0007_104122064161#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold a piece of paper'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002313/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 385, + 186 + ], + [ + 389, + 210 + ], + [ + 402, + 181 + ], + [ + 369, + 190 + ], + [ + 398, + 157 + ], + [ + 365, + 168 + ] + ], + [ + [ + 366, + 189 + ], + [ + 367, + 187 + ], + [ + 362, + 212 + ], + [ + 369, + 166 + ], + [ + 361, + 212 + ], + [ + 368, + 168 + ] + ], + [ + [ + 370, + 207 + ], + [ + 370, + 217 + ], + [ + 391, + 215 + ], + [ + 349, + 199 + ], + [ + 392, + 204 + ], + [ + 348, + 187 + ] + ], + [ + [ + 352, + 209 + ], + [ + 334, + 217 + ], + [ + 339, + 213 + ], + [ + 364, + 205 + ], + [ + 358, + 204 + ], + [ + 382, + 197 + ] + ] + ] + }, + { + "task_id": "0002314", + "id": "RH20T_cfg3_task_0092_user_0010_scene_0007_cfg_0003_104122062823#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the hands of a clock'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002314/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 179, + 76 + ], + [ + 178, + 48 + ], + [ + 202, + 87 + ], + [ + 156, + 65 + ], + [ + 201, + 112 + ], + [ + 157, + 91 + ] + ], + [ + [ + 179, + 96 + ], + [ + 169, + 95 + ], + [ + 195, + 73 + ], + [ + 162, + 120 + ], + [ + 206, + 73 + ], + [ + 171, + 122 + ] + ], + [ + [ + 139, + 101 + ], + [ + 111, + 89 + ], + [ + 152, + 106 + ], + [ + 126, + 97 + ], + [ + 178, + 117 + ], + [ + 151, + 108 + ] + ], + [ + [ + 197, + 92 + ], + [ + 203, + 68 + ], + [ + 211, + 81 + ], + [ + 181, + 104 + ], + [ + 205, + 102 + ], + [ + 176, + 126 + ] + ] + ] + }, + { + "task_id": "0002316", + "id": "RH20T_cfg3_task_0008_user_0016_scene_0008_cfg_0003_104122062823#196", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002316/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 420, + 64 + ], + [ + 418, + 85 + ], + [ + 429, + 36 + ], + [ + 412, + 88 + ], + [ + 430, + 17 + ], + [ + 414, + 69 + ] + ], + [ + [ + 425, + 106 + ], + [ + 430, + 79 + ], + [ + 457, + 114 + ], + [ + 393, + 99 + ], + [ + 450, + 139 + ], + [ + 390, + 123 + ] + ], + [ + [ + 420, + 85 + ], + [ + 426, + 100 + ], + [ + 451, + 65 + ], + [ + 392, + 105 + ], + [ + 444, + 53 + ], + [ + 389, + 91 + ] + ], + [ + [ + 422, + 116 + ], + [ + 414, + 130 + ], + [ + 395, + 100 + ], + [ + 450, + 131 + ], + [ + 403, + 89 + ], + [ + 456, + 117 + ] + ] + ] + }, + { + "task_id": "0002317", + "id": "RH20T_cfg2_task_0041_user_0007_scene_0009_cfg_0002_104122061850#482", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the tabletop with a sponge'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002317/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 322, + 320 + ], + [ + 318, + 312 + ], + [ + 308, + 348 + ], + [ + 335, + 295 + ], + [ + 313, + 354 + ], + [ + 338, + 304 + ] + ], + [ + [ + 337, + 328 + ], + [ + 365, + 335 + ], + [ + 348, + 320 + ], + [ + 325, + 338 + ], + [ + 323, + 314 + ], + [ + 298, + 331 + ] + ], + [ + [ + 329, + 327 + ], + [ + 338, + 301 + ], + [ + 352, + 343 + ], + [ + 307, + 312 + ], + [ + 343, + 360 + ], + [ + 299, + 337 + ] + ], + [ + [ + 303, + 303 + ], + [ + 275, + 301 + ], + [ + 308, + 277 + ], + [ + 297, + 332 + ], + [ + 335, + 279 + ], + [ + 327, + 335 + ] + ] + ] + }, + { + "task_id": "0002319", + "id": "RH20T_cfg5_task_0001_user_0014_scene_0002_cfg_0005_105422061350#179", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002319/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 229, + 84 + ], + [ + 258, + 86 + ], + [ + 230, + 54 + ], + [ + 228, + 113 + ], + [ + 200, + 52 + ], + [ + 199, + 111 + ] + ], + [ + [ + 232, + 71 + ], + [ + 214, + 83 + ], + [ + 252, + 95 + ], + [ + 211, + 47 + ], + [ + 267, + 84 + ], + [ + 228, + 38 + ] + ], + [ + [ + 228, + 93 + ], + [ + 210, + 74 + ], + [ + 226, + 109 + ], + [ + 231, + 74 + ], + [ + 243, + 127 + ], + [ + 250, + 93 + ] + ], + [ + [ + 217, + 87 + ], + [ + 217, + 67 + ], + [ + 225, + 65 + ], + [ + 209, + 110 + ], + [ + 225, + 83 + ], + [ + 210, + 127 + ] + ] + ] + }, + { + "task_id": "0002320", + "id": "RH20T_cfg3_task_0038_user_0010_scene_0003_cfg_0003_104422070011#149", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002320/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 276, + 86 + ], + [ + 287, + 59 + ], + [ + 267, + 86 + ], + [ + 284, + 86 + ], + [ + 255, + 114 + ], + [ + 273, + 110 + ] + ], + [ + [ + 289, + 102 + ], + [ + 296, + 87 + ], + [ + 263, + 107 + ], + [ + 313, + 99 + ], + [ + 257, + 120 + ], + [ + 306, + 112 + ] + ], + [ + [ + 291, + 105 + ], + [ + 269, + 95 + ], + [ + 305, + 113 + ], + [ + 278, + 98 + ], + [ + 326, + 122 + ], + [ + 298, + 108 + ] + ], + [ + [ + 318, + 95 + ], + [ + 322, + 117 + ], + [ + 292, + 91 + ], + [ + 342, + 100 + ], + [ + 288, + 70 + ], + [ + 337, + 80 + ] + ] + ] + }, + { + "task_id": "0002321", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_038522062288#157", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002321/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 144, + 167 + ], + [ + 137, + 146 + ], + [ + 122, + 155 + ], + [ + 167, + 179 + ], + [ + 130, + 175 + ], + [ + 173, + 198 + ] + ], + [ + [ + 142, + 150 + ], + [ + 159, + 132 + ], + [ + 123, + 127 + ], + [ + 160, + 171 + ], + [ + 104, + 146 + ], + [ + 144, + 190 + ] + ], + [ + [ + 164, + 195 + ], + [ + 153, + 172 + ], + [ + 171, + 207 + ], + [ + 156, + 180 + ], + [ + 181, + 230 + ], + [ + 167, + 205 + ] + ], + [ + [ + 161, + 184 + ], + [ + 179, + 163 + ], + [ + 136, + 164 + ], + [ + 185, + 203 + ], + [ + 118, + 185 + ], + [ + 168, + 223 + ] + ] + ] + }, + { + "task_id": "0002323", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_038522062288#338", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002323/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 167, + 222 + ], + [ + 139, + 243 + ], + [ + 179, + 222 + ], + [ + 155, + 222 + ], + [ + 206, + 201 + ], + [ + 180, + 203 + ] + ], + [ + [ + 166, + 260 + ], + [ + 201, + 255 + ], + [ + 172, + 294 + ], + [ + 160, + 225 + ], + [ + 138, + 299 + ], + [ + 125, + 230 + ] + ], + [ + [ + 199, + 240 + ], + [ + 193, + 259 + ], + [ + 166, + 229 + ], + [ + 231, + 251 + ], + [ + 173, + 212 + ], + [ + 234, + 233 + ] + ], + [ + [ + 176, + 237 + ], + [ + 172, + 214 + ], + [ + 152, + 222 + ], + [ + 202, + 252 + ], + [ + 157, + 243 + ], + [ + 204, + 272 + ] + ] + ] + }, + { + "task_id": "0002325", + "id": "RH20T_cfg1_task_0108_user_0014_scene_0004_cfg_0001_038522063145#121", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Sharpen the pencil with a pencil sharpener'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002325/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 388, + 97 + ], + [ + 393, + 77 + ], + [ + 410, + 98 + ], + [ + 366, + 96 + ], + [ + 405, + 115 + ], + [ + 362, + 114 + ] + ], + [ + [ + 403, + 105 + ], + [ + 413, + 117 + ], + [ + 396, + 123 + ], + [ + 410, + 87 + ], + [ + 386, + 110 + ], + [ + 400, + 73 + ] + ], + [ + [ + 368, + 75 + ], + [ + 359, + 97 + ], + [ + 379, + 83 + ], + [ + 356, + 67 + ], + [ + 388, + 61 + ], + [ + 366, + 44 + ] + ], + [ + [ + 375, + 97 + ], + [ + 368, + 108 + ], + [ + 353, + 87 + ], + [ + 397, + 107 + ], + [ + 360, + 74 + ], + [ + 406, + 96 + ] + ] + ] + }, + { + "task_id": "0002327", + "id": "RH20T_cfg1_task_0064_user_0015_scene_0010_cfg_0001_038522062547#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002327/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 459, + 85 + ], + [ + 458, + 77 + ], + [ + 484, + 95 + ], + [ + 434, + 75 + ], + [ + 488, + 104 + ], + [ + 435, + 84 + ] + ], + [ + [ + 430, + 82 + ], + [ + 403, + 79 + ], + [ + 430, + 57 + ], + [ + 431, + 104 + ], + [ + 459, + 60 + ], + [ + 458, + 108 + ] + ], + [ + [ + 434, + 87 + ], + [ + 445, + 77 + ], + [ + 453, + 106 + ], + [ + 415, + 68 + ], + [ + 442, + 114 + ], + [ + 406, + 78 + ] + ], + [ + [ + 425, + 108 + ], + [ + 439, + 91 + ], + [ + 443, + 125 + ], + [ + 407, + 90 + ], + [ + 430, + 140 + ], + [ + 395, + 106 + ] + ] + ] + }, + { + "task_id": "0002328", + "id": "RH20T_cfg1_task_0045_user_0015_scene_0002_cfg_0001_038522063145#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the brush on the pen rack'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002328/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 391, + 111 + ], + [ + 396, + 94 + ], + [ + 411, + 111 + ], + [ + 371, + 111 + ], + [ + 407, + 127 + ], + [ + 367, + 127 + ] + ], + [ + [ + 378, + 132 + ], + [ + 359, + 132 + ], + [ + 381, + 112 + ], + [ + 376, + 151 + ], + [ + 401, + 112 + ], + [ + 396, + 151 + ] + ], + [ + [ + 414, + 112 + ], + [ + 399, + 98 + ], + [ + 402, + 128 + ], + [ + 426, + 94 + ], + [ + 418, + 142 + ], + [ + 443, + 108 + ] + ], + [ + [ + 398, + 102 + ], + [ + 404, + 106 + ], + [ + 382, + 114 + ], + [ + 415, + 89 + ], + [ + 377, + 110 + ], + [ + 409, + 87 + ] + ] + ] + }, + { + "task_id": "0002329", + "id": "RH20T_cfg4_task_0091_user_0014_scene_0007_cfg_0004_104122062823#142", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002329/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 295, + 179 + ], + [ + 295, + 142 + ], + [ + 337, + 173 + ], + [ + 251, + 186 + ], + [ + 335, + 206 + ], + [ + 253, + 220 + ] + ], + [ + [ + 261, + 192 + ], + [ + 258, + 177 + ], + [ + 251, + 148 + ], + [ + 269, + 234 + ], + [ + 253, + 162 + ], + [ + 272, + 256 + ] + ], + [ + [ + 310, + 189 + ], + [ + 341, + 187 + ], + [ + 290, + 170 + ], + [ + 334, + 210 + ], + [ + 256, + 172 + ], + [ + 298, + 214 + ] + ], + [ + [ + 318, + 170 + ], + [ + 298, + 139 + ], + [ + 287, + 167 + ], + [ + 344, + 172 + ], + [ + 309, + 203 + ], + [ + 365, + 203 + ] + ] + ] + }, + { + "task_id": "0002330", + "id": "RH20T_cfg2_task_0071_user_0014_scene_0003_cfg_0002_037522062165#184", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002330/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 187, + 238 + ], + [ + 182, + 223 + ], + [ + 174, + 221 + ], + [ + 200, + 257 + ], + [ + 179, + 235 + ], + [ + 204, + 270 + ] + ], + [ + [ + 173, + 257 + ], + [ + 148, + 243 + ], + [ + 180, + 239 + ], + [ + 166, + 276 + ], + [ + 203, + 251 + ], + [ + 191, + 289 + ] + ], + [ + [ + 187, + 227 + ], + [ + 197, + 226 + ], + [ + 160, + 222 + ], + [ + 212, + 232 + ], + [ + 151, + 224 + ], + [ + 202, + 233 + ] + ], + [ + [ + 177, + 223 + ], + [ + 200, + 209 + ], + [ + 174, + 224 + ], + [ + 179, + 223 + ], + [ + 152, + 237 + ], + [ + 154, + 238 + ] + ] + ] + }, + { + "task_id": "0002332", + "id": "RH20T_cfg2_task_0015_user_0006_scene_0003_cfg_0002_104122061850#356", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002332/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 252, + 226 + ], + [ + 260, + 241 + ], + [ + 248, + 246 + ], + [ + 255, + 206 + ], + [ + 242, + 232 + ], + [ + 248, + 193 + ] + ], + [ + [ + 240, + 227 + ], + [ + 243, + 217 + ], + [ + 253, + 246 + ], + [ + 227, + 209 + ], + [ + 250, + 255 + ], + [ + 225, + 219 + ] + ], + [ + [ + 243, + 229 + ], + [ + 220, + 234 + ], + [ + 249, + 213 + ], + [ + 238, + 245 + ], + [ + 271, + 209 + ], + [ + 260, + 240 + ] + ], + [ + [ + 256, + 207 + ], + [ + 247, + 195 + ], + [ + 234, + 212 + ], + [ + 279, + 202 + ], + [ + 243, + 223 + ], + [ + 286, + 213 + ] + ] + ] + }, + { + "task_id": "0002335", + "id": "RH20T_cfg4_task_0064_user_0016_scene_0003_cfg_0004_036422060909#191", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002335/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 340, + 90 + ], + [ + 364, + 94 + ], + [ + 332, + 107 + ], + [ + 348, + 75 + ], + [ + 309, + 103 + ], + [ + 326, + 72 + ] + ], + [ + [ + 350, + 68 + ], + [ + 342, + 77 + ], + [ + 376, + 74 + ], + [ + 325, + 62 + ], + [ + 381, + 65 + ], + [ + 332, + 53 + ] + ], + [ + [ + 346, + 78 + ], + [ + 353, + 58 + ], + [ + 333, + 61 + ], + [ + 359, + 95 + ], + [ + 326, + 80 + ], + [ + 351, + 113 + ] + ], + [ + [ + 339, + 79 + ], + [ + 314, + 78 + ], + [ + 337, + 60 + ], + [ + 341, + 99 + ], + [ + 361, + 61 + ], + [ + 366, + 99 + ] + ] + ] + }, + { + "task_id": "0002336", + "id": "RH20T_cfg1_task_0077_user_0010_scene_0006_cfg_0001_750612070853#229", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002336/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 290, + 342 + ], + [ + 247, + 325 + ], + [ + 320, + 311 + ], + [ + 262, + 360 + ], + [ + 359, + 328 + ], + [ + 302, + 360 + ] + ], + [ + [ + 264, + 349 + ], + [ + 295, + 311 + ], + [ + 229, + 313 + ], + [ + 300, + 360 + ], + [ + 191, + 353 + ], + [ + 267, + 360 + ] + ], + [ + [ + 241, + 281 + ], + [ + 247, + 283 + ], + [ + 220, + 235 + ], + [ + 262, + 326 + ], + [ + 216, + 237 + ], + [ + 255, + 320 + ] + ], + [ + [ + 233, + 318 + ], + [ + 222, + 283 + ], + [ + 208, + 290 + ], + [ + 262, + 351 + ], + [ + 218, + 321 + ], + [ + 271, + 360 + ] + ] + ] + }, + { + "task_id": "0002337", + "id": "RH20T_cfg5_task_0006_user_0010_scene_0009_cfg_0005_104122063678#230", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002337/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 273, + 299 + ], + [ + 257, + 293 + ], + [ + 298, + 284 + ], + [ + 246, + 314 + ], + [ + 317, + 289 + ], + [ + 263, + 322 + ] + ], + [ + [ + 274, + 332 + ], + [ + 291, + 334 + ], + [ + 250, + 312 + ], + [ + 300, + 354 + ], + [ + 230, + 309 + ], + [ + 282, + 353 + ] + ], + [ + [ + 280, + 275 + ], + [ + 263, + 302 + ], + [ + 263, + 272 + ], + [ + 295, + 278 + ], + [ + 280, + 243 + ], + [ + 310, + 252 + ] + ], + [ + [ + 291, + 312 + ], + [ + 297, + 296 + ], + [ + 297, + 343 + ], + [ + 286, + 284 + ], + [ + 291, + 357 + ], + [ + 281, + 299 + ] + ] + ] + }, + { + "task_id": "0002338", + "id": "RH20T_cfg5_task_0007_user_0010_scene_0007_cfg_0005_104122061850#272", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002338/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 216 + ], + [ + 280, + 204 + ], + [ + 272, + 239 + ], + [ + 295, + 193 + ], + [ + 276, + 248 + ], + [ + 298, + 205 + ] + ], + [ + [ + 277, + 222 + ], + [ + 252, + 215 + ], + [ + 266, + 230 + ], + [ + 287, + 215 + ], + [ + 293, + 238 + ], + [ + 311, + 222 + ] + ], + [ + [ + 292, + 208 + ], + [ + 303, + 209 + ], + [ + 276, + 190 + ], + [ + 308, + 228 + ], + [ + 263, + 188 + ], + [ + 297, + 228 + ] + ], + [ + [ + 312, + 217 + ], + [ + 320, + 206 + ], + [ + 295, + 233 + ], + [ + 328, + 202 + ], + [ + 288, + 242 + ], + [ + 321, + 213 + ] + ] + ] + }, + { + "task_id": "0002339", + "id": "RH20T_cfg1_task_0043_user_0015_scene_0004_cfg_0001_750612070851#170", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002339/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 350, + 186 + ], + [ + 352, + 163 + ], + [ + 326, + 211 + ], + [ + 372, + 163 + ], + [ + 325, + 231 + ], + [ + 369, + 184 + ] + ], + [ + [ + 362, + 199 + ], + [ + 366, + 227 + ], + [ + 390, + 217 + ], + [ + 331, + 179 + ], + [ + 385, + 193 + ], + [ + 329, + 155 + ] + ], + [ + [ + 370, + 178 + ], + [ + 367, + 145 + ], + [ + 333, + 181 + ], + [ + 407, + 175 + ], + [ + 335, + 216 + ], + [ + 411, + 211 + ] + ], + [ + [ + 367, + 216 + ], + [ + 349, + 201 + ], + [ + 397, + 199 + ], + [ + 335, + 234 + ], + [ + 419, + 214 + ], + [ + 354, + 253 + ] + ] + ] + }, + { + "task_id": "0002340", + "id": "RH20T_cfg1_task_0072_user_0010_scene_0010_cfg_0001_750612070851#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drop coins into a piggy bank'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002340/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 184, + 283 + ], + [ + 182, + 318 + ], + [ + 227, + 292 + ], + [ + 142, + 275 + ], + [ + 231, + 255 + ], + [ + 144, + 237 + ] + ], + [ + [ + 212, + 306 + ], + [ + 199, + 310 + ], + [ + 174, + 299 + ], + [ + 247, + 312 + ], + [ + 188, + 294 + ], + [ + 264, + 308 + ] + ], + [ + [ + 219, + 333 + ], + [ + 232, + 294 + ], + [ + 185, + 333 + ], + [ + 256, + 333 + ], + [ + 168, + 360 + ], + [ + 243, + 360 + ] + ], + [ + [ + 219, + 321 + ], + [ + 247, + 294 + ], + [ + 190, + 326 + ], + [ + 243, + 317 + ], + [ + 162, + 354 + ], + [ + 218, + 341 + ] + ] + ] + }, + { + "task_id": "0002342", + "id": "RH20T_cfg1_task_0072_user_0010_scene_0010_cfg_0001_750612070851#597", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drop coins into a piggy bank'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002342/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 222, + 302 + ], + [ + 213, + 347 + ], + [ + 265, + 309 + ], + [ + 179, + 295 + ], + [ + 273, + 266 + ], + [ + 189, + 251 + ] + ], + [ + [ + 252, + 316 + ], + [ + 284, + 322 + ], + [ + 268, + 277 + ], + [ + 236, + 355 + ], + [ + 233, + 268 + ], + [ + 199, + 351 + ] + ], + [ + [ + 208, + 329 + ], + [ + 174, + 354 + ], + [ + 234, + 347 + ], + [ + 179, + 309 + ], + [ + 266, + 324 + ], + [ + 215, + 282 + ] + ], + [ + [ + 215, + 321 + ], + [ + 244, + 289 + ], + [ + 193, + 323 + ], + [ + 234, + 319 + ], + [ + 161, + 358 + ], + [ + 208, + 348 + ] + ] + ] + }, + { + "task_id": "0002344", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122061330#224", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002344/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 272, + 122 + ], + [ + 260, + 117 + ], + [ + 264, + 145 + ], + [ + 279, + 98 + ], + [ + 275, + 148 + ], + [ + 290, + 103 + ] + ], + [ + [ + 296, + 115 + ], + [ + 293, + 103 + ], + [ + 319, + 122 + ], + [ + 273, + 108 + ], + [ + 324, + 134 + ], + [ + 275, + 120 + ] + ], + [ + [ + 284, + 103 + ], + [ + 308, + 103 + ], + [ + 288, + 114 + ], + [ + 280, + 90 + ], + [ + 265, + 114 + ], + [ + 255, + 90 + ] + ], + [ + [ + 278, + 113 + ], + [ + 266, + 118 + ], + [ + 288, + 94 + ], + [ + 267, + 132 + ], + [ + 302, + 88 + ], + [ + 281, + 128 + ] + ] + ] + }, + { + "task_id": "0002345", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122061330#642", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002345/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 372, + 101 + ], + [ + 363, + 100 + ], + [ + 365, + 125 + ], + [ + 378, + 77 + ], + [ + 374, + 125 + ], + [ + 386, + 79 + ] + ], + [ + [ + 361, + 111 + ], + [ + 366, + 90 + ], + [ + 377, + 120 + ], + [ + 345, + 101 + ], + [ + 372, + 141 + ], + [ + 340, + 122 + ] + ], + [ + [ + 384, + 103 + ], + [ + 377, + 126 + ], + [ + 375, + 113 + ], + [ + 394, + 92 + ], + [ + 381, + 92 + ], + [ + 401, + 69 + ] + ], + [ + [ + 377, + 112 + ], + [ + 356, + 104 + ], + [ + 389, + 101 + ], + [ + 367, + 122 + ], + [ + 411, + 110 + ], + [ + 387, + 130 + ] + ] + ] + }, + { + "task_id": "0002346", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122061330#986", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002346/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 379, + 180 + ], + [ + 354, + 168 + ], + [ + 364, + 194 + ], + [ + 394, + 166 + ], + [ + 392, + 208 + ], + [ + 421, + 177 + ] + ], + [ + [ + 384, + 207 + ], + [ + 407, + 198 + ], + [ + 381, + 182 + ], + [ + 387, + 233 + ], + [ + 360, + 191 + ], + [ + 365, + 241 + ] + ], + [ + [ + 360, + 207 + ], + [ + 356, + 198 + ], + [ + 349, + 234 + ], + [ + 370, + 181 + ], + [ + 353, + 241 + ], + [ + 373, + 191 + ] + ], + [ + [ + 345, + 174 + ], + [ + 344, + 153 + ], + [ + 316, + 179 + ], + [ + 374, + 168 + ], + [ + 318, + 198 + ], + [ + 373, + 187 + ] + ] + ] + }, + { + "task_id": "0002347", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122061330#1303", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002347/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 257, + 239 + ], + [ + 241, + 241 + ], + [ + 248, + 265 + ], + [ + 267, + 213 + ], + [ + 263, + 262 + ], + [ + 280, + 212 + ] + ], + [ + [ + 249, + 226 + ], + [ + 228, + 215 + ], + [ + 253, + 243 + ], + [ + 246, + 208 + ], + [ + 274, + 255 + ], + [ + 269, + 221 + ] + ], + [ + [ + 282, + 239 + ], + [ + 261, + 225 + ], + [ + 291, + 214 + ], + [ + 272, + 264 + ], + [ + 310, + 227 + ], + [ + 292, + 277 + ] + ], + [ + [ + 236, + 241 + ], + [ + 217, + 235 + ], + [ + 214, + 242 + ], + [ + 256, + 241 + ], + [ + 235, + 249 + ], + [ + 277, + 247 + ] + ] + ] + }, + { + "task_id": "0002348", + "id": "RH20T_cfg6_task_0091_user_0020_scene_0003_cfg_0006_104122064161#176", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002348/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 366, + 184 + ], + [ + 392, + 192 + ], + [ + 369, + 191 + ], + [ + 362, + 175 + ], + [ + 345, + 183 + ], + [ + 335, + 167 + ] + ], + [ + [ + 379, + 184 + ], + [ + 381, + 164 + ], + [ + 380, + 203 + ], + [ + 378, + 166 + ], + [ + 378, + 220 + ], + [ + 376, + 184 + ] + ], + [ + [ + 373, + 173 + ], + [ + 354, + 163 + ], + [ + 374, + 193 + ], + [ + 372, + 154 + ], + [ + 393, + 202 + ], + [ + 390, + 163 + ] + ], + [ + [ + 373, + 189 + ], + [ + 350, + 198 + ], + [ + 380, + 211 + ], + [ + 367, + 165 + ], + [ + 402, + 202 + ], + [ + 390, + 156 + ] + ] + ] + }, + { + "task_id": "0002351", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_104122064161#190", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002351/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 403, + 223 + ], + [ + 416, + 243 + ], + [ + 393, + 241 + ], + [ + 414, + 203 + ], + [ + 382, + 222 + ], + [ + 401, + 184 + ] + ], + [ + [ + 395, + 198 + ], + [ + 391, + 205 + ], + [ + 393, + 222 + ], + [ + 396, + 173 + ], + [ + 397, + 215 + ], + [ + 400, + 168 + ] + ], + [ + [ + 373, + 202 + ], + [ + 378, + 207 + ], + [ + 383, + 225 + ], + [ + 362, + 178 + ], + [ + 378, + 219 + ], + [ + 359, + 175 + ] + ], + [ + [ + 345, + 189 + ], + [ + 346, + 205 + ], + [ + 371, + 188 + ], + [ + 319, + 189 + ], + [ + 369, + 173 + ], + [ + 319, + 174 + ] + ] + ] + }, + { + "task_id": "0002353", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_104122064161#262", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002353/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 394, + 206 + ], + [ + 400, + 228 + ], + [ + 402, + 222 + ], + [ + 385, + 189 + ], + [ + 396, + 203 + ], + [ + 380, + 170 + ] + ], + [ + [ + 425, + 213 + ], + [ + 441, + 211 + ], + [ + 438, + 238 + ], + [ + 412, + 188 + ], + [ + 423, + 238 + ], + [ + 398, + 191 + ] + ], + [ + [ + 375, + 235 + ], + [ + 400, + 223 + ], + [ + 368, + 233 + ], + [ + 381, + 238 + ], + [ + 342, + 245 + ], + [ + 358, + 249 + ] + ], + [ + [ + 397, + 189 + ], + [ + 422, + 200 + ], + [ + 399, + 179 + ], + [ + 394, + 198 + ], + [ + 373, + 168 + ], + [ + 371, + 188 + ] + ] + ] + }, + { + "task_id": "0002354", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_104122064161#307", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002354/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 402, + 205 + ], + [ + 428, + 215 + ], + [ + 402, + 180 + ], + [ + 402, + 228 + ], + [ + 377, + 171 + ], + [ + 379, + 218 + ] + ], + [ + [ + 378, + 235 + ], + [ + 375, + 221 + ], + [ + 373, + 213 + ], + [ + 383, + 256 + ], + [ + 377, + 228 + ], + [ + 387, + 272 + ] + ], + [ + [ + 380, + 207 + ], + [ + 361, + 225 + ], + [ + 360, + 188 + ], + [ + 399, + 225 + ], + [ + 380, + 169 + ], + [ + 419, + 207 + ] + ], + [ + [ + 382, + 223 + ], + [ + 387, + 237 + ], + [ + 392, + 245 + ], + [ + 372, + 200 + ], + [ + 387, + 232 + ], + [ + 368, + 189 + ] + ] + ] + }, + { + "task_id": "0002358", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_104122062295#200", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002358/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 0, + 193 + ], + [ + 0, + 167 + ], + [ + 0, + 153 + ], + [ + 0, + 238 + ], + [ + 0, + 177 + ], + [ + 0, + 259 + ] + ], + [ + [ + 0, + 194 + ], + [ + 0, + 152 + ], + [ + 0, + 170 + ], + [ + 0, + 220 + ], + [ + 0, + 207 + ], + [ + 0, + 258 + ] + ], + [ + [ + 0, + 176 + ], + [ + 0, + 217 + ], + [ + 0, + 208 + ], + [ + 0, + 140 + ], + [ + 0, + 172 + ], + [ + 0, + 104 + ] + ], + [ + [ + 0, + 203 + ], + [ + 27, + 180 + ], + [ + 10, + 224 + ], + [ + 0, + 183 + ], + [ + 0, + 252 + ], + [ + 0, + 207 + ] + ] + ] + }, + { + "task_id": "0002360", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_104122062295#761", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002360/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 0, + 236 + ], + [ + 0, + 210 + ], + [ + 0, + 269 + ], + [ + 20, + 205 + ], + [ + 0, + 302 + ], + [ + 35, + 231 + ] + ], + [ + [ + 10, + 209 + ], + [ + 0, + 213 + ], + [ + 0, + 242 + ], + [ + 44, + 177 + ], + [ + 7, + 236 + ], + [ + 69, + 177 + ] + ], + [ + [ + 0, + 237 + ], + [ + 0, + 196 + ], + [ + 0, + 212 + ], + [ + 30, + 262 + ], + [ + 0, + 249 + ], + [ + 31, + 298 + ] + ], + [ + [ + 0, + 238 + ], + [ + 0, + 300 + ], + [ + 51, + 249 + ], + [ + 0, + 226 + ], + [ + 65, + 196 + ], + [ + 0, + 166 + ] + ] + ] + }, + { + "task_id": "0002361", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_104122062295#914", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002361/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 11, + 263 + ], + [ + 0, + 297 + ], + [ + 0, + 218 + ], + [ + 44, + 306 + ], + [ + 31, + 188 + ], + [ + 94, + 272 + ] + ], + [ + [ + 66, + 319 + ], + [ + 120, + 334 + ], + [ + 82, + 280 + ], + [ + 51, + 355 + ], + [ + 22, + 262 + ], + [ + 0, + 341 + ] + ], + [ + [ + 23, + 271 + ], + [ + 19, + 229 + ], + [ + 6, + 237 + ], + [ + 42, + 309 + ], + [ + 10, + 273 + ], + [ + 45, + 346 + ] + ], + [ + [ + 14, + 291 + ], + [ + 68, + 256 + ], + [ + 3, + 252 + ], + [ + 24, + 324 + ], + [ + 0, + 294 + ], + [ + 0, + 360 + ] + ] + ] + }, + { + "task_id": "0002363", + "id": "RH20T_cfg1_task_0070_user_0014_scene_0008_cfg_0001_038522063145#147", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002363/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 465, + 141 + ], + [ + 446, + 133 + ], + [ + 471, + 149 + ], + [ + 459, + 133 + ], + [ + 490, + 158 + ], + [ + 479, + 142 + ] + ], + [ + [ + 438, + 142 + ], + [ + 423, + 153 + ], + [ + 432, + 121 + ], + [ + 445, + 161 + ], + [ + 446, + 110 + ], + [ + 458, + 150 + ] + ], + [ + [ + 444, + 139 + ], + [ + 448, + 121 + ], + [ + 463, + 148 + ], + [ + 425, + 130 + ], + [ + 458, + 165 + ], + [ + 422, + 146 + ] + ], + [ + [ + 433, + 131 + ], + [ + 432, + 145 + ], + [ + 455, + 134 + ], + [ + 412, + 128 + ], + [ + 456, + 121 + ], + [ + 414, + 115 + ] + ] + ] + }, + { + "task_id": "0002369", + "id": "RH20T_cfg4_task_0077_user_0015_scene_0009_cfg_0004_104422070011#105", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002369/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 427, + 192 + ], + [ + 424, + 190 + ], + [ + 424, + 217 + ], + [ + 430, + 168 + ], + [ + 427, + 218 + ], + [ + 432, + 172 + ] + ], + [ + [ + 421, + 193 + ], + [ + 433, + 216 + ], + [ + 400, + 198 + ], + [ + 442, + 189 + ], + [ + 389, + 176 + ], + [ + 431, + 168 + ] + ], + [ + [ + 419, + 181 + ], + [ + 430, + 169 + ], + [ + 399, + 185 + ], + [ + 438, + 177 + ], + [ + 389, + 197 + ], + [ + 427, + 189 + ] + ], + [ + [ + 422, + 194 + ], + [ + 428, + 175 + ], + [ + 403, + 182 + ], + [ + 442, + 208 + ], + [ + 398, + 199 + ], + [ + 436, + 226 + ] + ] + ] + }, + { + "task_id": "0002370", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0008_cfg_0006_104122061330#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002370/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 164 + ], + [ + 251, + 175 + ], + [ + 264, + 189 + ], + [ + 269, + 138 + ], + [ + 278, + 178 + ], + [ + 283, + 130 + ] + ], + [ + [ + 272, + 181 + ], + [ + 256, + 197 + ], + [ + 273, + 160 + ], + [ + 271, + 200 + ], + [ + 288, + 145 + ], + [ + 284, + 184 + ] + ], + [ + [ + 266, + 161 + ], + [ + 287, + 144 + ], + [ + 280, + 185 + ], + [ + 252, + 138 + ], + [ + 259, + 201 + ], + [ + 232, + 155 + ] + ], + [ + [ + 291, + 188 + ], + [ + 317, + 191 + ], + [ + 296, + 205 + ], + [ + 285, + 169 + ], + [ + 272, + 201 + ], + [ + 259, + 166 + ] + ] + ] + }, + { + "task_id": "0002372", + "id": "RH20T_cfg1_task_0043_user_0015_scene_0004_cfg_0001_819112070044#167", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002372/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 397, + 205 + ], + [ + 395, + 198 + ], + [ + 396, + 227 + ], + [ + 398, + 183 + ], + [ + 398, + 232 + ], + [ + 399, + 190 + ] + ], + [ + [ + 363, + 198 + ], + [ + 375, + 204 + ], + [ + 343, + 201 + ], + [ + 384, + 194 + ], + [ + 330, + 195 + ], + [ + 372, + 187 + ] + ], + [ + [ + 374, + 195 + ], + [ + 385, + 182 + ], + [ + 379, + 180 + ], + [ + 369, + 209 + ], + [ + 367, + 194 + ], + [ + 357, + 223 + ] + ], + [ + [ + 381, + 202 + ], + [ + 384, + 185 + ], + [ + 385, + 188 + ], + [ + 376, + 217 + ], + [ + 382, + 204 + ], + [ + 373, + 234 + ] + ] + ] + }, + { + "task_id": "0002374", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_035622060973#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002374/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 206, + 170 + ], + [ + 204, + 191 + ], + [ + 187, + 167 + ], + [ + 223, + 172 + ], + [ + 189, + 145 + ], + [ + 225, + 152 + ] + ], + [ + [ + 223, + 162 + ], + [ + 222, + 142 + ], + [ + 233, + 167 + ], + [ + 213, + 158 + ], + [ + 234, + 187 + ], + [ + 214, + 177 + ] + ], + [ + [ + 226, + 174 + ], + [ + 245, + 166 + ], + [ + 217, + 156 + ], + [ + 236, + 193 + ], + [ + 197, + 164 + ], + [ + 216, + 202 + ] + ], + [ + [ + 216, + 170 + ], + [ + 213, + 157 + ], + [ + 207, + 155 + ], + [ + 226, + 185 + ], + [ + 210, + 167 + ], + [ + 228, + 196 + ] + ] + ] + }, + { + "task_id": "0002375", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_035622060973#387", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002375/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 209 + ], + [ + 301, + 198 + ], + [ + 285, + 194 + ], + [ + 313, + 224 + ], + [ + 284, + 204 + ], + [ + 311, + 233 + ] + ], + [ + [ + 311, + 212 + ], + [ + 326, + 199 + ], + [ + 295, + 202 + ], + [ + 329, + 223 + ], + [ + 279, + 215 + ], + [ + 313, + 238 + ] + ], + [ + [ + 294, + 206 + ], + [ + 305, + 198 + ], + [ + 302, + 189 + ], + [ + 287, + 222 + ], + [ + 291, + 197 + ], + [ + 275, + 231 + ] + ], + [ + [ + 301, + 216 + ], + [ + 309, + 222 + ], + [ + 282, + 210 + ], + [ + 319, + 222 + ], + [ + 274, + 205 + ], + [ + 310, + 217 + ] + ] + ] + }, + { + "task_id": "0002376", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_035622060973#596", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002376/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 219, + 234 + ], + [ + 239, + 222 + ], + [ + 221, + 247 + ], + [ + 216, + 221 + ], + [ + 201, + 260 + ], + [ + 194, + 234 + ] + ], + [ + [ + 223, + 245 + ], + [ + 237, + 263 + ], + [ + 222, + 249 + ], + [ + 223, + 242 + ], + [ + 207, + 230 + ], + [ + 210, + 225 + ] + ], + [ + [ + 229, + 226 + ], + [ + 225, + 215 + ], + [ + 220, + 208 + ], + [ + 238, + 244 + ], + [ + 224, + 218 + ], + [ + 241, + 254 + ] + ], + [ + [ + 221, + 202 + ], + [ + 244, + 195 + ], + [ + 225, + 219 + ], + [ + 218, + 185 + ], + [ + 201, + 226 + ], + [ + 195, + 192 + ] + ] + ] + }, + { + "task_id": "0002378", + "id": "RH20T_cfg1_task_0048_user_0001_scene_0003_cfg_0001_750612070853#165", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002378/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 598, + 0 + ], + [ + 634, + 0 + ], + [ + 525, + 0 + ], + [ + 640, + 14 + ], + [ + 501, + 53 + ], + [ + 638, + 82 + ] + ], + [ + [ + 640, + 0 + ], + [ + 640, + 14 + ], + [ + 640, + 0 + ], + [ + 609, + 0 + ], + [ + 640, + 0 + ], + [ + 617, + 0 + ] + ], + [ + [ + 640, + 49 + ], + [ + 640, + 90 + ], + [ + 640, + 22 + ], + [ + 616, + 69 + ], + [ + 613, + 0 + ], + [ + 566, + 35 + ] + ], + [ + [ + 640, + 0 + ], + [ + 640, + 0 + ], + [ + 640, + 0 + ], + [ + 620, + 0 + ], + [ + 640, + 0 + ], + [ + 602, + 0 + ] + ] + ] + }, + { + "task_id": "0002379", + "id": "RH20T_cfg4_task_0111_user_0007_scene_0003_cfg_0004_104122062823#134", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object with the corresponding shape into the corresponding hole'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002379/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 24 + ], + [ + 284, + 8 + ], + [ + 304, + 41 + ], + [ + 265, + 7 + ], + [ + 306, + 58 + ], + [ + 266, + 24 + ] + ], + [ + [ + 273, + 73 + ], + [ + 283, + 64 + ], + [ + 267, + 97 + ], + [ + 280, + 49 + ], + [ + 258, + 104 + ], + [ + 270, + 58 + ] + ], + [ + [ + 280, + 55 + ], + [ + 289, + 81 + ], + [ + 287, + 60 + ], + [ + 273, + 49 + ], + [ + 279, + 36 + ], + [ + 264, + 21 + ] + ], + [ + [ + 299, + 53 + ], + [ + 297, + 29 + ], + [ + 323, + 60 + ], + [ + 276, + 47 + ], + [ + 324, + 83 + ], + [ + 279, + 69 + ] + ] + ] + }, + { + "task_id": "0002380", + "id": "RH20T_cfg4_task_0111_user_0007_scene_0003_cfg_0004_104122062823#281", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object with the corresponding shape into the corresponding hole'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002380/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 229, + 138 + ], + [ + 236, + 99 + ], + [ + 200, + 139 + ], + [ + 261, + 137 + ], + [ + 192, + 177 + ], + [ + 254, + 179 + ] + ], + [ + [ + 206, + 166 + ], + [ + 187, + 186 + ], + [ + 173, + 148 + ], + [ + 240, + 184 + ], + [ + 192, + 131 + ], + [ + 255, + 165 + ] + ], + [ + [ + 215, + 152 + ], + [ + 209, + 119 + ], + [ + 242, + 165 + ], + [ + 190, + 139 + ], + [ + 247, + 194 + ], + [ + 196, + 168 + ] + ], + [ + [ + 238, + 164 + ], + [ + 206, + 165 + ], + [ + 225, + 189 + ], + [ + 253, + 137 + ], + [ + 253, + 188 + ], + [ + 283, + 139 + ] + ] + ] + }, + { + "task_id": "0002381", + "id": "RH20T_cfg4_task_0111_user_0007_scene_0003_cfg_0004_104122062823#327", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object with the corresponding shape into the corresponding hole'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002381/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 277, + 148 + ], + [ + 282, + 138 + ], + [ + 248, + 133 + ], + [ + 307, + 163 + ], + [ + 240, + 144 + ], + [ + 303, + 176 + ] + ], + [ + [ + 254, + 132 + ], + [ + 247, + 101 + ], + [ + 287, + 135 + ], + [ + 223, + 129 + ], + [ + 292, + 165 + ], + [ + 231, + 157 + ] + ], + [ + [ + 233, + 126 + ], + [ + 231, + 111 + ], + [ + 214, + 159 + ], + [ + 251, + 94 + ], + [ + 217, + 169 + ], + [ + 251, + 109 + ] + ], + [ + [ + 239, + 125 + ], + [ + 251, + 154 + ], + [ + 218, + 138 + ], + [ + 258, + 112 + ], + [ + 205, + 106 + ], + [ + 247, + 82 + ] + ] + ] + }, + { + "task_id": "0002382", + "id": "RH20T_cfg5_task_0052_user_0007_scene_0010_cfg_0005_105422061350#227", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab a basketball, release it and shoot it into the basket'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002382/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 185, + 149 + ], + [ + 210, + 158 + ], + [ + 202, + 154 + ], + [ + 165, + 143 + ], + [ + 179, + 145 + ], + [ + 140, + 134 + ] + ], + [ + [ + 197, + 144 + ], + [ + 196, + 126 + ], + [ + 219, + 129 + ], + [ + 174, + 161 + ], + [ + 219, + 145 + ], + [ + 176, + 177 + ] + ], + [ + [ + 206, + 165 + ], + [ + 232, + 163 + ], + [ + 197, + 179 + ], + [ + 215, + 150 + ], + [ + 171, + 183 + ], + [ + 186, + 152 + ] + ], + [ + [ + 197, + 104 + ], + [ + 198, + 74 + ], + [ + 182, + 98 + ], + [ + 214, + 110 + ], + [ + 181, + 125 + ], + [ + 213, + 140 + ] + ] + ] + }, + { + "task_id": "0002387", + "id": "RH20T_cfg5_task_0085_user_0007_scene_0006_cfg_0005_105422061350#295", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the bubble ring from the assembled bubble ring and ball'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002387/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 239, + 169 + ], + [ + 231, + 147 + ], + [ + 260, + 149 + ], + [ + 219, + 187 + ], + [ + 270, + 172 + ], + [ + 227, + 210 + ] + ], + [ + [ + 216, + 203 + ], + [ + 211, + 231 + ], + [ + 183, + 199 + ], + [ + 249, + 206 + ], + [ + 189, + 173 + ], + [ + 252, + 180 + ] + ], + [ + [ + 215, + 189 + ], + [ + 214, + 173 + ], + [ + 244, + 196 + ], + [ + 185, + 183 + ], + [ + 244, + 210 + ], + [ + 188, + 197 + ] + ], + [ + [ + 242, + 193 + ], + [ + 239, + 166 + ], + [ + 227, + 185 + ], + [ + 256, + 200 + ], + [ + 230, + 215 + ], + [ + 259, + 227 + ] + ] + ] + }, + { + "task_id": "0002389", + "id": "RH20T_cfg6_task_0125_user_0020_scene_0004_cfg_0006_037522061512#552", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull up the lever'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002389/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 136, + 143 + ], + [ + 159, + 144 + ], + [ + 136, + 118 + ], + [ + 136, + 167 + ], + [ + 114, + 117 + ], + [ + 114, + 166 + ] + ], + [ + [ + 131, + 170 + ], + [ + 141, + 186 + ], + [ + 107, + 183 + ], + [ + 154, + 158 + ], + [ + 94, + 165 + ], + [ + 143, + 139 + ] + ], + [ + [ + 143, + 136 + ], + [ + 163, + 152 + ], + [ + 125, + 150 + ], + [ + 159, + 122 + ], + [ + 103, + 133 + ], + [ + 140, + 106 + ] + ], + [ + [ + 108, + 141 + ], + [ + 104, + 122 + ], + [ + 125, + 126 + ], + [ + 90, + 158 + ], + [ + 128, + 144 + ], + [ + 95, + 175 + ] + ] + ] + }, + { + "task_id": "0002393", + "id": "RH20T_cfg4_task_0057_user_0014_scene_0008_cfg_0004_038522062288#130", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002393/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 288, + 145 + ], + [ + 294, + 115 + ], + [ + 260, + 142 + ], + [ + 318, + 149 + ], + [ + 254, + 172 + ], + [ + 313, + 180 + ] + ], + [ + [ + 265, + 174 + ], + [ + 285, + 151 + ], + [ + 266, + 169 + ], + [ + 264, + 178 + ], + [ + 245, + 194 + ], + [ + 246, + 200 + ] + ], + [ + [ + 275, + 161 + ], + [ + 277, + 145 + ], + [ + 245, + 161 + ], + [ + 305, + 160 + ], + [ + 246, + 176 + ], + [ + 302, + 175 + ] + ], + [ + [ + 286, + 188 + ], + [ + 284, + 214 + ], + [ + 257, + 184 + ], + [ + 314, + 192 + ], + [ + 260, + 159 + ], + [ + 315, + 167 + ] + ] + ] + }, + { + "task_id": "0002397", + "id": "RH20T_cfg4_task_0057_user_0014_scene_0008_cfg_0004_038522062288#250", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002397/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 261, + 145 + ], + [ + 237, + 158 + ], + [ + 275, + 168 + ], + [ + 247, + 121 + ], + [ + 298, + 155 + ], + [ + 271, + 108 + ] + ], + [ + [ + 272, + 164 + ], + [ + 267, + 143 + ], + [ + 244, + 170 + ], + [ + 299, + 159 + ], + [ + 249, + 193 + ], + [ + 305, + 181 + ] + ], + [ + [ + 264, + 135 + ], + [ + 292, + 128 + ], + [ + 268, + 158 + ], + [ + 258, + 109 + ], + [ + 241, + 165 + ], + [ + 228, + 116 + ] + ], + [ + [ + 269, + 153 + ], + [ + 271, + 137 + ], + [ + 242, + 160 + ], + [ + 296, + 146 + ], + [ + 242, + 174 + ], + [ + 294, + 161 + ] + ] + ] + }, + { + "task_id": "0002405", + "id": "RH20T_cfg4_task_0012_user_0010_scene_0005_cfg_0004_104422070011#232", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push the soccer ball into the goal'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002405/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 252, + 108 + ], + [ + 237, + 128 + ], + [ + 232, + 101 + ], + [ + 272, + 115 + ], + [ + 246, + 81 + ], + [ + 286, + 95 + ] + ], + [ + [ + 191, + 90 + ], + [ + 217, + 101 + ], + [ + 186, + 99 + ], + [ + 197, + 79 + ], + [ + 161, + 87 + ], + [ + 169, + 66 + ] + ], + [ + [ + 229, + 83 + ], + [ + 208, + 102 + ], + [ + 220, + 71 + ], + [ + 239, + 96 + ], + [ + 241, + 53 + ], + [ + 261, + 76 + ] + ], + [ + [ + 226, + 89 + ], + [ + 235, + 86 + ], + [ + 200, + 96 + ], + [ + 250, + 83 + ], + [ + 193, + 98 + ], + [ + 241, + 86 + ] + ] + ] + }, + { + "task_id": "0002406", + "id": "RH20T_cfg4_task_0012_user_0010_scene_0005_cfg_0004_104422070011#250", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push the soccer ball into the goal'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002406/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 194, + 82 + ], + [ + 203, + 55 + ], + [ + 195, + 86 + ], + [ + 192, + 78 + ], + [ + 187, + 112 + ], + [ + 183, + 107 + ] + ], + [ + [ + 208, + 90 + ], + [ + 216, + 115 + ], + [ + 209, + 100 + ], + [ + 207, + 80 + ], + [ + 199, + 71 + ], + [ + 199, + 54 + ] + ], + [ + [ + 222, + 67 + ], + [ + 216, + 37 + ], + [ + 200, + 73 + ], + [ + 241, + 62 + ], + [ + 206, + 102 + ], + [ + 246, + 90 + ] + ], + [ + [ + 197, + 66 + ], + [ + 208, + 67 + ], + [ + 170, + 71 + ], + [ + 223, + 62 + ], + [ + 161, + 70 + ], + [ + 213, + 61 + ] + ] + ] + }, + { + "task_id": "0002407", + "id": "RH20T_cfg7_task_0044_user_0020_scene_0001_cfg_0007_104122061602#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002407/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 475, + 219 + ], + [ + 482, + 202 + ], + [ + 494, + 240 + ], + [ + 457, + 200 + ], + [ + 485, + 255 + ], + [ + 451, + 216 + ] + ], + [ + [ + 472, + 242 + ], + [ + 493, + 223 + ], + [ + 495, + 260 + ], + [ + 450, + 223 + ], + [ + 475, + 278 + ], + [ + 431, + 242 + ] + ], + [ + [ + 461, + 193 + ], + [ + 435, + 199 + ], + [ + 470, + 179 + ], + [ + 452, + 206 + ], + [ + 496, + 173 + ], + [ + 475, + 200 + ] + ], + [ + [ + 489, + 180 + ], + [ + 466, + 190 + ], + [ + 481, + 193 + ], + [ + 498, + 166 + ], + [ + 502, + 183 + ], + [ + 521, + 156 + ] + ] + ] + }, + { + "task_id": "0002410", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122061330#1131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002410/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 342, + 151 + ], + [ + 322, + 162 + ], + [ + 341, + 169 + ], + [ + 344, + 131 + ], + [ + 359, + 158 + ], + [ + 363, + 120 + ] + ], + [ + [ + 338, + 192 + ], + [ + 359, + 190 + ], + [ + 348, + 171 + ], + [ + 326, + 214 + ], + [ + 329, + 173 + ], + [ + 307, + 215 + ] + ], + [ + [ + 322, + 172 + ], + [ + 347, + 179 + ], + [ + 316, + 155 + ], + [ + 327, + 189 + ], + [ + 291, + 148 + ], + [ + 303, + 182 + ] + ], + [ + [ + 354, + 172 + ], + [ + 351, + 156 + ], + [ + 343, + 194 + ], + [ + 365, + 151 + ], + [ + 346, + 207 + ], + [ + 367, + 165 + ] + ] + ] + }, + { + "task_id": "0002415", + "id": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_104122062823#246", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002415/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 123, + 44 + ], + [ + 119, + 18 + ], + [ + 150, + 39 + ], + [ + 94, + 49 + ], + [ + 152, + 62 + ], + [ + 99, + 73 + ] + ], + [ + [ + 157, + 62 + ], + [ + 169, + 51 + ], + [ + 148, + 83 + ], + [ + 166, + 43 + ], + [ + 137, + 94 + ], + [ + 155, + 54 + ] + ], + [ + [ + 120, + 61 + ], + [ + 109, + 76 + ], + [ + 134, + 84 + ], + [ + 107, + 39 + ], + [ + 146, + 68 + ], + [ + 119, + 22 + ] + ], + [ + [ + 155, + 63 + ], + [ + 167, + 69 + ], + [ + 180, + 68 + ], + [ + 130, + 58 + ], + [ + 167, + 62 + ], + [ + 114, + 52 + ] + ] + ] + }, + { + "task_id": "0002416", + "id": "RH20T_cfg1_task_0106_user_0015_scene_0001_cfg_0001_750612070851#156", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the object down from the shelf'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002416/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 574, + 70 + ], + [ + 605, + 95 + ], + [ + 556, + 15 + ], + [ + 590, + 118 + ], + [ + 531, + 0 + ], + [ + 565, + 95 + ] + ], + [ + [ + 595, + 32 + ], + [ + 575, + 7 + ], + [ + 557, + 6 + ], + [ + 630, + 56 + ], + [ + 578, + 34 + ], + [ + 640, + 87 + ] + ], + [ + [ + 569, + 90 + ], + [ + 593, + 43 + ], + [ + 516, + 93 + ], + [ + 621, + 88 + ], + [ + 500, + 132 + ], + [ + 596, + 127 + ] + ], + [ + [ + 640, + 76 + ], + [ + 633, + 87 + ], + [ + 640, + 56 + ], + [ + 578, + 94 + ], + [ + 640, + 38 + ], + [ + 584, + 83 + ] + ] + ] + }, + { + "task_id": "0002419", + "id": "RH20T_cfg1_task_0023_user_0014_scene_0002_cfg_0001_038522062547#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002419/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 395, + 139 + ], + [ + 374, + 135 + ], + [ + 388, + 116 + ], + [ + 401, + 160 + ], + [ + 411, + 120 + ], + [ + 423, + 165 + ] + ], + [ + [ + 392, + 145 + ], + [ + 409, + 126 + ], + [ + 398, + 156 + ], + [ + 386, + 135 + ], + [ + 381, + 176 + ], + [ + 371, + 154 + ] + ], + [ + [ + 385, + 139 + ], + [ + 393, + 165 + ], + [ + 360, + 146 + ], + [ + 411, + 131 + ], + [ + 353, + 121 + ], + [ + 404, + 104 + ] + ], + [ + [ + 373, + 136 + ], + [ + 376, + 124 + ], + [ + 379, + 159 + ], + [ + 367, + 114 + ], + [ + 376, + 169 + ], + [ + 364, + 126 + ] + ] + ] + }, + { + "task_id": "0002420", + "id": "RH20T_cfg7_task_0013_user_0020_scene_0010_cfg_0007_104122061602#185", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002420/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 498, + 215 + ], + [ + 518, + 205 + ], + [ + 512, + 211 + ], + [ + 485, + 219 + ], + [ + 491, + 222 + ], + [ + 466, + 229 + ] + ], + [ + [ + 528, + 216 + ], + [ + 537, + 198 + ], + [ + 549, + 235 + ], + [ + 509, + 200 + ], + [ + 540, + 252 + ], + [ + 502, + 217 + ] + ], + [ + [ + 546, + 184 + ], + [ + 551, + 206 + ], + [ + 563, + 194 + ], + [ + 529, + 173 + ], + [ + 558, + 173 + ], + [ + 525, + 152 + ] + ], + [ + [ + 502, + 208 + ], + [ + 489, + 225 + ], + [ + 479, + 203 + ], + [ + 524, + 213 + ], + [ + 493, + 185 + ], + [ + 539, + 196 + ] + ] + ] + }, + { + "task_id": "0002422", + "id": "RH20T_cfg6_task_0077_user_0014_scene_0004_cfg_0006_104122061602#204", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002422/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 478, + 173 + ], + [ + 482, + 192 + ], + [ + 468, + 165 + ], + [ + 488, + 181 + ], + [ + 463, + 146 + ], + [ + 483, + 161 + ] + ], + [ + [ + 454, + 185 + ], + [ + 457, + 192 + ], + [ + 431, + 185 + ], + [ + 479, + 186 + ], + [ + 427, + 177 + ], + [ + 477, + 178 + ] + ], + [ + [ + 480, + 159 + ], + [ + 481, + 147 + ], + [ + 488, + 178 + ], + [ + 472, + 141 + ], + [ + 486, + 189 + ], + [ + 472, + 153 + ] + ], + [ + [ + 484, + 176 + ], + [ + 487, + 155 + ], + [ + 503, + 187 + ], + [ + 466, + 166 + ], + [ + 499, + 207 + ], + [ + 463, + 185 + ] + ] + ] + }, + { + "task_id": "0002424", + "id": "RH20T_cfg5_task_0011_user_0007_scene_0004_cfg_0005_036422060215#238", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002424/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 210, + 136 + ], + [ + 213, + 117 + ], + [ + 230, + 148 + ], + [ + 189, + 124 + ], + [ + 226, + 165 + ], + [ + 187, + 142 + ] + ], + [ + [ + 205, + 136 + ], + [ + 217, + 118 + ], + [ + 228, + 144 + ], + [ + 180, + 127 + ], + [ + 217, + 160 + ], + [ + 169, + 145 + ] + ], + [ + [ + 219, + 140 + ], + [ + 208, + 126 + ], + [ + 199, + 156 + ], + [ + 239, + 124 + ], + [ + 211, + 172 + ], + [ + 252, + 139 + ] + ], + [ + [ + 239, + 129 + ], + [ + 252, + 127 + ], + [ + 233, + 151 + ], + [ + 245, + 106 + ], + [ + 218, + 155 + ], + [ + 230, + 108 + ] + ] + ] + }, + { + "task_id": "0002426", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_037522062165#430", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002426/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 215, + 276 + ], + [ + 213, + 262 + ], + [ + 192, + 263 + ], + [ + 238, + 290 + ], + [ + 195, + 277 + ], + [ + 238, + 303 + ] + ], + [ + [ + 204, + 275 + ], + [ + 230, + 262 + ], + [ + 214, + 298 + ], + [ + 194, + 251 + ], + [ + 187, + 311 + ], + [ + 166, + 264 + ] + ], + [ + [ + 221, + 246 + ], + [ + 214, + 219 + ], + [ + 201, + 250 + ], + [ + 243, + 242 + ], + [ + 209, + 276 + ], + [ + 251, + 269 + ] + ], + [ + [ + 223, + 315 + ], + [ + 236, + 288 + ], + [ + 232, + 315 + ], + [ + 214, + 314 + ], + [ + 218, + 345 + ], + [ + 201, + 340 + ] + ] + ] + }, + { + "task_id": "0002428", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_037522062165#867", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002428/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 313, + 305 + ], + [ + 333, + 296 + ], + [ + 332, + 319 + ], + [ + 296, + 293 + ], + [ + 310, + 331 + ], + [ + 275, + 302 + ] + ], + [ + [ + 315, + 296 + ], + [ + 342, + 304 + ], + [ + 308, + 321 + ], + [ + 322, + 270 + ], + [ + 282, + 313 + ], + [ + 295, + 262 + ] + ], + [ + [ + 292, + 307 + ], + [ + 295, + 295 + ], + [ + 277, + 285 + ], + [ + 308, + 331 + ], + [ + 275, + 296 + ], + [ + 304, + 341 + ] + ], + [ + [ + 278, + 344 + ], + [ + 290, + 360 + ], + [ + 292, + 320 + ], + [ + 262, + 360 + ], + [ + 280, + 303 + ], + [ + 248, + 351 + ] + ] + ] + }, + { + "task_id": "0002429", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0006_cfg_0004_104122062823#135", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002429/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 339, + 39 + ], + [ + 338, + 0 + ], + [ + 357, + 23 + ], + [ + 318, + 57 + ], + [ + 357, + 59 + ], + [ + 320, + 95 + ] + ], + [ + [ + 358, + 24 + ], + [ + 359, + 0 + ], + [ + 373, + 19 + ], + [ + 342, + 30 + ], + [ + 371, + 58 + ], + [ + 341, + 74 + ] + ], + [ + [ + 318, + 0 + ], + [ + 324, + 0 + ], + [ + 311, + 0 + ], + [ + 325, + 39 + ], + [ + 307, + 0 + ], + [ + 320, + 42 + ] + ], + [ + [ + 340, + 22 + ], + [ + 340, + 0 + ], + [ + 335, + 49 + ], + [ + 345, + 0 + ], + [ + 335, + 85 + ], + [ + 345, + 34 + ] + ] + ] + }, + { + "task_id": "0002431", + "id": "RH20T_cfg1_task_0083_user_0014_scene_0003_cfg_0001_819112070044#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Swat with a flyswatter'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002431/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 278, + 194 + ], + [ + 296, + 201 + ], + [ + 278, + 177 + ], + [ + 278, + 209 + ], + [ + 260, + 171 + ], + [ + 261, + 203 + ] + ], + [ + [ + 302, + 209 + ], + [ + 315, + 225 + ], + [ + 318, + 197 + ], + [ + 286, + 222 + ], + [ + 306, + 181 + ], + [ + 274, + 206 + ] + ], + [ + [ + 296, + 202 + ], + [ + 295, + 186 + ], + [ + 307, + 191 + ], + [ + 284, + 214 + ], + [ + 307, + 206 + ], + [ + 285, + 229 + ] + ], + [ + [ + 302, + 205 + ], + [ + 305, + 203 + ], + [ + 281, + 200 + ], + [ + 323, + 211 + ], + [ + 277, + 202 + ], + [ + 320, + 214 + ] + ] + ] + }, + { + "task_id": "0002433", + "id": "RH20T_cfg7_task_0036_user_0014_scene_0001_cfg_0007_104122061018#85", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Shake the test tube'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002433/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 282, + 29 + ], + [ + 282, + 10 + ], + [ + 251, + 27 + ], + [ + 312, + 31 + ], + [ + 252, + 44 + ], + [ + 310, + 48 + ] + ], + [ + [ + 299, + 79 + ], + [ + 323, + 88 + ], + [ + 284, + 93 + ], + [ + 315, + 64 + ], + [ + 259, + 84 + ], + [ + 288, + 53 + ] + ], + [ + [ + 281, + 59 + ], + [ + 265, + 56 + ], + [ + 261, + 81 + ], + [ + 300, + 38 + ], + [ + 278, + 86 + ], + [ + 319, + 41 + ] + ], + [ + [ + 291, + 64 + ], + [ + 296, + 49 + ], + [ + 280, + 38 + ], + [ + 302, + 90 + ], + [ + 275, + 53 + ], + [ + 296, + 102 + ] + ] + ] + }, + { + "task_id": "0002434", + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104122062823#157", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002434/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 384, + 37 + ], + [ + 411, + 7 + ], + [ + 378, + 47 + ], + [ + 390, + 27 + ], + [ + 353, + 76 + ], + [ + 367, + 53 + ] + ], + [ + [ + 429, + 0 + ], + [ + 473, + 0 + ], + [ + 430, + 0 + ], + [ + 428, + 44 + ], + [ + 389, + 0 + ], + [ + 388, + 47 + ] + ], + [ + [ + 415, + 29 + ], + [ + 420, + 0 + ], + [ + 456, + 35 + ], + [ + 377, + 24 + ], + [ + 449, + 73 + ], + [ + 375, + 60 + ] + ], + [ + [ + 383, + 69 + ], + [ + 402, + 54 + ], + [ + 368, + 31 + ], + [ + 397, + 104 + ], + [ + 345, + 45 + ], + [ + 376, + 122 + ] + ] + ] + }, + { + "task_id": "0002436", + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104122062823#267", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002436/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 394, + 0 + ], + [ + 380, + 0 + ], + [ + 363, + 1 + ], + [ + 429, + 0 + ], + [ + 377, + 0 + ], + [ + 441, + 0 + ] + ], + [ + [ + 454, + 22 + ], + [ + 460, + 39 + ], + [ + 428, + 3 + ], + [ + 482, + 42 + ], + [ + 420, + 0 + ], + [ + 477, + 24 + ] + ], + [ + [ + 377, + 27 + ], + [ + 396, + 0 + ], + [ + 344, + 10 + ], + [ + 410, + 43 + ], + [ + 329, + 36 + ], + [ + 392, + 65 + ] + ], + [ + [ + 418, + 16 + ], + [ + 424, + 0 + ], + [ + 457, + 22 + ], + [ + 381, + 11 + ], + [ + 449, + 58 + ], + [ + 377, + 47 + ] + ] + ] + }, + { + "task_id": "0002440", + "id": "RH20T_cfg1_task_0110_user_0015_scene_0003_cfg_0001_035622060973#97", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the pencil out from the pencil sharpener'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002440/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 263, + 199 + ], + [ + 261, + 188 + ], + [ + 254, + 182 + ], + [ + 272, + 217 + ], + [ + 256, + 193 + ], + [ + 274, + 226 + ] + ], + [ + [ + 288, + 192 + ], + [ + 273, + 194 + ], + [ + 272, + 200 + ], + [ + 303, + 184 + ], + [ + 288, + 198 + ], + [ + 319, + 182 + ] + ], + [ + [ + 267, + 220 + ], + [ + 266, + 217 + ], + [ + 247, + 227 + ], + [ + 288, + 213 + ], + [ + 247, + 230 + ], + [ + 290, + 216 + ] + ], + [ + [ + 272, + 221 + ], + [ + 267, + 200 + ], + [ + 293, + 215 + ], + [ + 251, + 227 + ], + [ + 298, + 236 + ], + [ + 257, + 248 + ] + ] + ] + }, + { + "task_id": "0002443", + "id": "RH20T_cfg4_task_0103_user_0007_scene_0003_cfg_0004_104122062823#105", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002443/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 312, + 151 + ], + [ + 320, + 181 + ], + [ + 316, + 135 + ], + [ + 309, + 164 + ], + [ + 308, + 105 + ], + [ + 302, + 137 + ] + ], + [ + [ + 306, + 134 + ], + [ + 304, + 105 + ], + [ + 338, + 143 + ], + [ + 276, + 125 + ], + [ + 338, + 169 + ], + [ + 279, + 151 + ] + ], + [ + [ + 301, + 119 + ], + [ + 302, + 87 + ], + [ + 336, + 120 + ], + [ + 266, + 118 + ], + [ + 333, + 149 + ], + [ + 266, + 147 + ] + ], + [ + [ + 326, + 144 + ], + [ + 326, + 132 + ], + [ + 304, + 116 + ], + [ + 346, + 170 + ], + [ + 302, + 128 + ], + [ + 347, + 185 + ] + ] + ] + }, + { + "task_id": "0002446", + "id": "RH20T_cfg5_task_0044_user_0010_scene_0002_cfg_0005_104122063678#275", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002446/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 280, + 225 + ], + [ + 288, + 208 + ], + [ + 283, + 250 + ], + [ + 277, + 202 + ], + [ + 276, + 264 + ], + [ + 270, + 218 + ] + ], + [ + [ + 282, + 221 + ], + [ + 269, + 248 + ], + [ + 256, + 208 + ], + [ + 308, + 234 + ], + [ + 269, + 182 + ], + [ + 320, + 208 + ] + ], + [ + [ + 288, + 213 + ], + [ + 286, + 221 + ], + [ + 311, + 226 + ], + [ + 264, + 200 + ], + [ + 311, + 219 + ], + [ + 266, + 193 + ] + ], + [ + [ + 309, + 218 + ], + [ + 281, + 216 + ], + [ + 314, + 191 + ], + [ + 304, + 245 + ], + [ + 341, + 194 + ], + [ + 331, + 246 + ] + ] + ] + }, + { + "task_id": "0002448", + "id": "RH20T_cfg5_task_0025_user_0010_scene_0006_cfg_0005_037522062165#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold a piece of paper'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002448/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 236, + 251 + ], + [ + 238, + 244 + ], + [ + 213, + 234 + ], + [ + 260, + 269 + ], + [ + 212, + 242 + ], + [ + 256, + 274 + ] + ], + [ + [ + 243, + 234 + ], + [ + 227, + 232 + ], + [ + 256, + 206 + ], + [ + 231, + 261 + ], + [ + 270, + 209 + ], + [ + 247, + 261 + ] + ], + [ + [ + 204, + 243 + ], + [ + 183, + 251 + ], + [ + 224, + 256 + ], + [ + 181, + 228 + ], + [ + 247, + 248 + ], + [ + 204, + 219 + ] + ], + [ + [ + 245, + 254 + ], + [ + 233, + 274 + ], + [ + 223, + 258 + ], + [ + 269, + 248 + ], + [ + 234, + 240 + ], + [ + 278, + 229 + ] + ] + ] + }, + { + "task_id": "0002449", + "id": "RH20T_cfg5_task_0043_user_0007_scene_0002_cfg_0005_037522062165#325", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002449/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 271, + 194 + ], + [ + 272, + 178 + ], + [ + 282, + 172 + ], + [ + 260, + 216 + ], + [ + 281, + 187 + ], + [ + 260, + 230 + ] + ], + [ + [ + 276, + 203 + ], + [ + 250, + 193 + ], + [ + 287, + 192 + ], + [ + 263, + 215 + ], + [ + 313, + 201 + ], + [ + 291, + 226 + ] + ], + [ + [ + 248, + 200 + ], + [ + 248, + 192 + ], + [ + 252, + 229 + ], + [ + 243, + 172 + ], + [ + 252, + 235 + ], + [ + 243, + 180 + ] + ], + [ + [ + 277, + 193 + ], + [ + 252, + 183 + ], + [ + 286, + 169 + ], + [ + 267, + 217 + ], + [ + 311, + 179 + ], + [ + 291, + 227 + ] + ] + ] + }, + { + "task_id": "0002453", + "id": "RH20T_cfg5_task_0042_user_0007_scene_0008_cfg_0005_104422070044#108", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002453/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 463, + 188 + ], + [ + 467, + 177 + ], + [ + 471, + 201 + ], + [ + 455, + 173 + ], + [ + 468, + 213 + ], + [ + 451, + 184 + ] + ], + [ + [ + 444, + 175 + ], + [ + 449, + 173 + ], + [ + 426, + 170 + ], + [ + 462, + 179 + ], + [ + 422, + 172 + ], + [ + 456, + 181 + ] + ], + [ + [ + 438, + 180 + ], + [ + 438, + 168 + ], + [ + 449, + 192 + ], + [ + 427, + 169 + ], + [ + 449, + 203 + ], + [ + 428, + 180 + ] + ], + [ + [ + 426, + 188 + ], + [ + 429, + 205 + ], + [ + 445, + 186 + ], + [ + 408, + 189 + ], + [ + 442, + 169 + ], + [ + 406, + 171 + ] + ] + ] + }, + { + "task_id": "0002457", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0009_cfg_0004_104122062295#105", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002457/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 169, + 237 + ], + [ + 139, + 241 + ], + [ + 180, + 291 + ], + [ + 158, + 184 + ], + [ + 215, + 291 + ], + [ + 192, + 176 + ] + ], + [ + [ + 142, + 228 + ], + [ + 105, + 268 + ], + [ + 167, + 209 + ], + [ + 121, + 245 + ], + [ + 198, + 175 + ], + [ + 152, + 212 + ] + ], + [ + [ + 190, + 204 + ], + [ + 197, + 162 + ], + [ + 151, + 181 + ], + [ + 231, + 227 + ], + [ + 147, + 218 + ], + [ + 222, + 264 + ] + ], + [ + [ + 154, + 201 + ], + [ + 137, + 233 + ], + [ + 99, + 217 + ], + [ + 199, + 188 + ], + [ + 117, + 178 + ], + [ + 220, + 154 + ] + ] + ] + }, + { + "task_id": "0002459", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_039422060546#72", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002459/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 245 + ], + [ + 374, + 247 + ], + [ + 338, + 210 + ], + [ + 339, + 278 + ], + [ + 303, + 209 + ], + [ + 305, + 275 + ] + ], + [ + [ + 317, + 271 + ], + [ + 335, + 260 + ], + [ + 323, + 305 + ], + [ + 310, + 236 + ], + [ + 303, + 320 + ], + [ + 288, + 246 + ] + ], + [ + [ + 323, + 253 + ], + [ + 292, + 254 + ], + [ + 335, + 286 + ], + [ + 310, + 218 + ], + [ + 369, + 286 + ], + [ + 345, + 215 + ] + ], + [ + [ + 333, + 248 + ], + [ + 330, + 228 + ], + [ + 328, + 280 + ], + [ + 338, + 218 + ], + [ + 331, + 296 + ], + [ + 341, + 236 + ] + ] + ] + }, + { + "task_id": "0002460", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_039422060546#177", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002460/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 282, + 221 + ], + [ + 306, + 246 + ], + [ + 255, + 247 + ], + [ + 309, + 196 + ], + [ + 234, + 222 + ], + [ + 286, + 173 + ] + ], + [ + [ + 272, + 216 + ], + [ + 264, + 196 + ], + [ + 268, + 245 + ], + [ + 275, + 189 + ], + [ + 275, + 262 + ], + [ + 282, + 207 + ] + ], + [ + [ + 283, + 192 + ], + [ + 248, + 187 + ], + [ + 286, + 163 + ], + [ + 281, + 218 + ], + [ + 323, + 168 + ], + [ + 315, + 223 + ] + ], + [ + [ + 262, + 242 + ], + [ + 293, + 234 + ], + [ + 271, + 269 + ], + [ + 252, + 214 + ], + [ + 241, + 276 + ], + [ + 220, + 222 + ] + ] + ] + }, + { + "task_id": "0002462", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_039422060546#606", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002462/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 404, + 166 + ], + [ + 424, + 152 + ], + [ + 417, + 152 + ], + [ + 390, + 181 + ], + [ + 399, + 165 + ], + [ + 372, + 193 + ] + ], + [ + [ + 396, + 174 + ], + [ + 428, + 175 + ], + [ + 397, + 150 + ], + [ + 395, + 199 + ], + [ + 367, + 150 + ], + [ + 363, + 198 + ] + ], + [ + [ + 409, + 154 + ], + [ + 409, + 130 + ], + [ + 403, + 177 + ], + [ + 414, + 133 + ], + [ + 404, + 198 + ], + [ + 414, + 156 + ] + ], + [ + [ + 390, + 123 + ], + [ + 366, + 143 + ], + [ + 372, + 99 + ], + [ + 407, + 146 + ], + [ + 397, + 78 + ], + [ + 431, + 126 + ] + ] + ] + }, + { + "task_id": "0002469", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_039422060546#700", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002469/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 335, + 223 + ], + [ + 307, + 244 + ], + [ + 325, + 210 + ], + [ + 344, + 235 + ], + [ + 355, + 188 + ], + [ + 371, + 216 + ] + ], + [ + [ + 360, + 234 + ], + [ + 330, + 249 + ], + [ + 344, + 205 + ], + [ + 376, + 262 + ], + [ + 374, + 189 + ], + [ + 406, + 247 + ] + ], + [ + [ + 355, + 197 + ], + [ + 356, + 175 + ], + [ + 349, + 225 + ], + [ + 360, + 172 + ], + [ + 348, + 244 + ], + [ + 359, + 192 + ] + ], + [ + [ + 354, + 206 + ], + [ + 325, + 189 + ], + [ + 371, + 184 + ], + [ + 336, + 230 + ], + [ + 401, + 201 + ], + [ + 368, + 250 + ] + ] + ] + }, + { + "task_id": "0002471", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122061330#390", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002471/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 369, + 205 + ], + [ + 395, + 204 + ], + [ + 354, + 186 + ], + [ + 383, + 223 + ], + [ + 331, + 189 + ], + [ + 360, + 224 + ] + ], + [ + [ + 346, + 219 + ], + [ + 334, + 244 + ], + [ + 349, + 230 + ], + [ + 342, + 207 + ], + [ + 359, + 207 + ], + [ + 353, + 182 + ] + ], + [ + [ + 342, + 201 + ], + [ + 364, + 208 + ], + [ + 352, + 177 + ], + [ + 331, + 226 + ], + [ + 332, + 171 + ], + [ + 310, + 219 + ] + ], + [ + [ + 310, + 216 + ], + [ + 321, + 199 + ], + [ + 334, + 213 + ], + [ + 284, + 221 + ], + [ + 323, + 228 + ], + [ + 274, + 237 + ] + ] + ] + }, + { + "task_id": "0002472", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122061330#668", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002472/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 162 + ], + [ + 288, + 174 + ], + [ + 280, + 138 + ], + [ + 289, + 185 + ], + [ + 277, + 128 + ], + [ + 286, + 173 + ] + ], + [ + [ + 300, + 189 + ], + [ + 293, + 215 + ], + [ + 308, + 193 + ], + [ + 290, + 184 + ], + [ + 315, + 168 + ], + [ + 297, + 156 + ] + ], + [ + [ + 310, + 185 + ], + [ + 318, + 159 + ], + [ + 286, + 178 + ], + [ + 335, + 192 + ], + [ + 277, + 204 + ], + [ + 326, + 219 + ] + ], + [ + [ + 317, + 196 + ], + [ + 328, + 196 + ], + [ + 291, + 194 + ], + [ + 342, + 199 + ], + [ + 283, + 194 + ], + [ + 332, + 199 + ] + ] + ] + }, + { + "task_id": "0002474", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122061330#1689", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002474/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 387, + 58 + ], + [ + 382, + 50 + ], + [ + 378, + 82 + ], + [ + 396, + 35 + ], + [ + 383, + 89 + ], + [ + 399, + 44 + ] + ], + [ + [ + 391, + 72 + ], + [ + 414, + 78 + ], + [ + 397, + 47 + ], + [ + 385, + 95 + ], + [ + 374, + 41 + ], + [ + 363, + 89 + ] + ], + [ + [ + 371, + 48 + ], + [ + 356, + 58 + ], + [ + 390, + 60 + ], + [ + 352, + 35 + ], + [ + 406, + 50 + ], + [ + 367, + 23 + ] + ], + [ + [ + 393, + 75 + ], + [ + 377, + 90 + ], + [ + 413, + 88 + ], + [ + 374, + 62 + ], + [ + 429, + 73 + ], + [ + 390, + 48 + ] + ] + ] + }, + { + "task_id": "0002475", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122061330#2082", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002475/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 289, + 102 + ], + [ + 269, + 100 + ], + [ + 292, + 127 + ], + [ + 287, + 78 + ], + [ + 310, + 128 + ], + [ + 305, + 80 + ] + ], + [ + [ + 321, + 105 + ], + [ + 313, + 94 + ], + [ + 312, + 127 + ], + [ + 330, + 83 + ], + [ + 320, + 135 + ], + [ + 337, + 93 + ] + ], + [ + [ + 320, + 120 + ], + [ + 299, + 125 + ], + [ + 331, + 109 + ], + [ + 310, + 131 + ], + [ + 352, + 104 + ], + [ + 329, + 126 + ] + ], + [ + [ + 339, + 100 + ], + [ + 355, + 121 + ], + [ + 339, + 99 + ], + [ + 338, + 101 + ], + [ + 324, + 79 + ], + [ + 321, + 79 + ] + ] + ] + }, + { + "task_id": "0002476", + "id": "RH20T_cfg4_task_0047_user_0014_scene_0005_cfg_0004_036422060909#114", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002476/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 262, + 85 + ], + [ + 259, + 69 + ], + [ + 247, + 67 + ], + [ + 278, + 103 + ], + [ + 250, + 82 + ], + [ + 280, + 116 + ] + ], + [ + [ + 276, + 99 + ], + [ + 282, + 76 + ], + [ + 263, + 106 + ], + [ + 291, + 92 + ], + [ + 257, + 128 + ], + [ + 285, + 117 + ] + ], + [ + [ + 289, + 83 + ], + [ + 288, + 100 + ], + [ + 304, + 68 + ], + [ + 272, + 99 + ], + [ + 306, + 49 + ], + [ + 272, + 81 + ] + ], + [ + [ + 256, + 62 + ], + [ + 279, + 47 + ], + [ + 251, + 50 + ], + [ + 260, + 74 + ], + [ + 227, + 66 + ], + [ + 238, + 88 + ] + ] + ] + }, + { + "task_id": "0002481", + "id": "RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_104122063633#233", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002481/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 176, + 172 + ], + [ + 173, + 193 + ], + [ + 154, + 168 + ], + [ + 198, + 176 + ], + [ + 158, + 147 + ], + [ + 202, + 155 + ] + ], + [ + [ + 189, + 168 + ], + [ + 188, + 151 + ], + [ + 194, + 155 + ], + [ + 184, + 183 + ], + [ + 195, + 171 + ], + [ + 185, + 199 + ] + ], + [ + [ + 173, + 175 + ], + [ + 192, + 173 + ], + [ + 184, + 190 + ], + [ + 163, + 161 + ], + [ + 164, + 193 + ], + [ + 143, + 162 + ] + ], + [ + [ + 173, + 154 + ], + [ + 165, + 141 + ], + [ + 181, + 171 + ], + [ + 165, + 138 + ], + [ + 187, + 182 + ], + [ + 172, + 151 + ] + ] + ] + }, + { + "task_id": "0002482", + "id": "RH20T_cfg1_task_0080_user_0015_scene_0005_cfg_0001_035622060973#135", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Flip over and spread out the paper that is laid flat on the table'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002482/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 249, + 234 + ], + [ + 259, + 215 + ], + [ + 246, + 225 + ], + [ + 253, + 244 + ], + [ + 237, + 243 + ], + [ + 243, + 264 + ] + ], + [ + [ + 285, + 237 + ], + [ + 275, + 221 + ], + [ + 303, + 231 + ], + [ + 265, + 243 + ], + [ + 313, + 247 + ], + [ + 276, + 261 + ] + ], + [ + [ + 267, + 229 + ], + [ + 265, + 223 + ], + [ + 255, + 212 + ], + [ + 279, + 247 + ], + [ + 257, + 218 + ], + [ + 279, + 252 + ] + ], + [ + [ + 286, + 218 + ], + [ + 308, + 219 + ], + [ + 282, + 214 + ], + [ + 290, + 222 + ], + [ + 260, + 212 + ], + [ + 267, + 221 + ] + ] + ] + }, + { + "task_id": "0002486", + "id": "RH20T_cfg2_task_0015_user_0006_scene_0003_cfg_0002_104122063678#366", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002486/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 252, + 357 + ], + [ + 248, + 345 + ], + [ + 233, + 360 + ], + [ + 269, + 334 + ], + [ + 237, + 360 + ], + [ + 272, + 346 + ] + ], + [ + [ + 254, + 346 + ], + [ + 255, + 345 + ], + [ + 269, + 322 + ], + [ + 238, + 360 + ], + [ + 269, + 322 + ], + [ + 236, + 360 + ] + ], + [ + [ + 250, + 348 + ], + [ + 270, + 360 + ], + [ + 240, + 341 + ], + [ + 259, + 354 + ], + [ + 220, + 326 + ], + [ + 240, + 340 + ] + ], + [ + [ + 245, + 333 + ], + [ + 238, + 305 + ], + [ + 236, + 338 + ], + [ + 255, + 328 + ], + [ + 243, + 360 + ], + [ + 262, + 359 + ] + ] + ] + }, + { + "task_id": "0002489", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104122062295#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002489/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 131 + ], + [ + 366, + 131 + ], + [ + 391, + 129 + ], + [ + 307, + 132 + ], + [ + 371, + 129 + ], + [ + 282, + 132 + ] + ], + [ + [ + 318, + 117 + ], + [ + 363, + 110 + ], + [ + 318, + 164 + ], + [ + 318, + 65 + ], + [ + 271, + 173 + ], + [ + 266, + 70 + ] + ], + [ + [ + 321, + 73 + ], + [ + 366, + 58 + ], + [ + 308, + 24 + ], + [ + 335, + 121 + ], + [ + 268, + 41 + ], + [ + 294, + 133 + ] + ], + [ + [ + 289, + 99 + ], + [ + 299, + 53 + ], + [ + 259, + 76 + ], + [ + 322, + 123 + ], + [ + 252, + 116 + ], + [ + 311, + 164 + ] + ] + ] + }, + { + "task_id": "0002490", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104122062295#156", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002490/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 265, + 115 + ], + [ + 276, + 72 + ], + [ + 233, + 88 + ], + [ + 299, + 145 + ], + [ + 226, + 126 + ], + [ + 287, + 181 + ] + ], + [ + [ + 281, + 103 + ], + [ + 280, + 81 + ], + [ + 327, + 92 + ], + [ + 234, + 114 + ], + [ + 323, + 111 + ], + [ + 239, + 131 + ] + ], + [ + [ + 215, + 116 + ], + [ + 201, + 94 + ], + [ + 259, + 84 + ], + [ + 171, + 148 + ], + [ + 280, + 107 + ], + [ + 184, + 177 + ] + ], + [ + [ + 299, + 120 + ], + [ + 262, + 106 + ], + [ + 325, + 101 + ], + [ + 270, + 141 + ], + [ + 363, + 115 + ], + [ + 312, + 158 + ] + ] + ] + }, + { + "task_id": "0002492", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104122062295#286", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002492/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 242, + 140 + ], + [ + 255, + 91 + ], + [ + 206, + 113 + ], + [ + 282, + 170 + ], + [ + 197, + 156 + ], + [ + 268, + 213 + ] + ], + [ + [ + 227, + 145 + ], + [ + 262, + 133 + ], + [ + 189, + 146 + ], + [ + 271, + 143 + ], + [ + 146, + 160 + ], + [ + 230, + 159 + ] + ], + [ + [ + 222, + 124 + ], + [ + 184, + 122 + ], + [ + 274, + 133 + ], + [ + 174, + 117 + ], + [ + 299, + 134 + ], + [ + 207, + 120 + ] + ], + [ + [ + 277, + 122 + ], + [ + 233, + 136 + ], + [ + 315, + 87 + ], + [ + 246, + 152 + ], + [ + 350, + 79 + ], + [ + 282, + 139 + ] + ] + ] + }, + { + "task_id": "0002496", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0002_cfg_0004_036422060909#179", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002496/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 343, + 157 + ], + [ + 344, + 140 + ], + [ + 329, + 140 + ], + [ + 358, + 175 + ], + [ + 328, + 157 + ], + [ + 357, + 191 + ] + ], + [ + [ + 352, + 132 + ], + [ + 348, + 108 + ], + [ + 356, + 119 + ], + [ + 348, + 146 + ], + [ + 360, + 141 + ], + [ + 353, + 169 + ] + ], + [ + [ + 334, + 176 + ], + [ + 316, + 190 + ], + [ + 313, + 164 + ], + [ + 353, + 187 + ], + [ + 333, + 148 + ], + [ + 373, + 173 + ] + ], + [ + [ + 352, + 176 + ], + [ + 357, + 183 + ], + [ + 377, + 168 + ], + [ + 327, + 183 + ], + [ + 371, + 163 + ], + [ + 324, + 177 + ] + ] + ] + }, + { + "task_id": "0002499", + "id": "RH20T_cfg6_task_0042_user_0014_scene_0004_cfg_0006_104122064161#226", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002499/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 337, + 157 + ], + [ + 355, + 145 + ], + [ + 339, + 180 + ], + [ + 334, + 133 + ], + [ + 320, + 194 + ], + [ + 314, + 147 + ] + ], + [ + [ + 341, + 179 + ], + [ + 349, + 189 + ], + [ + 335, + 157 + ], + [ + 347, + 202 + ], + [ + 325, + 145 + ], + [ + 337, + 192 + ] + ], + [ + [ + 346, + 172 + ], + [ + 350, + 146 + ], + [ + 346, + 186 + ], + [ + 346, + 159 + ], + [ + 341, + 210 + ], + [ + 342, + 182 + ] + ], + [ + [ + 319, + 160 + ], + [ + 322, + 146 + ], + [ + 305, + 138 + ], + [ + 334, + 183 + ], + [ + 303, + 151 + ], + [ + 330, + 194 + ] + ] + ] + }, + { + "task_id": "0002500", + "id": "RH20T_cfg6_task_0046_user_0020_scene_0004_cfg_0006_104122061602#380", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002500/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 478, + 153 + ], + [ + 452, + 147 + ], + [ + 493, + 139 + ], + [ + 462, + 169 + ], + [ + 521, + 145 + ], + [ + 491, + 177 + ] + ], + [ + [ + 440, + 141 + ], + [ + 426, + 137 + ], + [ + 429, + 119 + ], + [ + 452, + 163 + ], + [ + 442, + 123 + ], + [ + 465, + 166 + ] + ], + [ + [ + 465, + 140 + ], + [ + 476, + 165 + ], + [ + 446, + 150 + ], + [ + 485, + 130 + ], + [ + 436, + 127 + ], + [ + 473, + 104 + ] + ], + [ + [ + 486, + 120 + ], + [ + 482, + 93 + ], + [ + 509, + 118 + ], + [ + 465, + 122 + ], + [ + 513, + 146 + ], + [ + 468, + 148 + ] + ] + ] + }, + { + "task_id": "0002504", + "id": "RH20T_cfg7_task_0034_user_0014_scene_0001_cfg_0007_104122064161#384", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002504/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 330, + 285 + ], + [ + 302, + 290 + ], + [ + 332, + 309 + ], + [ + 328, + 264 + ], + [ + 361, + 305 + ], + [ + 354, + 260 + ] + ], + [ + [ + 332, + 270 + ], + [ + 361, + 272 + ], + [ + 335, + 242 + ], + [ + 330, + 299 + ], + [ + 306, + 240 + ], + [ + 301, + 297 + ] + ], + [ + [ + 379, + 271 + ], + [ + 405, + 264 + ], + [ + 372, + 263 + ], + [ + 385, + 279 + ], + [ + 345, + 271 + ], + [ + 361, + 285 + ] + ], + [ + [ + 357, + 269 + ], + [ + 362, + 251 + ], + [ + 359, + 293 + ], + [ + 354, + 248 + ], + [ + 354, + 310 + ], + [ + 350, + 265 + ] + ] + ] + }, + { + "task_id": "0002505", + "id": "RH20T_cfg7_task_0034_user_0014_scene_0001_cfg_0007_104122064161#483", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002505/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 396, + 257 + ], + [ + 389, + 233 + ], + [ + 424, + 255 + ], + [ + 369, + 259 + ], + [ + 431, + 279 + ], + [ + 377, + 282 + ] + ], + [ + [ + 370, + 262 + ], + [ + 366, + 281 + ], + [ + 395, + 252 + ], + [ + 346, + 272 + ], + [ + 398, + 236 + ], + [ + 351, + 255 + ] + ], + [ + [ + 387, + 266 + ], + [ + 389, + 284 + ], + [ + 382, + 286 + ], + [ + 393, + 244 + ], + [ + 380, + 269 + ], + [ + 390, + 228 + ] + ], + [ + [ + 403, + 259 + ], + [ + 380, + 276 + ], + [ + 397, + 246 + ], + [ + 409, + 271 + ], + [ + 420, + 228 + ], + [ + 430, + 255 + ] + ] + ] + }, + { + "task_id": "0002508", + "id": "RH20T_cfg4_task_0068_user_0015_scene_0006_cfg_0004_038522062288#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button horizontally with obstacles'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002508/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 121, + 111 + ], + [ + 105, + 101 + ], + [ + 126, + 136 + ], + [ + 116, + 86 + ], + [ + 140, + 144 + ], + [ + 130, + 96 + ] + ], + [ + [ + 95, + 129 + ], + [ + 81, + 134 + ], + [ + 122, + 125 + ], + [ + 66, + 132 + ], + [ + 137, + 120 + ], + [ + 80, + 127 + ] + ], + [ + [ + 95, + 130 + ], + [ + 90, + 114 + ], + [ + 65, + 124 + ], + [ + 125, + 135 + ], + [ + 72, + 139 + ], + [ + 128, + 149 + ] + ], + [ + [ + 88, + 99 + ], + [ + 97, + 102 + ], + [ + 66, + 117 + ], + [ + 110, + 81 + ], + [ + 54, + 116 + ], + [ + 102, + 77 + ] + ] + ] + }, + { + "task_id": "0002511", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0004_cfg_0004_038522062288#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002511/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 273, + 100 + ], + [ + 261, + 82 + ], + [ + 242, + 112 + ], + [ + 303, + 88 + ], + [ + 255, + 127 + ], + [ + 312, + 104 + ] + ], + [ + [ + 254, + 110 + ], + [ + 252, + 142 + ], + [ + 279, + 106 + ], + [ + 231, + 114 + ], + [ + 281, + 73 + ], + [ + 233, + 84 + ] + ], + [ + [ + 292, + 91 + ], + [ + 314, + 107 + ], + [ + 311, + 64 + ], + [ + 275, + 117 + ], + [ + 288, + 46 + ], + [ + 251, + 101 + ] + ], + [ + [ + 277, + 91 + ], + [ + 296, + 68 + ], + [ + 297, + 100 + ], + [ + 259, + 83 + ], + [ + 277, + 125 + ], + [ + 241, + 105 + ] + ] + ] + }, + { + "task_id": "0002514", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0004_cfg_0004_038522062288#282", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002514/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 183, + 144 + ], + [ + 170, + 174 + ], + [ + 193, + 160 + ], + [ + 170, + 126 + ], + [ + 204, + 133 + ], + [ + 184, + 97 + ] + ], + [ + [ + 193, + 101 + ], + [ + 201, + 111 + ], + [ + 212, + 132 + ], + [ + 173, + 67 + ], + [ + 204, + 121 + ], + [ + 167, + 60 + ] + ], + [ + [ + 195, + 104 + ], + [ + 206, + 95 + ], + [ + 170, + 129 + ], + [ + 218, + 81 + ], + [ + 162, + 135 + ], + [ + 207, + 89 + ] + ], + [ + [ + 211, + 126 + ], + [ + 205, + 99 + ], + [ + 176, + 138 + ], + [ + 245, + 114 + ], + [ + 183, + 161 + ], + [ + 248, + 138 + ] + ] + ] + }, + { + "task_id": "0002516", + "id": "RH20T_cfg7_task_0017_user_0014_scene_0002_cfg_0007_104122060811#171", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the pen into the pen holder'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002516/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 346, + 154 + ], + [ + 351, + 149 + ], + [ + 360, + 156 + ], + [ + 333, + 152 + ], + [ + 354, + 161 + ], + [ + 327, + 158 + ] + ], + [ + [ + 352, + 156 + ], + [ + 347, + 143 + ], + [ + 356, + 150 + ], + [ + 348, + 161 + ], + [ + 361, + 163 + ], + [ + 352, + 174 + ] + ], + [ + [ + 342, + 154 + ], + [ + 349, + 147 + ], + [ + 355, + 154 + ], + [ + 330, + 155 + ], + [ + 349, + 161 + ], + [ + 324, + 161 + ] + ], + [ + [ + 349, + 144 + ], + [ + 353, + 156 + ], + [ + 347, + 136 + ], + [ + 350, + 152 + ], + [ + 343, + 125 + ], + [ + 345, + 141 + ] + ] + ] + }, + { + "task_id": "0002517", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0006_cfg_0005_105422061350#237", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002517/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 224, + 84 + ], + [ + 250, + 93 + ], + [ + 234, + 66 + ], + [ + 213, + 104 + ], + [ + 208, + 57 + ], + [ + 185, + 95 + ] + ], + [ + [ + 209, + 86 + ], + [ + 209, + 93 + ], + [ + 207, + 110 + ], + [ + 210, + 60 + ], + [ + 207, + 102 + ], + [ + 210, + 54 + ] + ], + [ + [ + 240, + 99 + ], + [ + 244, + 84 + ], + [ + 245, + 120 + ], + [ + 235, + 77 + ], + [ + 240, + 137 + ], + [ + 229, + 93 + ] + ], + [ + [ + 227, + 81 + ], + [ + 225, + 62 + ], + [ + 240, + 62 + ], + [ + 214, + 101 + ], + [ + 241, + 79 + ], + [ + 217, + 117 + ] + ] + ] + }, + { + "task_id": "0002518", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0006_cfg_0005_105422061350#465", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002518/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 191, + 163 + ], + [ + 223, + 155 + ], + [ + 196, + 193 + ], + [ + 186, + 135 + ], + [ + 163, + 200 + ], + [ + 155, + 142 + ] + ], + [ + [ + 205, + 161 + ], + [ + 199, + 145 + ], + [ + 221, + 138 + ], + [ + 189, + 185 + ], + [ + 226, + 153 + ], + [ + 195, + 198 + ] + ], + [ + [ + 164, + 157 + ], + [ + 193, + 172 + ], + [ + 179, + 131 + ], + [ + 148, + 185 + ], + [ + 149, + 115 + ], + [ + 116, + 169 + ] + ], + [ + [ + 219, + 164 + ], + [ + 236, + 179 + ], + [ + 201, + 152 + ], + [ + 236, + 175 + ], + [ + 186, + 138 + ], + [ + 221, + 161 + ] + ] + ] + }, + { + "task_id": "0002522", + "id": "RH20T_cfg7_task_0036_user_0014_scene_0005_cfg_0007_104122061602#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Shake the test tube'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002522/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 466, + 152 + ], + [ + 441, + 166 + ], + [ + 477, + 161 + ], + [ + 454, + 141 + ], + [ + 503, + 147 + ], + [ + 481, + 125 + ] + ], + [ + [ + 460, + 126 + ], + [ + 446, + 129 + ], + [ + 437, + 127 + ], + [ + 484, + 126 + ], + [ + 452, + 123 + ], + [ + 501, + 123 + ] + ], + [ + [ + 464, + 183 + ], + [ + 449, + 173 + ], + [ + 444, + 175 + ], + [ + 486, + 191 + ], + [ + 459, + 184 + ], + [ + 501, + 200 + ] + ], + [ + [ + 461, + 153 + ], + [ + 478, + 171 + ], + [ + 452, + 171 + ], + [ + 470, + 133 + ], + [ + 437, + 154 + ], + [ + 452, + 116 + ] + ] + ] + }, + { + "task_id": "0002528", + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_105422061350#498", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002528/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 373, + 178 + ], + [ + 398, + 171 + ], + [ + 386, + 197 + ], + [ + 360, + 158 + ], + [ + 363, + 203 + ], + [ + 337, + 165 + ] + ], + [ + [ + 403, + 166 + ], + [ + 407, + 141 + ], + [ + 426, + 176 + ], + [ + 383, + 158 + ], + [ + 422, + 200 + ], + [ + 380, + 181 + ] + ], + [ + [ + 384, + 155 + ], + [ + 387, + 129 + ], + [ + 403, + 153 + ], + [ + 364, + 157 + ], + [ + 400, + 178 + ], + [ + 361, + 183 + ] + ], + [ + [ + 379, + 179 + ], + [ + 387, + 163 + ], + [ + 404, + 179 + ], + [ + 353, + 178 + ], + [ + 395, + 193 + ], + [ + 347, + 193 + ] + ] + ] + }, + { + "task_id": "0002531", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0003_cfg_0006_104122061602#196", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002531/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 389, + 89 + ], + [ + 413, + 84 + ], + [ + 392, + 67 + ], + [ + 386, + 109 + ], + [ + 366, + 72 + ], + [ + 362, + 115 + ] + ], + [ + [ + 399, + 88 + ], + [ + 400, + 62 + ], + [ + 413, + 93 + ], + [ + 387, + 84 + ], + [ + 412, + 119 + ], + [ + 386, + 108 + ] + ], + [ + [ + 415, + 100 + ], + [ + 414, + 81 + ], + [ + 389, + 105 + ], + [ + 442, + 94 + ], + [ + 390, + 124 + ], + [ + 445, + 115 + ] + ], + [ + [ + 386, + 89 + ], + [ + 411, + 89 + ], + [ + 388, + 65 + ], + [ + 383, + 113 + ], + [ + 363, + 66 + ], + [ + 359, + 112 + ] + ] + ] + }, + { + "task_id": "0002534", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_819112070044#282", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002534/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 214 + ], + [ + 391, + 217 + ], + [ + 414, + 194 + ], + [ + 397, + 232 + ], + [ + 428, + 192 + ], + [ + 411, + 229 + ] + ], + [ + [ + 396, + 195 + ], + [ + 412, + 202 + ], + [ + 382, + 203 + ], + [ + 410, + 186 + ], + [ + 365, + 196 + ], + [ + 393, + 179 + ] + ], + [ + [ + 384, + 184 + ], + [ + 402, + 191 + ], + [ + 389, + 194 + ], + [ + 379, + 175 + ], + [ + 369, + 186 + ], + [ + 360, + 168 + ] + ], + [ + [ + 389, + 203 + ], + [ + 393, + 188 + ], + [ + 395, + 190 + ], + [ + 383, + 218 + ], + [ + 391, + 204 + ], + [ + 379, + 233 + ] + ] + ] + }, + { + "task_id": "0002535", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_819112070044#446", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002535/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 389, + 208 + ], + [ + 376, + 196 + ], + [ + 393, + 189 + ], + [ + 385, + 228 + ], + [ + 405, + 200 + ], + [ + 398, + 239 + ] + ], + [ + [ + 376, + 226 + ], + [ + 376, + 250 + ], + [ + 365, + 224 + ], + [ + 389, + 229 + ], + [ + 365, + 202 + ], + [ + 390, + 204 + ] + ], + [ + [ + 381, + 241 + ], + [ + 376, + 265 + ], + [ + 358, + 236 + ], + [ + 405, + 246 + ], + [ + 363, + 212 + ], + [ + 410, + 222 + ] + ], + [ + [ + 375, + 234 + ], + [ + 381, + 219 + ], + [ + 388, + 223 + ], + [ + 361, + 246 + ], + [ + 382, + 238 + ], + [ + 356, + 261 + ] + ] + ] + }, + { + "task_id": "0002538", + "id": "RH20T_cfg1_task_0012_user_0014_scene_0004_cfg_0001_750612070853#104", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push the soccer ball into the goal'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002538/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 353, + 285 + ], + [ + 339, + 251 + ], + [ + 327, + 259 + ], + [ + 384, + 316 + ], + [ + 340, + 289 + ], + [ + 395, + 347 + ] + ], + [ + [ + 402, + 336 + ], + [ + 376, + 321 + ], + [ + 420, + 360 + ], + [ + 387, + 292 + ], + [ + 442, + 360 + ], + [ + 408, + 307 + ] + ], + [ + [ + 401, + 316 + ], + [ + 362, + 346 + ], + [ + 379, + 303 + ], + [ + 427, + 332 + ], + [ + 413, + 277 + ], + [ + 467, + 300 + ] + ], + [ + [ + 317, + 304 + ], + [ + 338, + 329 + ], + [ + 280, + 284 + ], + [ + 361, + 327 + ], + [ + 254, + 255 + ], + [ + 338, + 297 + ] + ] + ] + }, + { + "task_id": "0002539", + "id": "RH20T_cfg1_task_0012_user_0014_scene_0004_cfg_0001_750612070853#155", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push the soccer ball into the goal'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002539/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 418, + 232 + ], + [ + 406, + 197 + ], + [ + 393, + 212 + ], + [ + 447, + 256 + ], + [ + 404, + 242 + ], + [ + 457, + 289 + ] + ], + [ + [ + 363, + 224 + ], + [ + 343, + 265 + ], + [ + 361, + 228 + ], + [ + 365, + 218 + ], + [ + 378, + 191 + ], + [ + 387, + 174 + ] + ], + [ + [ + 418, + 196 + ], + [ + 459, + 179 + ], + [ + 408, + 166 + ], + [ + 429, + 229 + ], + [ + 371, + 182 + ], + [ + 387, + 245 + ] + ], + [ + [ + 437, + 232 + ], + [ + 477, + 208 + ], + [ + 435, + 198 + ], + [ + 439, + 261 + ], + [ + 390, + 223 + ], + [ + 401, + 284 + ] + ] + ] + }, + { + "task_id": "0002540", + "id": "RH20T_cfg1_task_0012_user_0014_scene_0004_cfg_0001_750612070853#207", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push the soccer ball into the goal'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002540/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 481, + 182 + ], + [ + 500, + 135 + ], + [ + 512, + 185 + ], + [ + 456, + 179 + ], + [ + 490, + 238 + ], + [ + 438, + 222 + ] + ], + [ + [ + 438, + 203 + ], + [ + 430, + 165 + ], + [ + 414, + 186 + ], + [ + 466, + 224 + ], + [ + 422, + 218 + ], + [ + 472, + 261 + ] + ], + [ + [ + 444, + 178 + ], + [ + 430, + 213 + ], + [ + 484, + 187 + ], + [ + 403, + 169 + ], + [ + 500, + 150 + ], + [ + 415, + 130 + ] + ], + [ + [ + 400, + 198 + ], + [ + 380, + 192 + ], + [ + 438, + 205 + ], + [ + 359, + 191 + ], + [ + 465, + 214 + ], + [ + 380, + 198 + ] + ] + ] + }, + { + "task_id": "0002543", + "id": "RH20T_cfg4_task_0127_user_0010_scene_0001_cfg_0004_036422060909#385", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002543/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 134 + ], + [ + 334, + 112 + ], + [ + 329, + 119 + ], + [ + 339, + 150 + ], + [ + 330, + 139 + ], + [ + 339, + 170 + ] + ], + [ + [ + 355, + 134 + ], + [ + 362, + 149 + ], + [ + 380, + 131 + ], + [ + 331, + 137 + ], + [ + 373, + 114 + ], + [ + 323, + 121 + ] + ], + [ + [ + 345, + 139 + ], + [ + 354, + 128 + ], + [ + 326, + 121 + ], + [ + 363, + 157 + ], + [ + 319, + 132 + ], + [ + 353, + 166 + ] + ], + [ + [ + 352, + 137 + ], + [ + 328, + 127 + ], + [ + 342, + 159 + ], + [ + 363, + 115 + ], + [ + 366, + 169 + ], + [ + 388, + 125 + ] + ] + ] + }, + { + "task_id": "0002548", + "id": "RH20T_cfg4_task_0069_user_0015_scene_0001_cfg_0004_104122062823#384", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Assemble one piece of a puzzle'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002548/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 402, + 81 + ], + [ + 406, + 54 + ], + [ + 416, + 69 + ], + [ + 387, + 95 + ], + [ + 411, + 93 + ], + [ + 383, + 119 + ] + ], + [ + [ + 428, + 86 + ], + [ + 456, + 93 + ], + [ + 437, + 85 + ], + [ + 420, + 87 + ], + [ + 407, + 77 + ], + [ + 393, + 79 + ] + ], + [ + [ + 418, + 88 + ], + [ + 421, + 110 + ], + [ + 444, + 76 + ], + [ + 393, + 99 + ], + [ + 440, + 55 + ], + [ + 391, + 79 + ] + ], + [ + [ + 392, + 117 + ], + [ + 407, + 142 + ], + [ + 409, + 112 + ], + [ + 373, + 122 + ], + [ + 395, + 88 + ], + [ + 358, + 96 + ] + ] + ] + }, + { + "task_id": "0002549", + "id": "RH20T_cfg4_task_0069_user_0015_scene_0001_cfg_0004_104122062823#474", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Assemble one piece of a puzzle'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002549/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 373, + 82 + ], + [ + 372, + 63 + ], + [ + 361, + 104 + ], + [ + 387, + 58 + ], + [ + 362, + 124 + ], + [ + 389, + 78 + ] + ], + [ + [ + 410, + 65 + ], + [ + 379, + 69 + ], + [ + 415, + 59 + ], + [ + 406, + 71 + ], + [ + 449, + 55 + ], + [ + 436, + 68 + ] + ], + [ + [ + 398, + 75 + ], + [ + 406, + 46 + ], + [ + 415, + 68 + ], + [ + 381, + 82 + ], + [ + 407, + 94 + ], + [ + 374, + 109 + ] + ], + [ + [ + 380, + 90 + ], + [ + 396, + 110 + ], + [ + 354, + 100 + ], + [ + 405, + 80 + ], + [ + 340, + 80 + ], + [ + 389, + 61 + ] + ] + ] + }, + { + "task_id": "0002550", + "id": "RH20T_cfg4_task_0069_user_0015_scene_0001_cfg_0004_104122062823#780", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Assemble one piece of a puzzle'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002550/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 41 + ], + [ + 437, + 47 + ], + [ + 415, + 13 + ], + [ + 398, + 70 + ], + [ + 385, + 8 + ], + [ + 369, + 64 + ] + ], + [ + [ + 403, + 70 + ], + [ + 410, + 42 + ], + [ + 423, + 62 + ], + [ + 382, + 79 + ], + [ + 416, + 86 + ], + [ + 376, + 105 + ] + ], + [ + [ + 417, + 66 + ], + [ + 396, + 53 + ], + [ + 396, + 88 + ], + [ + 438, + 43 + ], + [ + 417, + 102 + ], + [ + 461, + 56 + ] + ], + [ + [ + 399, + 80 + ], + [ + 404, + 49 + ], + [ + 367, + 74 + ], + [ + 431, + 87 + ], + [ + 364, + 103 + ], + [ + 425, + 116 + ] + ] + ] + }, + { + "task_id": "0002552", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0010_cfg_0001_038522062547#80", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002552/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 364, + 189 + ], + [ + 334, + 190 + ], + [ + 366, + 219 + ], + [ + 363, + 158 + ], + [ + 397, + 218 + ], + [ + 394, + 156 + ] + ], + [ + [ + 378, + 168 + ], + [ + 357, + 183 + ], + [ + 356, + 149 + ], + [ + 400, + 187 + ], + [ + 377, + 132 + ], + [ + 423, + 172 + ] + ], + [ + [ + 393, + 194 + ], + [ + 391, + 185 + ], + [ + 395, + 224 + ], + [ + 391, + 166 + ], + [ + 397, + 230 + ], + [ + 393, + 176 + ] + ], + [ + [ + 376, + 225 + ], + [ + 376, + 216 + ], + [ + 399, + 207 + ], + [ + 352, + 244 + ], + [ + 397, + 215 + ], + [ + 353, + 250 + ] + ] + ] + }, + { + "task_id": "0002554", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0010_cfg_0001_038522062547#597", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002554/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 351, + 101 + ], + [ + 377, + 108 + ], + [ + 342, + 127 + ], + [ + 359, + 76 + ], + [ + 317, + 120 + ], + [ + 333, + 70 + ] + ], + [ + [ + 350, + 92 + ], + [ + 324, + 83 + ], + [ + 352, + 76 + ], + [ + 348, + 110 + ], + [ + 377, + 84 + ], + [ + 375, + 119 + ] + ], + [ + [ + 378, + 119 + ], + [ + 376, + 104 + ], + [ + 384, + 142 + ], + [ + 372, + 97 + ], + [ + 386, + 155 + ], + [ + 375, + 111 + ] + ], + [ + [ + 348, + 100 + ], + [ + 375, + 100 + ], + [ + 345, + 94 + ], + [ + 350, + 105 + ], + [ + 317, + 93 + ], + [ + 325, + 104 + ] + ] + ] + }, + { + "task_id": "0002557", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0004_cfg_0006_037522061512#238", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002557/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 227, + 101 + ], + [ + 240, + 99 + ], + [ + 226, + 122 + ], + [ + 228, + 80 + ], + [ + 212, + 125 + ], + [ + 213, + 81 + ] + ], + [ + [ + 250, + 106 + ], + [ + 261, + 111 + ], + [ + 261, + 119 + ], + [ + 238, + 91 + ], + [ + 250, + 114 + ], + [ + 228, + 87 + ] + ], + [ + [ + 228, + 95 + ], + [ + 231, + 115 + ], + [ + 243, + 98 + ], + [ + 211, + 92 + ], + [ + 240, + 79 + ], + [ + 208, + 71 + ] + ], + [ + [ + 227, + 87 + ], + [ + 205, + 88 + ], + [ + 230, + 108 + ], + [ + 223, + 66 + ], + [ + 251, + 106 + ], + [ + 244, + 65 + ] + ] + ] + }, + { + "task_id": "0002558", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0004_cfg_0006_037522061512#572", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002558/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 236, + 71 + ], + [ + 244, + 52 + ], + [ + 230, + 69 + ], + [ + 243, + 72 + ], + [ + 222, + 87 + ], + [ + 235, + 92 + ] + ], + [ + [ + 260, + 89 + ], + [ + 258, + 73 + ], + [ + 256, + 78 + ], + [ + 263, + 99 + ], + [ + 258, + 95 + ], + [ + 264, + 115 + ] + ], + [ + [ + 247, + 86 + ], + [ + 259, + 75 + ], + [ + 252, + 103 + ], + [ + 243, + 69 + ], + [ + 241, + 114 + ], + [ + 230, + 80 + ] + ], + [ + [ + 251, + 116 + ], + [ + 242, + 98 + ], + [ + 255, + 109 + ], + [ + 247, + 123 + ], + [ + 264, + 127 + ], + [ + 256, + 140 + ] + ] + ] + }, + { + "task_id": "0002559", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0004_cfg_0006_037522061512#924", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002559/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 255, + 137 + ], + [ + 256, + 152 + ], + [ + 255, + 118 + ], + [ + 255, + 154 + ], + [ + 254, + 104 + ], + [ + 254, + 140 + ] + ], + [ + [ + 254, + 112 + ], + [ + 241, + 94 + ], + [ + 238, + 126 + ], + [ + 270, + 98 + ], + [ + 251, + 143 + ], + [ + 282, + 115 + ] + ], + [ + [ + 256, + 121 + ], + [ + 270, + 118 + ], + [ + 255, + 142 + ], + [ + 258, + 100 + ], + [ + 241, + 147 + ], + [ + 243, + 102 + ] + ], + [ + [ + 254, + 138 + ], + [ + 258, + 150 + ], + [ + 238, + 126 + ], + [ + 270, + 151 + ], + [ + 232, + 114 + ], + [ + 265, + 139 + ] + ] + ] + }, + { + "task_id": "0002566", + "id": "RH20T_cfg4_task_0019_user_0016_scene_0005_cfg_0004_104422070011#461", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the first move as black on the 3-4 point in the upper right corner of the Go board'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002566/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 362, + 326 + ], + [ + 365, + 354 + ], + [ + 385, + 333 + ], + [ + 337, + 318 + ], + [ + 382, + 307 + ], + [ + 336, + 292 + ] + ], + [ + [ + 325, + 313 + ], + [ + 309, + 300 + ], + [ + 302, + 309 + ], + [ + 346, + 316 + ], + [ + 320, + 323 + ], + [ + 364, + 329 + ] + ], + [ + [ + 359, + 305 + ], + [ + 370, + 302 + ], + [ + 336, + 325 + ], + [ + 380, + 286 + ], + [ + 328, + 326 + ], + [ + 370, + 289 + ] + ], + [ + [ + 358, + 315 + ], + [ + 346, + 303 + ], + [ + 345, + 344 + ], + [ + 371, + 288 + ], + [ + 357, + 355 + ], + [ + 381, + 301 + ] + ] + ] + }, + { + "task_id": "0002568", + "id": "RH20T_cfg3_task_0016_user_0011_scene_0010_cfg_0003_036422060909#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002568/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 367, + 85 + ], + [ + 373, + 67 + ], + [ + 364, + 68 + ], + [ + 370, + 103 + ], + [ + 359, + 85 + ], + [ + 365, + 119 + ] + ], + [ + [ + 345, + 95 + ], + [ + 325, + 94 + ], + [ + 330, + 103 + ], + [ + 360, + 86 + ], + [ + 348, + 104 + ], + [ + 379, + 87 + ] + ], + [ + [ + 385, + 105 + ], + [ + 409, + 98 + ], + [ + 390, + 110 + ], + [ + 380, + 99 + ], + [ + 368, + 116 + ], + [ + 355, + 106 + ] + ], + [ + [ + 371, + 89 + ], + [ + 370, + 107 + ], + [ + 353, + 79 + ], + [ + 390, + 100 + ], + [ + 353, + 60 + ], + [ + 392, + 81 + ] + ] + ] + }, + { + "task_id": "0002570", + "id": "RH20T_cfg6_task_0129_user_0020_scene_0009_cfg_0006_104122064161#200", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002570/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 399, + 119 + ], + [ + 396, + 100 + ], + [ + 419, + 113 + ], + [ + 380, + 126 + ], + [ + 423, + 133 + ], + [ + 383, + 145 + ] + ], + [ + [ + 397, + 135 + ], + [ + 391, + 149 + ], + [ + 381, + 141 + ], + [ + 414, + 129 + ], + [ + 387, + 128 + ], + [ + 419, + 116 + ] + ], + [ + [ + 385, + 128 + ], + [ + 390, + 131 + ], + [ + 378, + 149 + ], + [ + 392, + 108 + ], + [ + 374, + 145 + ], + [ + 388, + 106 + ] + ], + [ + [ + 392, + 149 + ], + [ + 376, + 144 + ], + [ + 396, + 128 + ], + [ + 388, + 170 + ], + [ + 411, + 133 + ], + [ + 403, + 175 + ] + ] + ] + }, + { + "task_id": "0002573", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_039422060546#241", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002573/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 254, + 124 + ], + [ + 245, + 106 + ], + [ + 249, + 149 + ], + [ + 258, + 102 + ], + [ + 257, + 165 + ], + [ + 265, + 119 + ] + ], + [ + [ + 239, + 148 + ], + [ + 265, + 149 + ], + [ + 231, + 151 + ], + [ + 247, + 145 + ], + [ + 204, + 151 + ], + [ + 218, + 144 + ] + ], + [ + [ + 261, + 112 + ], + [ + 266, + 83 + ], + [ + 260, + 114 + ], + [ + 262, + 110 + ], + [ + 254, + 144 + ], + [ + 257, + 136 + ] + ], + [ + [ + 238, + 124 + ], + [ + 220, + 100 + ], + [ + 261, + 108 + ], + [ + 214, + 141 + ], + [ + 278, + 132 + ], + [ + 231, + 165 + ] + ] + ] + }, + { + "task_id": "0002574", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_039422060546#373", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002574/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 329, + 78 + ], + [ + 340, + 64 + ], + [ + 322, + 56 + ], + [ + 336, + 102 + ], + [ + 313, + 70 + ], + [ + 326, + 114 + ] + ], + [ + [ + 336, + 48 + ], + [ + 327, + 56 + ], + [ + 359, + 39 + ], + [ + 313, + 58 + ], + [ + 370, + 30 + ], + [ + 322, + 50 + ] + ], + [ + [ + 326, + 81 + ], + [ + 301, + 82 + ], + [ + 327, + 60 + ], + [ + 325, + 103 + ], + [ + 353, + 58 + ], + [ + 351, + 101 + ] + ], + [ + [ + 349, + 74 + ], + [ + 349, + 53 + ], + [ + 347, + 94 + ], + [ + 351, + 55 + ], + [ + 347, + 113 + ], + [ + 351, + 75 + ] + ] + ] + }, + { + "task_id": "0002576", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_039422060546#721", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002576/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 344, + 168 + ], + [ + 373, + 160 + ], + [ + 337, + 163 + ], + [ + 353, + 174 + ], + [ + 310, + 170 + ], + [ + 322, + 183 + ] + ], + [ + [ + 346, + 182 + ], + [ + 345, + 163 + ], + [ + 358, + 208 + ], + [ + 334, + 158 + ], + [ + 357, + 224 + ], + [ + 335, + 176 + ] + ], + [ + [ + 337, + 203 + ], + [ + 367, + 197 + ], + [ + 345, + 189 + ], + [ + 329, + 219 + ], + [ + 319, + 194 + ], + [ + 300, + 224 + ] + ], + [ + [ + 330, + 211 + ], + [ + 315, + 212 + ], + [ + 357, + 204 + ], + [ + 301, + 219 + ], + [ + 376, + 202 + ], + [ + 316, + 219 + ] + ] + ] + }, + { + "task_id": "0002577", + "id": "RH20T_cfg7_task_0004_user_0020_scene_0005_cfg_0007_104122061602#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002577/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 510, + 237 + ], + [ + 501, + 235 + ], + [ + 541, + 234 + ], + [ + 481, + 240 + ], + [ + 548, + 236 + ], + [ + 490, + 242 + ] + ], + [ + [ + 517, + 245 + ], + [ + 535, + 238 + ], + [ + 532, + 272 + ], + [ + 502, + 220 + ], + [ + 516, + 277 + ], + [ + 488, + 227 + ] + ], + [ + [ + 498, + 223 + ], + [ + 508, + 247 + ], + [ + 475, + 226 + ], + [ + 524, + 219 + ], + [ + 466, + 203 + ], + [ + 514, + 194 + ] + ], + [ + [ + 532, + 267 + ], + [ + 515, + 283 + ], + [ + 516, + 246 + ], + [ + 549, + 290 + ], + [ + 533, + 229 + ], + [ + 567, + 273 + ] + ] + ] + }, + { + "task_id": "0002578", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0010_cfg_0003_104122062295#218", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002578/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 188, + 131 + ], + [ + 210, + 124 + ], + [ + 184, + 159 + ], + [ + 192, + 101 + ], + [ + 161, + 168 + ], + [ + 168, + 106 + ] + ], + [ + [ + 200, + 142 + ], + [ + 171, + 134 + ], + [ + 179, + 163 + ], + [ + 222, + 121 + ], + [ + 206, + 169 + ], + [ + 248, + 130 + ] + ], + [ + [ + 208, + 133 + ], + [ + 209, + 103 + ], + [ + 184, + 122 + ], + [ + 234, + 145 + ], + [ + 183, + 150 + ], + [ + 232, + 173 + ] + ], + [ + [ + 193, + 114 + ], + [ + 219, + 108 + ], + [ + 194, + 145 + ], + [ + 192, + 81 + ], + [ + 166, + 153 + ], + [ + 163, + 86 + ] + ] + ] + }, + { + "task_id": "0002590", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0008_cfg_0004_104122062295#192", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002590/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 239, + 98 + ], + [ + 243, + 47 + ], + [ + 271, + 128 + ], + [ + 210, + 70 + ], + [ + 265, + 171 + ], + [ + 209, + 113 + ] + ], + [ + [ + 256, + 109 + ], + [ + 264, + 61 + ], + [ + 215, + 89 + ], + [ + 301, + 131 + ], + [ + 211, + 131 + ], + [ + 291, + 173 + ] + ], + [ + [ + 306, + 131 + ], + [ + 296, + 118 + ], + [ + 303, + 178 + ], + [ + 308, + 80 + ], + [ + 314, + 198 + ], + [ + 320, + 91 + ] + ], + [ + [ + 241, + 109 + ], + [ + 192, + 98 + ], + [ + 250, + 89 + ], + [ + 233, + 125 + ], + [ + 301, + 100 + ], + [ + 276, + 134 + ] + ] + ] + }, + { + "task_id": "0002591", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0008_cfg_0004_104122062295#248", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002591/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 235, + 106 + ], + [ + 199, + 134 + ], + [ + 241, + 142 + ], + [ + 227, + 63 + ], + [ + 271, + 117 + ], + [ + 262, + 39 + ] + ], + [ + [ + 223, + 65 + ], + [ + 204, + 103 + ], + [ + 185, + 63 + ], + [ + 265, + 68 + ], + [ + 203, + 30 + ], + [ + 281, + 31 + ] + ], + [ + [ + 184, + 33 + ], + [ + 175, + 0 + ], + [ + 124, + 54 + ], + [ + 240, + 14 + ], + [ + 139, + 81 + ], + [ + 241, + 45 + ] + ], + [ + [ + 194, + 67 + ], + [ + 201, + 16 + ], + [ + 158, + 46 + ], + [ + 232, + 89 + ], + [ + 153, + 91 + ], + [ + 223, + 134 + ] + ] + ] + }, + { + "task_id": "0002592", + "id": "RH20T_cfg1_task_0106_user_0015_scene_0001_cfg_0001_819112070044#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the object down from the shelf'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002592/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 322, + 119 + ], + [ + 337, + 95 + ], + [ + 323, + 114 + ], + [ + 321, + 124 + ], + [ + 310, + 136 + ], + [ + 306, + 148 + ] + ], + [ + [ + 297, + 137 + ], + [ + 300, + 115 + ], + [ + 322, + 132 + ], + [ + 272, + 143 + ], + [ + 318, + 152 + ], + [ + 270, + 164 + ] + ], + [ + [ + 315, + 161 + ], + [ + 306, + 184 + ], + [ + 331, + 154 + ], + [ + 301, + 167 + ], + [ + 340, + 131 + ], + [ + 310, + 146 + ] + ], + [ + [ + 309, + 118 + ], + [ + 321, + 91 + ], + [ + 302, + 115 + ], + [ + 318, + 121 + ], + [ + 291, + 140 + ], + [ + 306, + 148 + ] + ] + ] + }, + { + "task_id": "0002593", + "id": "RH20T_cfg5_task_0118_user_0007_scene_0010_cfg_0005_036422060215#165", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the water tap'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002593/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 249, + 118 + ], + [ + 253, + 102 + ], + [ + 269, + 122 + ], + [ + 227, + 114 + ], + [ + 265, + 137 + ], + [ + 224, + 129 + ] + ], + [ + [ + 249, + 102 + ], + [ + 255, + 120 + ], + [ + 261, + 90 + ], + [ + 238, + 112 + ], + [ + 256, + 72 + ], + [ + 234, + 94 + ] + ], + [ + [ + 250, + 143 + ], + [ + 253, + 134 + ], + [ + 256, + 161 + ], + [ + 243, + 125 + ], + [ + 253, + 172 + ], + [ + 239, + 135 + ] + ], + [ + [ + 268, + 97 + ], + [ + 267, + 81 + ], + [ + 274, + 110 + ], + [ + 262, + 84 + ], + [ + 275, + 127 + ], + [ + 263, + 101 + ] + ] + ] + }, + { + "task_id": "0002594", + "id": "RH20T_cfg3_task_0073_user_0010_scene_0010_cfg_0003_036422060909#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002594/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 424, + 136 + ], + [ + 400, + 144 + ], + [ + 426, + 148 + ], + [ + 423, + 124 + ], + [ + 449, + 141 + ], + [ + 449, + 116 + ] + ], + [ + [ + 416, + 155 + ], + [ + 437, + 169 + ], + [ + 405, + 158 + ], + [ + 427, + 153 + ], + [ + 385, + 144 + ], + [ + 408, + 140 + ] + ], + [ + [ + 440, + 174 + ], + [ + 463, + 184 + ], + [ + 443, + 172 + ], + [ + 436, + 177 + ], + [ + 422, + 163 + ], + [ + 412, + 167 + ] + ], + [ + [ + 428, + 147 + ], + [ + 435, + 130 + ], + [ + 442, + 135 + ], + [ + 414, + 159 + ], + [ + 436, + 150 + ], + [ + 408, + 175 + ] + ] + ] + }, + { + "task_id": "0002597", + "id": "RH20T_cfg5_task_0043_user_0007_scene_0002_cfg_0005_104122063678#364", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002597/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 238 + ], + [ + 315, + 260 + ], + [ + 287, + 253 + ], + [ + 325, + 222 + ], + [ + 279, + 233 + ], + [ + 316, + 202 + ] + ], + [ + [ + 331, + 223 + ], + [ + 357, + 234 + ], + [ + 341, + 200 + ], + [ + 321, + 245 + ], + [ + 316, + 190 + ], + [ + 296, + 235 + ] + ], + [ + [ + 301, + 211 + ], + [ + 306, + 193 + ], + [ + 276, + 215 + ], + [ + 326, + 207 + ], + [ + 272, + 232 + ], + [ + 321, + 223 + ] + ], + [ + [ + 280, + 212 + ], + [ + 274, + 191 + ], + [ + 277, + 195 + ], + [ + 281, + 228 + ], + [ + 283, + 219 + ], + [ + 287, + 251 + ] + ] + ] + }, + { + "task_id": "0002600", + "id": "RH20T_cfg1_task_0059_user_0001_scene_0001_cfg_0001_038522063145#394", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002600/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 474, + 204 + ], + [ + 450, + 213 + ], + [ + 489, + 217 + ], + [ + 461, + 192 + ], + [ + 514, + 207 + ], + [ + 484, + 183 + ] + ], + [ + [ + 461, + 192 + ], + [ + 464, + 177 + ], + [ + 436, + 185 + ], + [ + 487, + 200 + ], + [ + 432, + 200 + ], + [ + 485, + 217 + ] + ], + [ + [ + 455, + 181 + ], + [ + 461, + 162 + ], + [ + 459, + 164 + ], + [ + 451, + 199 + ], + [ + 454, + 182 + ], + [ + 445, + 217 + ] + ], + [ + [ + 450, + 194 + ], + [ + 442, + 172 + ], + [ + 425, + 201 + ], + [ + 475, + 187 + ], + [ + 432, + 225 + ], + [ + 484, + 211 + ] + ] + ] + }, + { + "task_id": "0002601", + "id": "RH20T_cfg1_task_0059_user_0001_scene_0001_cfg_0001_038522063145#565", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002601/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 453, + 181 + ], + [ + 459, + 161 + ], + [ + 463, + 167 + ], + [ + 441, + 195 + ], + [ + 457, + 185 + ], + [ + 436, + 214 + ] + ], + [ + [ + 446, + 175 + ], + [ + 422, + 187 + ], + [ + 453, + 186 + ], + [ + 439, + 166 + ], + [ + 478, + 173 + ], + [ + 462, + 154 + ] + ], + [ + [ + 446, + 193 + ], + [ + 472, + 203 + ], + [ + 452, + 179 + ], + [ + 440, + 206 + ], + [ + 425, + 169 + ], + [ + 416, + 196 + ] + ], + [ + [ + 427, + 187 + ], + [ + 403, + 177 + ], + [ + 438, + 166 + ], + [ + 415, + 209 + ], + [ + 463, + 175 + ], + [ + 440, + 220 + ] + ] + ] + }, + { + "task_id": "0002603", + "id": "RH20T_cfg2_task_0011_user_0013_scene_0001_cfg_0002_104422070044#247", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002603/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 262, + 145 + ], + [ + 253, + 159 + ], + [ + 254, + 147 + ], + [ + 270, + 142 + ], + [ + 263, + 132 + ], + [ + 279, + 128 + ] + ], + [ + [ + 244, + 143 + ], + [ + 239, + 147 + ], + [ + 245, + 126 + ], + [ + 243, + 161 + ], + [ + 251, + 122 + ], + [ + 249, + 157 + ] + ], + [ + [ + 258, + 147 + ], + [ + 253, + 134 + ], + [ + 242, + 154 + ], + [ + 274, + 140 + ], + [ + 246, + 167 + ], + [ + 277, + 153 + ] + ], + [ + [ + 278, + 150 + ], + [ + 273, + 140 + ], + [ + 274, + 164 + ], + [ + 282, + 135 + ], + [ + 278, + 174 + ], + [ + 287, + 146 + ] + ] + ] + }, + { + "task_id": "0002605", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0007_cfg_0004_036422060909#264", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002605/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 366, + 160 + ], + [ + 386, + 167 + ], + [ + 365, + 186 + ], + [ + 368, + 133 + ], + [ + 347, + 178 + ], + [ + 349, + 128 + ] + ], + [ + [ + 411, + 156 + ], + [ + 393, + 151 + ], + [ + 420, + 180 + ], + [ + 402, + 132 + ], + [ + 436, + 183 + ], + [ + 418, + 137 + ] + ], + [ + [ + 383, + 177 + ], + [ + 392, + 161 + ], + [ + 382, + 156 + ], + [ + 384, + 200 + ], + [ + 374, + 171 + ], + [ + 376, + 214 + ] + ], + [ + [ + 383, + 155 + ], + [ + 384, + 182 + ], + [ + 356, + 155 + ], + [ + 412, + 155 + ], + [ + 355, + 127 + ], + [ + 411, + 126 + ] + ] + ] + }, + { + "task_id": "0002607", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0007_cfg_0004_036422060909#443", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002607/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 341, + 156 + ], + [ + 326, + 147 + ], + [ + 355, + 170 + ], + [ + 327, + 143 + ], + [ + 368, + 178 + ], + [ + 341, + 151 + ] + ], + [ + [ + 327, + 144 + ], + [ + 329, + 126 + ], + [ + 331, + 126 + ], + [ + 322, + 163 + ], + [ + 329, + 143 + ], + [ + 321, + 179 + ] + ], + [ + [ + 317, + 142 + ], + [ + 335, + 128 + ], + [ + 330, + 133 + ], + [ + 304, + 153 + ], + [ + 314, + 146 + ], + [ + 288, + 166 + ] + ], + [ + [ + 323, + 120 + ], + [ + 323, + 139 + ], + [ + 301, + 111 + ], + [ + 344, + 129 + ], + [ + 301, + 94 + ], + [ + 343, + 112 + ] + ] + ] + }, + { + "task_id": "0002608", + "id": "RH20T_cfg4_task_0091_user_0014_scene_0007_cfg_0004_036422060909#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002608/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 454, + 164 + ], + [ + 424, + 163 + ], + [ + 452, + 186 + ], + [ + 455, + 144 + ], + [ + 484, + 188 + ], + [ + 484, + 145 + ] + ], + [ + [ + 405, + 161 + ], + [ + 383, + 180 + ], + [ + 386, + 145 + ], + [ + 425, + 179 + ], + [ + 407, + 127 + ], + [ + 448, + 160 + ] + ], + [ + [ + 428, + 165 + ], + [ + 433, + 144 + ], + [ + 424, + 145 + ], + [ + 432, + 185 + ], + [ + 420, + 165 + ], + [ + 427, + 204 + ] + ], + [ + [ + 442, + 187 + ], + [ + 470, + 177 + ], + [ + 440, + 172 + ], + [ + 444, + 200 + ], + [ + 412, + 182 + ], + [ + 419, + 209 + ] + ] + ] + }, + { + "task_id": "0002613", + "id": "RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_104122063633#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002613/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 180, + 238 + ], + [ + 171, + 264 + ], + [ + 180, + 230 + ], + [ + 180, + 244 + ], + [ + 190, + 204 + ], + [ + 189, + 221 + ] + ], + [ + [ + 150, + 250 + ], + [ + 158, + 242 + ], + [ + 148, + 278 + ], + [ + 152, + 222 + ], + [ + 138, + 289 + ], + [ + 143, + 230 + ] + ], + [ + [ + 131, + 214 + ], + [ + 157, + 201 + ], + [ + 121, + 192 + ], + [ + 140, + 234 + ], + [ + 91, + 205 + ], + [ + 113, + 248 + ] + ], + [ + [ + 138, + 263 + ], + [ + 164, + 273 + ], + [ + 123, + 289 + ], + [ + 153, + 239 + ], + [ + 97, + 279 + ], + [ + 129, + 230 + ] + ] + ] + }, + { + "task_id": "0002616", + "id": "RH20T_cfg7_task_0033_user_0020_scene_0001_cfg_0007_104122061602#249", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002616/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 411, + 117 + ], + [ + 423, + 129 + ], + [ + 407, + 134 + ], + [ + 415, + 99 + ], + [ + 397, + 122 + ], + [ + 403, + 88 + ] + ], + [ + [ + 406, + 99 + ], + [ + 389, + 103 + ], + [ + 417, + 114 + ], + [ + 396, + 85 + ], + [ + 434, + 111 + ], + [ + 412, + 83 + ] + ], + [ + [ + 415, + 89 + ], + [ + 413, + 82 + ], + [ + 408, + 108 + ], + [ + 423, + 69 + ], + [ + 410, + 116 + ], + [ + 426, + 75 + ] + ], + [ + [ + 397, + 133 + ], + [ + 402, + 114 + ], + [ + 412, + 126 + ], + [ + 382, + 140 + ], + [ + 407, + 144 + ], + [ + 379, + 158 + ] + ] + ] + }, + { + "task_id": "0002617", + "id": "RH20T_cfg7_task_0033_user_0020_scene_0001_cfg_0007_104122061602#296", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002617/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 431, + 113 + ], + [ + 429, + 132 + ], + [ + 411, + 114 + ], + [ + 453, + 111 + ], + [ + 413, + 96 + ], + [ + 455, + 91 + ] + ], + [ + [ + 414, + 103 + ], + [ + 418, + 123 + ], + [ + 434, + 101 + ], + [ + 396, + 105 + ], + [ + 430, + 80 + ], + [ + 392, + 85 + ] + ], + [ + [ + 422, + 119 + ], + [ + 435, + 132 + ], + [ + 418, + 135 + ], + [ + 426, + 102 + ], + [ + 407, + 122 + ], + [ + 414, + 89 + ] + ], + [ + [ + 422, + 103 + ], + [ + 401, + 103 + ], + [ + 425, + 104 + ], + [ + 420, + 102 + ], + [ + 446, + 103 + ], + [ + 443, + 101 + ] + ] + ] + }, + { + "task_id": "0002619", + "id": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_035622060973#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Insert the tip of a medium pipette into the holder for medium pipette tips'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002619/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 233, + 169 + ], + [ + 252, + 162 + ], + [ + 224, + 159 + ], + [ + 241, + 179 + ], + [ + 205, + 167 + ], + [ + 224, + 187 + ] + ], + [ + [ + 251, + 160 + ], + [ + 261, + 160 + ], + [ + 251, + 179 + ], + [ + 251, + 140 + ], + [ + 242, + 178 + ], + [ + 242, + 140 + ] + ], + [ + [ + 231, + 155 + ], + [ + 227, + 144 + ], + [ + 227, + 138 + ], + [ + 235, + 172 + ], + [ + 231, + 149 + ], + [ + 239, + 181 + ] + ], + [ + [ + 243, + 149 + ], + [ + 239, + 165 + ], + [ + 224, + 149 + ], + [ + 261, + 148 + ], + [ + 229, + 131 + ], + [ + 266, + 131 + ] + ] + ] + }, + { + "task_id": "0002620", + "id": "RH20T_cfg5_task_0038_user_0007_scene_0009_cfg_0005_037522062165#254", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002620/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 343, + 138 + ], + [ + 347, + 118 + ], + [ + 322, + 125 + ], + [ + 366, + 152 + ], + [ + 319, + 144 + ], + [ + 361, + 170 + ] + ], + [ + [ + 342, + 134 + ], + [ + 330, + 111 + ], + [ + 351, + 116 + ], + [ + 332, + 152 + ], + [ + 361, + 137 + ], + [ + 343, + 173 + ] + ], + [ + [ + 359, + 141 + ], + [ + 376, + 128 + ], + [ + 379, + 149 + ], + [ + 340, + 134 + ], + [ + 360, + 163 + ], + [ + 321, + 146 + ] + ], + [ + [ + 323, + 139 + ], + [ + 340, + 122 + ], + [ + 342, + 159 + ], + [ + 304, + 119 + ], + [ + 325, + 176 + ], + [ + 285, + 135 + ] + ] + ] + }, + { + "task_id": "0002621", + "id": "RH20T_cfg6_task_0011_user_0014_scene_0007_cfg_0006_104122061602#145", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002621/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 450, + 206 + ], + [ + 461, + 235 + ], + [ + 471, + 199 + ], + [ + 431, + 213 + ], + [ + 459, + 169 + ], + [ + 421, + 186 + ] + ], + [ + [ + 470, + 133 + ], + [ + 490, + 156 + ], + [ + 471, + 129 + ], + [ + 468, + 138 + ], + [ + 452, + 108 + ], + [ + 446, + 115 + ] + ], + [ + [ + 461, + 169 + ], + [ + 473, + 150 + ], + [ + 479, + 191 + ], + [ + 444, + 147 + ], + [ + 467, + 208 + ], + [ + 434, + 165 + ] + ], + [ + [ + 448, + 180 + ], + [ + 470, + 183 + ], + [ + 443, + 207 + ], + [ + 452, + 152 + ], + [ + 420, + 204 + ], + [ + 430, + 148 + ] + ] + ] + }, + { + "task_id": "0002625", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_750612070853#103", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002625/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 236, + 120 + ], + [ + 274, + 116 + ], + [ + 233, + 150 + ], + [ + 238, + 92 + ], + [ + 194, + 155 + ], + [ + 202, + 96 + ] + ], + [ + [ + 195, + 176 + ], + [ + 194, + 141 + ], + [ + 153, + 180 + ], + [ + 236, + 173 + ], + [ + 152, + 218 + ], + [ + 237, + 210 + ] + ], + [ + [ + 166, + 131 + ], + [ + 169, + 170 + ], + [ + 209, + 131 + ], + [ + 122, + 131 + ], + [ + 207, + 90 + ], + [ + 117, + 88 + ] + ], + [ + [ + 198, + 141 + ], + [ + 188, + 107 + ], + [ + 180, + 126 + ], + [ + 217, + 157 + ], + [ + 189, + 156 + ], + [ + 225, + 188 + ] + ] + ] + }, + { + "task_id": "0002626", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_750612070853#236", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002626/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 232, + 146 + ], + [ + 247, + 154 + ], + [ + 265, + 130 + ], + [ + 200, + 161 + ], + [ + 251, + 118 + ], + [ + 181, + 152 + ] + ], + [ + [ + 229, + 139 + ], + [ + 190, + 148 + ], + [ + 222, + 98 + ], + [ + 237, + 179 + ], + [ + 259, + 91 + ], + [ + 271, + 169 + ] + ], + [ + [ + 229, + 171 + ], + [ + 267, + 166 + ], + [ + 230, + 132 + ], + [ + 227, + 208 + ], + [ + 186, + 136 + ], + [ + 187, + 215 + ] + ], + [ + [ + 214, + 152 + ], + [ + 206, + 115 + ], + [ + 193, + 135 + ], + [ + 238, + 170 + ], + [ + 201, + 167 + ], + [ + 245, + 203 + ] + ] + ] + }, + { + "task_id": "0002627", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_750612070853#372", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002627/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 226, + 140 + ], + [ + 270, + 117 + ], + [ + 198, + 106 + ], + [ + 257, + 177 + ], + [ + 151, + 130 + ], + [ + 210, + 206 + ] + ], + [ + [ + 238, + 159 + ], + [ + 228, + 120 + ], + [ + 214, + 140 + ], + [ + 265, + 181 + ], + [ + 224, + 174 + ], + [ + 273, + 216 + ] + ], + [ + [ + 226, + 177 + ], + [ + 208, + 157 + ], + [ + 202, + 213 + ], + [ + 250, + 139 + ], + [ + 220, + 238 + ], + [ + 272, + 159 + ] + ], + [ + [ + 201, + 128 + ], + [ + 224, + 88 + ], + [ + 219, + 162 + ], + [ + 185, + 98 + ], + [ + 196, + 199 + ], + [ + 165, + 134 + ] + ] + ] + }, + { + "task_id": "0002630", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_750612070853#744", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002630/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 247 + ], + [ + 266, + 212 + ], + [ + 242, + 224 + ], + [ + 311, + 273 + ], + [ + 251, + 254 + ], + [ + 316, + 303 + ] + ], + [ + [ + 226, + 236 + ], + [ + 177, + 213 + ], + [ + 259, + 203 + ], + [ + 197, + 265 + ], + [ + 303, + 226 + ], + [ + 240, + 282 + ] + ], + [ + [ + 221, + 235 + ], + [ + 243, + 203 + ], + [ + 234, + 206 + ], + [ + 211, + 260 + ], + [ + 207, + 244 + ], + [ + 186, + 297 + ] + ], + [ + [ + 327, + 270 + ], + [ + 305, + 283 + ], + [ + 343, + 318 + ], + [ + 312, + 226 + ], + [ + 371, + 306 + ], + [ + 335, + 207 + ] + ] + ] + }, + { + "task_id": "0002632", + "id": "RH20T_cfg5_task_0096_user_0007_scene_0008_cfg_0005_104122061850#460", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002632/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 322, + 214 + ], + [ + 327, + 220 + ], + [ + 295, + 232 + ], + [ + 349, + 197 + ], + [ + 287, + 227 + ], + [ + 344, + 189 + ] + ], + [ + [ + 297, + 188 + ], + [ + 289, + 171 + ], + [ + 271, + 178 + ], + [ + 325, + 198 + ], + [ + 280, + 193 + ], + [ + 331, + 213 + ] + ], + [ + [ + 311, + 151 + ], + [ + 308, + 131 + ], + [ + 284, + 166 + ], + [ + 340, + 136 + ], + [ + 285, + 189 + ], + [ + 345, + 158 + ] + ], + [ + [ + 278, + 173 + ], + [ + 290, + 147 + ], + [ + 300, + 188 + ], + [ + 255, + 158 + ], + [ + 289, + 214 + ], + [ + 241, + 186 + ] + ] + ] + }, + { + "task_id": "0002636", + "id": "RH20T_cfg1_task_0210_user_0015_scene_0008_cfg_0001_038522062547#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Chop the onions'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002636/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 377, + 0 + ], + [ + 388, + 6 + ], + [ + 384, + 0 + ], + [ + 371, + 22 + ], + [ + 373, + 0 + ], + [ + 361, + 13 + ] + ], + [ + [ + 362, + 12 + ], + [ + 347, + 7 + ], + [ + 369, + 0 + ], + [ + 356, + 37 + ], + [ + 383, + 0 + ], + [ + 370, + 41 + ] + ], + [ + [ + 356, + 23 + ], + [ + 358, + 5 + ], + [ + 358, + 42 + ], + [ + 354, + 5 + ], + [ + 356, + 58 + ], + [ + 352, + 22 + ] + ], + [ + [ + 374, + 17 + ], + [ + 362, + 18 + ], + [ + 374, + 42 + ], + [ + 374, + 0 + ], + [ + 387, + 42 + ], + [ + 387, + 0 + ] + ] + ] + }, + { + "task_id": "0002637", + "id": "RH20T_cfg3_task_0059_user_0010_scene_0002_cfg_0003_036422060909#165", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002637/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 356, + 141 + ], + [ + 353, + 153 + ], + [ + 345, + 121 + ], + [ + 367, + 161 + ], + [ + 348, + 107 + ], + [ + 370, + 148 + ] + ], + [ + [ + 367, + 123 + ], + [ + 372, + 109 + ], + [ + 389, + 116 + ], + [ + 345, + 131 + ], + [ + 384, + 129 + ], + [ + 342, + 144 + ] + ], + [ + [ + 340, + 112 + ], + [ + 360, + 110 + ], + [ + 343, + 136 + ], + [ + 338, + 88 + ], + [ + 324, + 137 + ], + [ + 319, + 90 + ] + ], + [ + [ + 347, + 117 + ], + [ + 344, + 97 + ], + [ + 351, + 101 + ], + [ + 343, + 133 + ], + [ + 353, + 119 + ], + [ + 346, + 152 + ] + ] + ] + }, + { + "task_id": "0002639", + "id": "RH20T_cfg5_task_0023_user_0010_scene_0007_cfg_0005_037522062165#320", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002639/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 339, + 210 + ], + [ + 345, + 186 + ], + [ + 331, + 216 + ], + [ + 348, + 204 + ], + [ + 325, + 240 + ], + [ + 342, + 231 + ] + ], + [ + [ + 324, + 214 + ], + [ + 326, + 199 + ], + [ + 303, + 202 + ], + [ + 346, + 226 + ], + [ + 301, + 217 + ], + [ + 343, + 240 + ] + ], + [ + [ + 370, + 200 + ], + [ + 395, + 194 + ], + [ + 372, + 221 + ], + [ + 368, + 180 + ], + [ + 346, + 226 + ], + [ + 344, + 185 + ] + ], + [ + [ + 339, + 210 + ], + [ + 340, + 202 + ], + [ + 315, + 200 + ], + [ + 364, + 221 + ], + [ + 315, + 208 + ], + [ + 361, + 228 + ] + ] + ] + }, + { + "task_id": "0002641", + "id": "RH20T_cfg1_task_0055_user_0015_scene_0009_cfg_0001_038522062547#78", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002641/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 360, + 145 + ], + [ + 383, + 132 + ], + [ + 377, + 161 + ], + [ + 344, + 130 + ], + [ + 353, + 176 + ], + [ + 322, + 144 + ] + ], + [ + [ + 352, + 125 + ], + [ + 378, + 121 + ], + [ + 365, + 145 + ], + [ + 339, + 106 + ], + [ + 337, + 150 + ], + [ + 312, + 109 + ] + ], + [ + [ + 360, + 117 + ], + [ + 362, + 103 + ], + [ + 376, + 139 + ], + [ + 344, + 96 + ], + [ + 373, + 151 + ], + [ + 343, + 110 + ] + ], + [ + [ + 359, + 131 + ], + [ + 385, + 139 + ], + [ + 363, + 109 + ], + [ + 355, + 152 + ], + [ + 337, + 101 + ], + [ + 331, + 143 + ] + ] + ] + }, + { + "task_id": "0002642", + "id": "RH20T_cfg1_task_0055_user_0015_scene_0009_cfg_0001_038522062547#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002642/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 468, + 80 + ], + [ + 447, + 66 + ], + [ + 484, + 69 + ], + [ + 450, + 91 + ], + [ + 506, + 83 + ], + [ + 472, + 105 + ] + ], + [ + [ + 455, + 90 + ], + [ + 456, + 105 + ], + [ + 471, + 71 + ], + [ + 439, + 109 + ], + [ + 470, + 54 + ], + [ + 436, + 94 + ] + ], + [ + [ + 464, + 54 + ], + [ + 471, + 39 + ], + [ + 481, + 75 + ], + [ + 448, + 35 + ], + [ + 473, + 88 + ], + [ + 442, + 49 + ] + ], + [ + [ + 461, + 46 + ], + [ + 451, + 67 + ], + [ + 480, + 63 + ], + [ + 441, + 28 + ], + [ + 489, + 44 + ], + [ + 451, + 8 + ] + ] + ] + }, + { + "task_id": "0002645", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0006_cfg_0005_104422070044#178", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002645/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 482, + 160 + ], + [ + 490, + 177 + ], + [ + 478, + 158 + ], + [ + 487, + 161 + ], + [ + 471, + 142 + ], + [ + 480, + 143 + ] + ], + [ + [ + 492, + 158 + ], + [ + 489, + 142 + ], + [ + 505, + 166 + ], + [ + 480, + 150 + ], + [ + 507, + 181 + ], + [ + 483, + 165 + ] + ], + [ + [ + 500, + 172 + ], + [ + 484, + 183 + ], + [ + 499, + 163 + ], + [ + 501, + 180 + ], + [ + 516, + 152 + ], + [ + 517, + 169 + ] + ], + [ + [ + 496, + 168 + ], + [ + 504, + 154 + ], + [ + 479, + 158 + ], + [ + 513, + 178 + ], + [ + 470, + 172 + ], + [ + 505, + 193 + ] + ] + ] + }, + { + "task_id": "0002646", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0006_cfg_0005_104422070044#285", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002646/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 625, + 168 + ], + [ + 626, + 153 + ], + [ + 620, + 184 + ], + [ + 630, + 153 + ], + [ + 619, + 199 + ], + [ + 629, + 168 + ] + ], + [ + [ + 608, + 169 + ], + [ + 635, + 169 + ], + [ + 609, + 147 + ], + [ + 608, + 191 + ], + [ + 584, + 148 + ], + [ + 583, + 190 + ] + ], + [ + [ + 628, + 187 + ], + [ + 616, + 200 + ], + [ + 608, + 188 + ], + [ + 640, + 186 + ], + [ + 619, + 175 + ], + [ + 640, + 172 + ] + ], + [ + [ + 602, + 178 + ], + [ + 615, + 179 + ], + [ + 581, + 169 + ], + [ + 622, + 186 + ], + [ + 571, + 168 + ], + [ + 610, + 185 + ] + ] + ] + }, + { + "task_id": "0002647", + "id": "RH20T_cfg4_task_0091_user_0014_scene_0007_cfg_0004_104122063550#53", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002647/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 203, + 286 + ], + [ + 178, + 295 + ], + [ + 182, + 286 + ], + [ + 222, + 285 + ], + [ + 210, + 276 + ], + [ + 247, + 276 + ] + ], + [ + [ + 195, + 292 + ], + [ + 193, + 285 + ], + [ + 222, + 308 + ], + [ + 168, + 275 + ], + [ + 223, + 313 + ], + [ + 172, + 282 + ] + ], + [ + [ + 216, + 279 + ], + [ + 237, + 253 + ], + [ + 241, + 299 + ], + [ + 191, + 258 + ], + [ + 220, + 326 + ], + [ + 168, + 285 + ] + ], + [ + [ + 195, + 285 + ], + [ + 222, + 269 + ], + [ + 205, + 300 + ], + [ + 184, + 269 + ], + [ + 179, + 317 + ], + [ + 155, + 286 + ] + ] + ] + }, + { + "task_id": "0002650", + "id": "RH20T_cfg1_task_0066_user_0010_scene_0004_cfg_0001_750612070851#727", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002650/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 604, + 237 + ], + [ + 579, + 226 + ], + [ + 631, + 222 + ], + [ + 577, + 254 + ], + [ + 640, + 233 + ], + [ + 605, + 268 + ] + ], + [ + [ + 630, + 242 + ], + [ + 640, + 267 + ], + [ + 617, + 238 + ], + [ + 640, + 246 + ], + [ + 589, + 215 + ], + [ + 615, + 224 + ] + ], + [ + [ + 637, + 244 + ], + [ + 640, + 224 + ], + [ + 613, + 260 + ], + [ + 640, + 229 + ], + [ + 597, + 278 + ], + [ + 640, + 247 + ] + ], + [ + [ + 621, + 281 + ], + [ + 600, + 252 + ], + [ + 591, + 279 + ], + [ + 640, + 284 + ], + [ + 611, + 306 + ], + [ + 640, + 316 + ] + ] + ] + }, + { + "task_id": "0002654", + "id": "RH20T_cfg6_task_0130_user_0014_scene_0004_cfg_0006_104122060811#236", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002654/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 352, + 125 + ], + [ + 357, + 119 + ], + [ + 358, + 114 + ], + [ + 346, + 135 + ], + [ + 353, + 120 + ], + [ + 340, + 142 + ] + ], + [ + [ + 354, + 124 + ], + [ + 346, + 111 + ], + [ + 350, + 126 + ], + [ + 358, + 121 + ], + [ + 358, + 137 + ], + [ + 366, + 134 + ] + ], + [ + [ + 357, + 132 + ], + [ + 370, + 125 + ], + [ + 364, + 135 + ], + [ + 350, + 128 + ], + [ + 352, + 142 + ], + [ + 338, + 135 + ] + ], + [ + [ + 358, + 119 + ], + [ + 356, + 123 + ], + [ + 343, + 120 + ], + [ + 373, + 118 + ], + [ + 344, + 116 + ], + [ + 375, + 114 + ] + ] + ] + }, + { + "task_id": "0002655", + "id": "RH20T_cfg6_task_0032_user_0020_scene_0009_cfg_0006_104122060811#208", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002655/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 410, + 120 + ], + [ + 424, + 119 + ], + [ + 410, + 106 + ], + [ + 410, + 134 + ], + [ + 396, + 107 + ], + [ + 397, + 134 + ] + ], + [ + [ + 411, + 135 + ], + [ + 397, + 132 + ], + [ + 409, + 144 + ], + [ + 413, + 127 + ], + [ + 423, + 148 + ], + [ + 427, + 131 + ] + ], + [ + [ + 416, + 120 + ], + [ + 419, + 108 + ], + [ + 416, + 128 + ], + [ + 416, + 111 + ], + [ + 414, + 139 + ], + [ + 413, + 123 + ] + ], + [ + [ + 418, + 124 + ], + [ + 423, + 117 + ], + [ + 431, + 124 + ], + [ + 405, + 124 + ], + [ + 426, + 131 + ], + [ + 400, + 131 + ] + ] + ] + }, + { + "task_id": "0002661", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_037522061512#395", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002661/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 212, + 119 + ], + [ + 221, + 140 + ], + [ + 223, + 114 + ], + [ + 201, + 124 + ], + [ + 214, + 92 + ], + [ + 193, + 104 + ] + ], + [ + [ + 237, + 102 + ], + [ + 223, + 95 + ], + [ + 247, + 114 + ], + [ + 227, + 88 + ], + [ + 262, + 120 + ], + [ + 241, + 95 + ] + ], + [ + [ + 237, + 112 + ], + [ + 255, + 112 + ], + [ + 250, + 111 + ], + [ + 222, + 114 + ], + [ + 233, + 111 + ], + [ + 204, + 115 + ] + ], + [ + [ + 233, + 117 + ], + [ + 245, + 105 + ], + [ + 237, + 133 + ], + [ + 229, + 99 + ], + [ + 225, + 146 + ], + [ + 215, + 112 + ] + ] + ] + }, + { + "task_id": "0002663", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_037522061512#1236", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002663/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 162, + 200 + ], + [ + 174, + 177 + ], + [ + 178, + 205 + ], + [ + 145, + 193 + ], + [ + 167, + 227 + ], + [ + 133, + 217 + ] + ], + [ + [ + 170, + 177 + ], + [ + 146, + 189 + ], + [ + 160, + 155 + ], + [ + 179, + 197 + ], + [ + 184, + 143 + ], + [ + 201, + 186 + ] + ], + [ + [ + 166, + 174 + ], + [ + 174, + 198 + ], + [ + 150, + 181 + ], + [ + 183, + 167 + ], + [ + 143, + 158 + ], + [ + 175, + 143 + ] + ], + [ + [ + 125, + 211 + ], + [ + 151, + 218 + ], + [ + 120, + 204 + ], + [ + 129, + 217 + ], + [ + 92, + 196 + ], + [ + 105, + 210 + ] + ] + ] + }, + { + "task_id": "0002666", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_104122063678#243", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002666/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 340, + 191 + ], + [ + 328, + 196 + ], + [ + 320, + 199 + ], + [ + 361, + 182 + ], + [ + 331, + 193 + ], + [ + 371, + 177 + ] + ], + [ + [ + 319, + 202 + ], + [ + 323, + 186 + ], + [ + 319, + 224 + ], + [ + 318, + 183 + ], + [ + 315, + 239 + ], + [ + 315, + 198 + ] + ], + [ + [ + 338, + 186 + ], + [ + 326, + 187 + ], + [ + 361, + 181 + ], + [ + 316, + 190 + ], + [ + 370, + 180 + ], + [ + 327, + 189 + ] + ], + [ + [ + 304, + 216 + ], + [ + 298, + 219 + ], + [ + 297, + 191 + ], + [ + 312, + 241 + ], + [ + 303, + 186 + ], + [ + 319, + 238 + ] + ] + ] + }, + { + "task_id": "0002667", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_104122063678#458", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002667/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 312, + 204 + ], + [ + 315, + 188 + ], + [ + 308, + 224 + ], + [ + 316, + 185 + ], + [ + 305, + 239 + ], + [ + 313, + 200 + ] + ], + [ + [ + 314, + 206 + ], + [ + 331, + 219 + ], + [ + 297, + 222 + ], + [ + 332, + 189 + ], + [ + 279, + 209 + ], + [ + 314, + 175 + ] + ], + [ + [ + 296, + 199 + ], + [ + 321, + 206 + ], + [ + 299, + 186 + ], + [ + 293, + 213 + ], + [ + 275, + 178 + ], + [ + 266, + 205 + ] + ], + [ + [ + 306, + 220 + ], + [ + 327, + 208 + ], + [ + 306, + 236 + ], + [ + 306, + 205 + ], + [ + 285, + 248 + ], + [ + 287, + 216 + ] + ] + ] + }, + { + "task_id": "0002669", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_036422060215#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002669/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 296, + 100 + ], + [ + 318, + 98 + ], + [ + 296, + 121 + ], + [ + 296, + 79 + ], + [ + 274, + 122 + ], + [ + 275, + 80 + ] + ], + [ + [ + 282, + 112 + ], + [ + 284, + 126 + ], + [ + 301, + 122 + ], + [ + 261, + 102 + ], + [ + 298, + 108 + ], + [ + 260, + 89 + ] + ], + [ + [ + 304, + 127 + ], + [ + 325, + 122 + ], + [ + 300, + 108 + ], + [ + 308, + 147 + ], + [ + 279, + 113 + ], + [ + 287, + 152 + ] + ], + [ + [ + 294, + 114 + ], + [ + 297, + 95 + ], + [ + 315, + 116 + ], + [ + 271, + 113 + ], + [ + 311, + 133 + ], + [ + 269, + 131 + ] + ] + ] + }, + { + "task_id": "0002670", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_036422060215#199", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002670/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 279, + 123 + ], + [ + 282, + 104 + ], + [ + 301, + 125 + ], + [ + 256, + 121 + ], + [ + 298, + 143 + ], + [ + 254, + 139 + ] + ], + [ + [ + 278, + 118 + ], + [ + 278, + 140 + ], + [ + 257, + 121 + ], + [ + 299, + 115 + ], + [ + 257, + 98 + ], + [ + 299, + 93 + ] + ], + [ + [ + 287, + 150 + ], + [ + 296, + 162 + ], + [ + 267, + 156 + ], + [ + 308, + 143 + ], + [ + 257, + 143 + ], + [ + 299, + 130 + ] + ], + [ + [ + 275, + 126 + ], + [ + 270, + 110 + ], + [ + 286, + 137 + ], + [ + 266, + 116 + ], + [ + 291, + 152 + ], + [ + 271, + 130 + ] + ] + ] + }, + { + "task_id": "0002672", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_036422060215#464", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002672/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 294, + 103 + ], + [ + 309, + 114 + ], + [ + 279, + 116 + ], + [ + 309, + 89 + ], + [ + 263, + 105 + ], + [ + 293, + 77 + ] + ], + [ + [ + 256, + 113 + ], + [ + 251, + 112 + ], + [ + 276, + 116 + ], + [ + 234, + 110 + ], + [ + 283, + 117 + ], + [ + 239, + 111 + ] + ], + [ + [ + 278, + 120 + ], + [ + 280, + 101 + ], + [ + 299, + 119 + ], + [ + 256, + 121 + ], + [ + 296, + 137 + ], + [ + 254, + 139 + ] + ], + [ + [ + 260, + 121 + ], + [ + 265, + 133 + ], + [ + 238, + 120 + ], + [ + 282, + 122 + ], + [ + 234, + 110 + ], + [ + 277, + 112 + ] + ] + ] + }, + { + "task_id": "0002673", + "id": "RH20T_cfg7_task_0029_user_0020_scene_0010_cfg_0007_104122061602#166", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002673/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 438, + 140 + ], + [ + 452, + 154 + ], + [ + 431, + 157 + ], + [ + 446, + 121 + ], + [ + 419, + 144 + ], + [ + 432, + 108 + ] + ], + [ + [ + 434, + 158 + ], + [ + 451, + 154 + ], + [ + 436, + 137 + ], + [ + 432, + 178 + ], + [ + 418, + 140 + ], + [ + 414, + 182 + ] + ], + [ + [ + 433, + 170 + ], + [ + 419, + 166 + ], + [ + 440, + 148 + ], + [ + 428, + 192 + ], + [ + 455, + 151 + ], + [ + 443, + 197 + ] + ], + [ + [ + 456, + 134 + ], + [ + 474, + 137 + ], + [ + 468, + 149 + ], + [ + 445, + 120 + ], + [ + 449, + 146 + ], + [ + 427, + 116 + ] + ] + ] + }, + { + "task_id": "0002674", + "id": "RH20T_cfg7_task_0016_user_0020_scene_0004_cfg_0007_104122063633#136", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002674/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 168, + 168 + ], + [ + 144, + 158 + ], + [ + 156, + 184 + ], + [ + 180, + 153 + ], + [ + 182, + 194 + ], + [ + 203, + 163 + ] + ], + [ + [ + 185, + 199 + ], + [ + 193, + 180 + ], + [ + 181, + 212 + ], + [ + 190, + 184 + ], + [ + 173, + 232 + ], + [ + 181, + 205 + ] + ], + [ + [ + 206, + 180 + ], + [ + 182, + 169 + ], + [ + 217, + 161 + ], + [ + 195, + 200 + ], + [ + 240, + 171 + ], + [ + 220, + 211 + ] + ], + [ + [ + 206, + 186 + ], + [ + 182, + 175 + ], + [ + 199, + 206 + ], + [ + 214, + 166 + ], + [ + 223, + 216 + ], + [ + 239, + 177 + ] + ] + ] + }, + { + "task_id": "0002675", + "id": "RH20T_cfg1_task_0111_user_0010_scene_0006_cfg_0001_750612070851#114", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object with the corresponding shape into the corresponding hole'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002675/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 433, + 360 + ], + [ + 437, + 344 + ], + [ + 391, + 356 + ], + [ + 478, + 360 + ], + [ + 390, + 360 + ], + [ + 472, + 360 + ] + ], + [ + [ + 438, + 345 + ], + [ + 403, + 360 + ], + [ + 444, + 353 + ], + [ + 431, + 336 + ], + [ + 477, + 336 + ], + [ + 469, + 317 + ] + ], + [ + [ + 398, + 347 + ], + [ + 425, + 316 + ], + [ + 422, + 360 + ], + [ + 372, + 322 + ], + [ + 395, + 360 + ], + [ + 343, + 354 + ] + ], + [ + [ + 397, + 360 + ], + [ + 381, + 327 + ], + [ + 407, + 360 + ], + [ + 385, + 344 + ], + [ + 424, + 360 + ], + [ + 403, + 360 + ] + ] + ] + }, + { + "task_id": "0002676", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_035622060973#152", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002676/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 267, + 115 + ], + [ + 277, + 115 + ], + [ + 266, + 135 + ], + [ + 267, + 95 + ], + [ + 257, + 134 + ], + [ + 257, + 96 + ] + ], + [ + [ + 274, + 134 + ], + [ + 273, + 121 + ], + [ + 255, + 128 + ], + [ + 293, + 139 + ], + [ + 257, + 139 + ], + [ + 293, + 150 + ] + ], + [ + [ + 255, + 149 + ], + [ + 265, + 163 + ], + [ + 268, + 136 + ], + [ + 243, + 163 + ], + [ + 259, + 123 + ], + [ + 234, + 150 + ] + ], + [ + [ + 292, + 138 + ], + [ + 299, + 121 + ], + [ + 293, + 144 + ], + [ + 291, + 131 + ], + [ + 286, + 161 + ], + [ + 284, + 149 + ] + ] + ] + }, + { + "task_id": "0002677", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_035622060973#264", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002677/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 294, + 195 + ], + [ + 309, + 180 + ], + [ + 308, + 210 + ], + [ + 279, + 179 + ], + [ + 292, + 225 + ], + [ + 264, + 193 + ] + ], + [ + [ + 324, + 189 + ], + [ + 316, + 198 + ], + [ + 312, + 172 + ], + [ + 336, + 206 + ], + [ + 321, + 164 + ], + [ + 344, + 196 + ] + ], + [ + [ + 290, + 187 + ], + [ + 272, + 174 + ], + [ + 277, + 205 + ], + [ + 303, + 169 + ], + [ + 295, + 218 + ], + [ + 321, + 182 + ] + ], + [ + [ + 300, + 182 + ], + [ + 301, + 169 + ], + [ + 286, + 170 + ], + [ + 315, + 194 + ], + [ + 286, + 182 + ], + [ + 314, + 206 + ] + ] + ] + }, + { + "task_id": "0002680", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_035622060973#809", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002680/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 221, + 192 + ], + [ + 217, + 181 + ], + [ + 202, + 185 + ], + [ + 240, + 199 + ], + [ + 206, + 195 + ], + [ + 242, + 209 + ] + ], + [ + [ + 208, + 190 + ], + [ + 216, + 174 + ], + [ + 223, + 187 + ], + [ + 194, + 192 + ], + [ + 215, + 203 + ], + [ + 185, + 208 + ] + ], + [ + [ + 220, + 201 + ], + [ + 230, + 182 + ], + [ + 229, + 211 + ], + [ + 211, + 190 + ], + [ + 219, + 229 + ], + [ + 200, + 210 + ] + ], + [ + [ + 214, + 182 + ], + [ + 206, + 199 + ], + [ + 231, + 183 + ], + [ + 196, + 180 + ], + [ + 238, + 167 + ], + [ + 204, + 165 + ] + ] + ] + }, + { + "task_id": "0002681", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0005_cfg_0004_104122063550#122", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002681/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 235 + ], + [ + 328, + 245 + ], + [ + 311, + 209 + ], + [ + 286, + 263 + ], + [ + 284, + 200 + ], + [ + 257, + 252 + ] + ], + [ + [ + 268, + 223 + ], + [ + 281, + 247 + ], + [ + 285, + 208 + ], + [ + 252, + 238 + ], + [ + 273, + 185 + ], + [ + 241, + 216 + ] + ], + [ + [ + 286, + 260 + ], + [ + 273, + 288 + ], + [ + 262, + 246 + ], + [ + 310, + 274 + ], + [ + 275, + 220 + ], + [ + 322, + 248 + ] + ], + [ + [ + 302, + 226 + ], + [ + 303, + 214 + ], + [ + 327, + 213 + ], + [ + 276, + 238 + ], + [ + 325, + 224 + ], + [ + 276, + 248 + ] + ] + ] + }, + { + "task_id": "0002682", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0005_cfg_0004_104122063550#227", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002682/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 304, + 176 + ], + [ + 331, + 169 + ], + [ + 297, + 148 + ], + [ + 310, + 203 + ], + [ + 269, + 155 + ], + [ + 283, + 209 + ] + ], + [ + [ + 320, + 179 + ], + [ + 318, + 164 + ], + [ + 327, + 201 + ], + [ + 314, + 157 + ], + [ + 329, + 213 + ], + [ + 316, + 171 + ] + ], + [ + [ + 316, + 160 + ], + [ + 341, + 153 + ], + [ + 323, + 184 + ], + [ + 309, + 136 + ], + [ + 298, + 191 + ], + [ + 284, + 142 + ] + ], + [ + [ + 326, + 175 + ], + [ + 301, + 180 + ], + [ + 331, + 174 + ], + [ + 322, + 177 + ], + [ + 357, + 169 + ], + [ + 346, + 173 + ] + ] + ] + }, + { + "task_id": "0002683", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0005_cfg_0004_104122063550#1400", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002683/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 286, + 269 + ], + [ + 304, + 291 + ], + [ + 283, + 248 + ], + [ + 290, + 289 + ], + [ + 266, + 228 + ], + [ + 274, + 270 + ] + ], + [ + [ + 293, + 275 + ], + [ + 281, + 304 + ], + [ + 281, + 281 + ], + [ + 306, + 268 + ], + [ + 292, + 255 + ], + [ + 317, + 240 + ] + ], + [ + [ + 265, + 275 + ], + [ + 260, + 254 + ], + [ + 295, + 274 + ], + [ + 236, + 276 + ], + [ + 299, + 294 + ], + [ + 242, + 296 + ] + ], + [ + [ + 255, + 248 + ], + [ + 256, + 257 + ], + [ + 283, + 254 + ], + [ + 226, + 241 + ], + [ + 284, + 244 + ], + [ + 223, + 231 + ] + ] + ] + }, + { + "task_id": "0002688", + "id": "RH20T_cfg4_task_0089_user_0014_scene_0001_cfg_0004_104122062823#305", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push an object with obstacles'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002688/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 515, + 75 + ], + [ + 493, + 50 + ], + [ + 526, + 49 + ], + [ + 505, + 98 + ], + [ + 552, + 77 + ], + [ + 528, + 124 + ] + ], + [ + [ + 515, + 93 + ], + [ + 552, + 86 + ], + [ + 509, + 113 + ], + [ + 522, + 71 + ], + [ + 474, + 119 + ], + [ + 482, + 79 + ] + ], + [ + [ + 471, + 90 + ], + [ + 464, + 63 + ], + [ + 499, + 70 + ], + [ + 443, + 110 + ], + [ + 504, + 95 + ], + [ + 450, + 134 + ] + ], + [ + [ + 448, + 85 + ], + [ + 484, + 72 + ], + [ + 444, + 69 + ], + [ + 452, + 102 + ], + [ + 412, + 80 + ], + [ + 416, + 114 + ] + ] + ] + }, + { + "task_id": "0002690", + "id": "RH20T_cfg4_task_0089_user_0014_scene_0001_cfg_0004_104122062823#387", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push an object with obstacles'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002690/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 31 + ], + [ + 315, + 4 + ], + [ + 344, + 32 + ], + [ + 287, + 29 + ], + [ + 342, + 56 + ], + [ + 288, + 53 + ] + ], + [ + [ + 310, + 3 + ], + [ + 328, + 0 + ], + [ + 303, + 0 + ], + [ + 316, + 26 + ], + [ + 283, + 0 + ], + [ + 298, + 44 + ] + ], + [ + [ + 334, + 17 + ], + [ + 315, + 36 + ], + [ + 352, + 26 + ], + [ + 317, + 9 + ], + [ + 372, + 8 + ], + [ + 336, + 0 + ] + ], + [ + [ + 309, + 54 + ], + [ + 327, + 37 + ], + [ + 287, + 44 + ], + [ + 329, + 64 + ], + [ + 270, + 61 + ], + [ + 311, + 80 + ] + ] + ] + }, + { + "task_id": "0002694", + "id": "RH20T_cfg5_task_0038_user_0007_scene_0009_cfg_0005_104122063678#308", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002694/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 242, + 160 + ], + [ + 259, + 152 + ], + [ + 253, + 143 + ], + [ + 230, + 179 + ], + [ + 237, + 150 + ], + [ + 214, + 186 + ] + ], + [ + [ + 221, + 169 + ], + [ + 220, + 150 + ], + [ + 218, + 191 + ], + [ + 224, + 149 + ], + [ + 219, + 208 + ], + [ + 225, + 166 + ] + ], + [ + [ + 246, + 162 + ], + [ + 228, + 150 + ], + [ + 266, + 160 + ], + [ + 228, + 164 + ], + [ + 282, + 171 + ], + [ + 244, + 175 + ] + ], + [ + [ + 221, + 173 + ], + [ + 225, + 148 + ], + [ + 247, + 183 + ], + [ + 196, + 164 + ], + [ + 243, + 206 + ], + [ + 193, + 187 + ] + ] + ] + }, + { + "task_id": "0002695", + "id": "RH20T_cfg4_task_0127_user_0014_scene_0010_cfg_0004_104422070011#114", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002695/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 79 + ], + [ + 380, + 50 + ], + [ + 359, + 78 + ], + [ + 384, + 81 + ], + [ + 351, + 105 + ], + [ + 374, + 111 + ] + ], + [ + [ + 393, + 85 + ], + [ + 404, + 70 + ], + [ + 372, + 100 + ], + [ + 412, + 71 + ], + [ + 363, + 113 + ], + [ + 401, + 85 + ] + ], + [ + [ + 381, + 86 + ], + [ + 394, + 78 + ], + [ + 367, + 61 + ], + [ + 395, + 111 + ], + [ + 352, + 69 + ], + [ + 381, + 121 + ] + ], + [ + [ + 382, + 126 + ], + [ + 389, + 101 + ], + [ + 390, + 139 + ], + [ + 373, + 111 + ], + [ + 383, + 164 + ], + [ + 365, + 138 + ] + ] + ] + }, + { + "task_id": "0002700", + "id": "RH20T_cfg3_task_0032_user_0011_scene_0001_cfg_0003_104122062823#384", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002700/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 208, + 95 + ], + [ + 186, + 118 + ], + [ + 185, + 69 + ], + [ + 229, + 119 + ], + [ + 207, + 47 + ], + [ + 249, + 98 + ] + ], + [ + [ + 212, + 89 + ], + [ + 218, + 63 + ], + [ + 214, + 105 + ], + [ + 209, + 72 + ], + [ + 208, + 131 + ], + [ + 202, + 101 + ] + ], + [ + [ + 224, + 127 + ], + [ + 202, + 139 + ], + [ + 203, + 132 + ], + [ + 246, + 121 + ], + [ + 222, + 121 + ], + [ + 265, + 110 + ] + ], + [ + [ + 224, + 114 + ], + [ + 235, + 92 + ], + [ + 248, + 134 + ], + [ + 201, + 94 + ], + [ + 237, + 154 + ], + [ + 192, + 115 + ] + ] + ] + }, + { + "task_id": "0002703", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0002_cfg_0006_104122061602#368", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002703/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 415, + 102 + ], + [ + 390, + 113 + ], + [ + 411, + 101 + ], + [ + 419, + 104 + ], + [ + 436, + 90 + ], + [ + 447, + 91 + ] + ], + [ + [ + 415, + 127 + ], + [ + 397, + 129 + ], + [ + 421, + 105 + ], + [ + 409, + 150 + ], + [ + 441, + 102 + ], + [ + 429, + 149 + ] + ], + [ + [ + 410, + 116 + ], + [ + 426, + 113 + ], + [ + 416, + 141 + ], + [ + 404, + 90 + ], + [ + 402, + 142 + ], + [ + 390, + 93 + ] + ], + [ + [ + 410, + 141 + ], + [ + 393, + 127 + ], + [ + 413, + 160 + ], + [ + 407, + 123 + ], + [ + 430, + 173 + ], + [ + 423, + 137 + ] + ] + ] + }, + { + "task_id": "0002705", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122060811#356", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002705/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 405, + 138 + ], + [ + 414, + 136 + ], + [ + 408, + 124 + ], + [ + 403, + 150 + ], + [ + 398, + 126 + ], + [ + 393, + 153 + ] + ], + [ + [ + 415, + 137 + ], + [ + 428, + 130 + ], + [ + 421, + 140 + ], + [ + 409, + 134 + ], + [ + 408, + 147 + ], + [ + 397, + 141 + ] + ], + [ + [ + 413, + 124 + ], + [ + 420, + 111 + ], + [ + 415, + 122 + ], + [ + 411, + 127 + ], + [ + 408, + 135 + ], + [ + 405, + 139 + ] + ], + [ + [ + 427, + 142 + ], + [ + 415, + 138 + ], + [ + 431, + 152 + ], + [ + 424, + 131 + ], + [ + 443, + 156 + ], + [ + 435, + 135 + ] + ] + ] + }, + { + "task_id": "0002706", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122060811#751", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002706/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 418, + 106 + ], + [ + 421, + 118 + ], + [ + 431, + 106 + ], + [ + 404, + 107 + ], + [ + 428, + 95 + ], + [ + 401, + 96 + ] + ], + [ + [ + 415, + 94 + ], + [ + 408, + 83 + ], + [ + 418, + 99 + ], + [ + 411, + 89 + ], + [ + 426, + 109 + ], + [ + 418, + 100 + ] + ], + [ + [ + 412, + 110 + ], + [ + 420, + 101 + ], + [ + 423, + 109 + ], + [ + 401, + 111 + ], + [ + 415, + 118 + ], + [ + 394, + 119 + ] + ], + [ + [ + 402, + 120 + ], + [ + 388, + 117 + ], + [ + 399, + 119 + ], + [ + 404, + 121 + ], + [ + 413, + 122 + ], + [ + 419, + 124 + ] + ] + ] + }, + { + "task_id": "0002707", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122060811#1159", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002707/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 442, + 122 + ], + [ + 441, + 111 + ], + [ + 441, + 113 + ], + [ + 444, + 131 + ], + [ + 442, + 124 + ], + [ + 445, + 142 + ] + ], + [ + [ + 455, + 121 + ], + [ + 467, + 112 + ], + [ + 462, + 123 + ], + [ + 449, + 118 + ], + [ + 450, + 131 + ], + [ + 437, + 127 + ] + ], + [ + [ + 466, + 107 + ], + [ + 474, + 103 + ], + [ + 472, + 96 + ], + [ + 460, + 118 + ], + [ + 463, + 100 + ], + [ + 452, + 122 + ] + ], + [ + [ + 470, + 128 + ], + [ + 470, + 132 + ], + [ + 484, + 122 + ], + [ + 457, + 135 + ], + [ + 485, + 118 + ], + [ + 457, + 131 + ] + ] + ] + }, + { + "task_id": "0002708", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122060811#1560", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002708/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 280, + 142 + ], + [ + 276, + 131 + ], + [ + 293, + 135 + ], + [ + 267, + 149 + ], + [ + 297, + 147 + ], + [ + 271, + 162 + ] + ], + [ + [ + 280, + 134 + ], + [ + 276, + 120 + ], + [ + 286, + 127 + ], + [ + 274, + 140 + ], + [ + 291, + 141 + ], + [ + 278, + 154 + ] + ], + [ + [ + 277, + 147 + ], + [ + 279, + 139 + ], + [ + 291, + 144 + ], + [ + 263, + 150 + ], + [ + 289, + 152 + ], + [ + 262, + 158 + ] + ], + [ + [ + 278, + 133 + ], + [ + 289, + 125 + ], + [ + 271, + 119 + ], + [ + 285, + 146 + ], + [ + 260, + 127 + ], + [ + 274, + 153 + ] + ] + ] + }, + { + "task_id": "0002709", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122060811#1977", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002709/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 291, + 121 + ], + [ + 287, + 110 + ], + [ + 304, + 115 + ], + [ + 278, + 127 + ], + [ + 308, + 126 + ], + [ + 282, + 138 + ] + ], + [ + [ + 298, + 124 + ], + [ + 312, + 119 + ], + [ + 300, + 130 + ], + [ + 296, + 117 + ], + [ + 287, + 135 + ], + [ + 282, + 123 + ] + ], + [ + [ + 293, + 129 + ], + [ + 291, + 115 + ], + [ + 296, + 125 + ], + [ + 290, + 133 + ], + [ + 299, + 140 + ], + [ + 292, + 147 + ] + ], + [ + [ + 297, + 107 + ], + [ + 292, + 119 + ], + [ + 294, + 112 + ], + [ + 301, + 102 + ], + [ + 299, + 99 + ], + [ + 306, + 89 + ] + ] + ] + }, + { + "task_id": "0002711", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0007_cfg_0004_104422070011#179", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002711/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 128 + ], + [ + 260, + 133 + ], + [ + 296, + 149 + ], + [ + 273, + 108 + ], + [ + 319, + 143 + ], + [ + 296, + 103 + ] + ], + [ + [ + 269, + 105 + ], + [ + 247, + 103 + ], + [ + 261, + 82 + ], + [ + 276, + 126 + ], + [ + 285, + 84 + ], + [ + 298, + 128 + ] + ], + [ + [ + 271, + 122 + ], + [ + 290, + 133 + ], + [ + 255, + 145 + ], + [ + 286, + 98 + ], + [ + 237, + 133 + ], + [ + 267, + 88 + ] + ], + [ + [ + 276, + 113 + ], + [ + 279, + 96 + ], + [ + 262, + 130 + ], + [ + 290, + 97 + ], + [ + 260, + 145 + ], + [ + 287, + 112 + ] + ] + ] + }, + { + "task_id": "0002713", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0007_cfg_0004_104422070011#408", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002713/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 301, + 170 + ], + [ + 311, + 158 + ], + [ + 290, + 191 + ], + [ + 313, + 150 + ], + [ + 282, + 200 + ], + [ + 304, + 161 + ] + ], + [ + [ + 299, + 168 + ], + [ + 309, + 182 + ], + [ + 274, + 173 + ], + [ + 324, + 163 + ], + [ + 262, + 158 + ], + [ + 314, + 147 + ] + ], + [ + [ + 294, + 161 + ], + [ + 274, + 148 + ], + [ + 311, + 148 + ], + [ + 276, + 175 + ], + [ + 332, + 160 + ], + [ + 297, + 188 + ] + ], + [ + [ + 325, + 150 + ], + [ + 304, + 162 + ], + [ + 313, + 154 + ], + [ + 337, + 145 + ], + [ + 332, + 143 + ], + [ + 358, + 132 + ] + ] + ] + }, + { + "task_id": "0002715", + "id": "RH20T_cfg5_task_0085_user_0007_scene_0002_cfg_0005_104122061850#314", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the bubble ring from the assembled bubble ring and ball'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002715/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 360, + 120 + ], + [ + 374, + 132 + ], + [ + 374, + 99 + ], + [ + 347, + 140 + ], + [ + 360, + 86 + ], + [ + 331, + 128 + ] + ], + [ + [ + 375, + 140 + ], + [ + 377, + 120 + ], + [ + 375, + 153 + ], + [ + 375, + 127 + ], + [ + 373, + 174 + ], + [ + 372, + 149 + ] + ], + [ + [ + 369, + 110 + ], + [ + 352, + 128 + ], + [ + 385, + 126 + ], + [ + 352, + 94 + ], + [ + 402, + 108 + ], + [ + 369, + 76 + ] + ], + [ + [ + 383, + 127 + ], + [ + 383, + 112 + ], + [ + 360, + 121 + ], + [ + 407, + 133 + ], + [ + 362, + 135 + ], + [ + 407, + 147 + ] + ] + ] + }, + { + "task_id": "0002717", + "id": "RH20T_cfg1_task_0029_user_0001_scene_0004_cfg_0001_750612070851#179", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002717/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 470, + 319 + ], + [ + 457, + 278 + ], + [ + 508, + 312 + ], + [ + 431, + 327 + ], + [ + 523, + 355 + ], + [ + 444, + 360 + ] + ], + [ + [ + 474, + 247 + ], + [ + 497, + 279 + ], + [ + 442, + 270 + ], + [ + 505, + 225 + ], + [ + 419, + 239 + ], + [ + 481, + 194 + ] + ], + [ + [ + 460, + 281 + ], + [ + 476, + 274 + ], + [ + 421, + 286 + ], + [ + 499, + 276 + ], + [ + 411, + 291 + ], + [ + 483, + 282 + ] + ], + [ + [ + 476, + 282 + ], + [ + 478, + 327 + ], + [ + 431, + 281 + ], + [ + 522, + 283 + ], + [ + 430, + 239 + ], + [ + 518, + 240 + ] + ] + ] + }, + { + "task_id": "0002718", + "id": "RH20T_cfg1_task_0029_user_0001_scene_0004_cfg_0001_750612070851#199", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002718/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 426, + 311 + ], + [ + 416, + 295 + ], + [ + 408, + 277 + ], + [ + 446, + 349 + ], + [ + 418, + 292 + ], + [ + 454, + 360 + ] + ], + [ + [ + 457, + 303 + ], + [ + 453, + 262 + ], + [ + 416, + 305 + ], + [ + 500, + 301 + ], + [ + 421, + 346 + ], + [ + 505, + 344 + ] + ], + [ + [ + 489, + 312 + ], + [ + 524, + 343 + ], + [ + 508, + 290 + ], + [ + 469, + 336 + ], + [ + 477, + 263 + ], + [ + 436, + 305 + ] + ], + [ + [ + 458, + 277 + ], + [ + 474, + 271 + ], + [ + 419, + 281 + ], + [ + 497, + 274 + ], + [ + 409, + 286 + ], + [ + 481, + 279 + ] + ] + ] + }, + { + "task_id": "0002720", + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122060811#105", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002720/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 390, + 119 + ], + [ + 391, + 107 + ], + [ + 381, + 126 + ], + [ + 399, + 112 + ], + [ + 380, + 137 + ], + [ + 398, + 123 + ] + ], + [ + [ + 404, + 129 + ], + [ + 393, + 122 + ], + [ + 407, + 117 + ], + [ + 401, + 140 + ], + [ + 419, + 124 + ], + [ + 412, + 147 + ] + ], + [ + [ + 401, + 113 + ], + [ + 406, + 103 + ], + [ + 413, + 112 + ], + [ + 388, + 114 + ], + [ + 408, + 121 + ], + [ + 383, + 123 + ] + ], + [ + [ + 406, + 128 + ], + [ + 399, + 126 + ], + [ + 413, + 115 + ], + [ + 400, + 141 + ], + [ + 421, + 117 + ], + [ + 408, + 144 + ] + ] + ] + }, + { + "task_id": "0002721", + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122060811#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002721/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 399, + 104 + ], + [ + 391, + 96 + ], + [ + 411, + 101 + ], + [ + 387, + 107 + ], + [ + 420, + 110 + ], + [ + 396, + 116 + ] + ], + [ + [ + 392, + 121 + ], + [ + 401, + 111 + ], + [ + 401, + 123 + ], + [ + 382, + 120 + ], + [ + 392, + 132 + ], + [ + 372, + 130 + ] + ], + [ + [ + 386, + 114 + ], + [ + 378, + 103 + ], + [ + 399, + 106 + ], + [ + 374, + 122 + ], + [ + 407, + 117 + ], + [ + 382, + 133 + ] + ], + [ + [ + 397, + 107 + ], + [ + 412, + 103 + ], + [ + 402, + 114 + ], + [ + 392, + 100 + ], + [ + 389, + 118 + ], + [ + 378, + 104 + ] + ] + ] + }, + { + "task_id": "0002723", + "id": "RH20T_cfg7_task_0007_user_0020_scene_0004_cfg_0007_104122061602#221", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002723/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 530, + 272 + ], + [ + 539, + 284 + ], + [ + 561, + 273 + ], + [ + 501, + 271 + ], + [ + 552, + 259 + ], + [ + 491, + 258 + ] + ], + [ + [ + 527, + 266 + ], + [ + 498, + 270 + ], + [ + 528, + 252 + ], + [ + 527, + 283 + ], + [ + 558, + 248 + ], + [ + 560, + 279 + ] + ], + [ + [ + 561, + 233 + ], + [ + 564, + 219 + ], + [ + 531, + 220 + ], + [ + 593, + 247 + ], + [ + 527, + 234 + ], + [ + 591, + 263 + ] + ], + [ + [ + 537, + 254 + ], + [ + 547, + 236 + ], + [ + 560, + 277 + ], + [ + 516, + 234 + ], + [ + 550, + 293 + ], + [ + 508, + 250 + ] + ] + ] + }, + { + "task_id": "0002724", + "id": "RH20T_cfg1_task_0015_user_0008_scene_0008_cfg_0001_750612070853#123", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002724/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 20, + 253 + ], + [ + 24, + 237 + ], + [ + 45, + 204 + ], + [ + 0, + 301 + ], + [ + 43, + 218 + ], + [ + 0, + 326 + ] + ], + [ + [ + 81, + 241 + ], + [ + 88, + 267 + ], + [ + 127, + 231 + ], + [ + 30, + 252 + ], + [ + 118, + 210 + ], + [ + 27, + 228 + ] + ], + [ + [ + 69, + 222 + ], + [ + 56, + 183 + ], + [ + 15, + 239 + ], + [ + 117, + 206 + ], + [ + 30, + 275 + ], + [ + 126, + 240 + ] + ], + [ + [ + 47, + 284 + ], + [ + 26, + 260 + ], + [ + 5, + 326 + ], + [ + 86, + 246 + ], + [ + 27, + 356 + ], + [ + 112, + 269 + ] + ] + ] + }, + { + "task_id": "0002726", + "id": "RH20T_cfg4_task_0035_user_0017_scene_0005_cfg_0004_104122062823#122", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up one small block'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002726/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 104 + ], + [ + 273, + 127 + ], + [ + 254, + 82 + ], + [ + 316, + 126 + ], + [ + 267, + 62 + ], + [ + 325, + 105 + ] + ], + [ + [ + 263, + 117 + ], + [ + 258, + 87 + ], + [ + 296, + 117 + ], + [ + 230, + 117 + ], + [ + 299, + 144 + ], + [ + 236, + 145 + ] + ], + [ + [ + 277, + 133 + ], + [ + 275, + 119 + ], + [ + 309, + 139 + ], + [ + 243, + 126 + ], + [ + 314, + 155 + ], + [ + 243, + 141 + ] + ], + [ + [ + 239, + 118 + ], + [ + 228, + 95 + ], + [ + 273, + 104 + ], + [ + 205, + 132 + ], + [ + 287, + 128 + ], + [ + 216, + 158 + ] + ] + ] + }, + { + "task_id": "0002727", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0007_cfg_0006_104122061330#157", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002727/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 293, + 186 + ], + [ + 292, + 213 + ], + [ + 265, + 185 + ], + [ + 321, + 186 + ], + [ + 265, + 157 + ], + [ + 323, + 158 + ] + ], + [ + [ + 288, + 188 + ], + [ + 264, + 181 + ], + [ + 295, + 203 + ], + [ + 280, + 171 + ], + [ + 319, + 211 + ], + [ + 306, + 179 + ] + ], + [ + [ + 286, + 157 + ], + [ + 273, + 165 + ], + [ + 310, + 150 + ], + [ + 263, + 163 + ], + [ + 321, + 143 + ], + [ + 275, + 156 + ] + ], + [ + [ + 270, + 165 + ], + [ + 254, + 176 + ], + [ + 268, + 189 + ], + [ + 273, + 140 + ], + [ + 282, + 179 + ], + [ + 288, + 131 + ] + ] + ] + }, + { + "task_id": "0002728", + "id": "RH20T_cfg5_task_0008_user_0007_scene_0001_cfg_0005_104122061850#294", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002728/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 227, + 234 + ], + [ + 233, + 209 + ], + [ + 201, + 226 + ], + [ + 252, + 242 + ], + [ + 195, + 251 + ], + [ + 246, + 267 + ] + ], + [ + [ + 227, + 261 + ], + [ + 226, + 280 + ], + [ + 209, + 243 + ], + [ + 247, + 280 + ], + [ + 209, + 222 + ], + [ + 249, + 259 + ] + ], + [ + [ + 224, + 267 + ], + [ + 203, + 252 + ], + [ + 242, + 260 + ], + [ + 207, + 274 + ], + [ + 261, + 275 + ], + [ + 226, + 288 + ] + ], + [ + [ + 229, + 249 + ], + [ + 215, + 237 + ], + [ + 211, + 272 + ], + [ + 246, + 226 + ], + [ + 225, + 282 + ], + [ + 257, + 237 + ] + ] + ] + }, + { + "task_id": "0002729", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_036422060215#203", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002729/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 287, + 148 + ], + [ + 300, + 163 + ], + [ + 284, + 138 + ], + [ + 289, + 157 + ], + [ + 270, + 123 + ], + [ + 277, + 143 + ] + ], + [ + [ + 302, + 136 + ], + [ + 290, + 123 + ], + [ + 289, + 151 + ], + [ + 315, + 120 + ], + [ + 300, + 164 + ], + [ + 327, + 133 + ] + ], + [ + [ + 298, + 138 + ], + [ + 302, + 120 + ], + [ + 316, + 134 + ], + [ + 280, + 142 + ], + [ + 312, + 151 + ], + [ + 276, + 159 + ] + ], + [ + [ + 296, + 158 + ], + [ + 292, + 143 + ], + [ + 298, + 144 + ], + [ + 294, + 171 + ], + [ + 302, + 161 + ], + [ + 298, + 187 + ] + ] + ] + }, + { + "task_id": "0002730", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_036422060215#306", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002730/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 264, + 147 + ], + [ + 268, + 128 + ], + [ + 284, + 143 + ], + [ + 244, + 151 + ], + [ + 280, + 160 + ], + [ + 241, + 168 + ] + ], + [ + [ + 254, + 135 + ], + [ + 254, + 123 + ], + [ + 234, + 145 + ], + [ + 273, + 125 + ], + [ + 233, + 158 + ], + [ + 274, + 138 + ] + ], + [ + [ + 254, + 135 + ], + [ + 266, + 140 + ], + [ + 274, + 129 + ], + [ + 234, + 142 + ], + [ + 263, + 125 + ], + [ + 224, + 138 + ] + ], + [ + [ + 265, + 161 + ], + [ + 275, + 158 + ], + [ + 245, + 157 + ], + [ + 285, + 164 + ], + [ + 236, + 160 + ], + [ + 276, + 167 + ] + ] + ] + }, + { + "task_id": "0002733", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_037522062165#330", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002733/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 216, + 211 + ], + [ + 239, + 218 + ], + [ + 219, + 184 + ], + [ + 213, + 239 + ], + [ + 195, + 177 + ], + [ + 187, + 234 + ] + ], + [ + [ + 208, + 224 + ], + [ + 200, + 211 + ], + [ + 198, + 201 + ], + [ + 218, + 249 + ], + [ + 206, + 214 + ], + [ + 224, + 261 + ] + ], + [ + [ + 183, + 223 + ], + [ + 214, + 221 + ], + [ + 180, + 201 + ], + [ + 185, + 243 + ], + [ + 146, + 203 + ], + [ + 154, + 245 + ] + ], + [ + [ + 191, + 261 + ], + [ + 159, + 270 + ], + [ + 185, + 233 + ], + [ + 197, + 288 + ], + [ + 216, + 225 + ], + [ + 226, + 279 + ] + ] + ] + }, + { + "task_id": "0002734", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_037522062165#410", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002734/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 255, + 305 + ], + [ + 279, + 305 + ], + [ + 238, + 298 + ], + [ + 274, + 312 + ], + [ + 212, + 298 + ], + [ + 247, + 312 + ] + ], + [ + [ + 261, + 236 + ], + [ + 235, + 246 + ], + [ + 241, + 221 + ], + [ + 281, + 252 + ], + [ + 264, + 213 + ], + [ + 304, + 242 + ] + ], + [ + [ + 236, + 248 + ], + [ + 212, + 264 + ], + [ + 222, + 260 + ], + [ + 252, + 234 + ], + [ + 243, + 245 + ], + [ + 273, + 219 + ] + ], + [ + [ + 257, + 269 + ], + [ + 252, + 256 + ], + [ + 235, + 251 + ], + [ + 280, + 289 + ], + [ + 241, + 263 + ], + [ + 283, + 300 + ] + ] + ] + }, + { + "task_id": "0002736", + "id": "RH20T_cfg6_task_0209_user_0014_scene_0004_cfg_0006_037522061512#94", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slice the carrots'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002736/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 187, + 229 + ], + [ + 213, + 243 + ], + [ + 190, + 228 + ], + [ + 184, + 229 + ], + [ + 166, + 215 + ], + [ + 157, + 215 + ] + ], + [ + [ + 173, + 205 + ], + [ + 155, + 213 + ], + [ + 174, + 181 + ], + [ + 172, + 229 + ], + [ + 193, + 172 + ], + [ + 192, + 221 + ] + ], + [ + [ + 181, + 221 + ], + [ + 168, + 204 + ], + [ + 202, + 203 + ], + [ + 159, + 240 + ], + [ + 212, + 219 + ], + [ + 172, + 255 + ] + ], + [ + [ + 205, + 216 + ], + [ + 195, + 206 + ], + [ + 210, + 240 + ], + [ + 199, + 191 + ], + [ + 221, + 251 + ], + [ + 210, + 201 + ] + ] + ] + }, + { + "task_id": "0002742", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0004_cfg_0004_104122062823#497", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002742/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 316, + 49 + ], + [ + 322, + 87 + ], + [ + 355, + 45 + ], + [ + 278, + 54 + ], + [ + 351, + 3 + ], + [ + 271, + 14 + ] + ], + [ + [ + 328, + 56 + ], + [ + 323, + 93 + ], + [ + 299, + 46 + ], + [ + 359, + 66 + ], + [ + 303, + 7 + ], + [ + 366, + 24 + ] + ], + [ + [ + 303, + 45 + ], + [ + 337, + 30 + ], + [ + 305, + 84 + ], + [ + 301, + 9 + ], + [ + 275, + 96 + ], + [ + 271, + 24 + ] + ], + [ + [ + 315, + 44 + ], + [ + 317, + 2 + ], + [ + 352, + 39 + ], + [ + 275, + 50 + ], + [ + 348, + 76 + ], + [ + 275, + 88 + ] + ] + ] + }, + { + "task_id": "0002746", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0008_cfg_0006_104122064161#708", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002746/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 342, + 194 + ], + [ + 340, + 191 + ], + [ + 356, + 172 + ], + [ + 328, + 215 + ], + [ + 357, + 176 + ], + [ + 330, + 216 + ] + ], + [ + [ + 335, + 206 + ], + [ + 309, + 200 + ], + [ + 339, + 196 + ], + [ + 330, + 218 + ], + [ + 363, + 201 + ], + [ + 357, + 224 + ] + ], + [ + [ + 360, + 201 + ], + [ + 364, + 192 + ], + [ + 357, + 227 + ], + [ + 362, + 176 + ], + [ + 354, + 234 + ], + [ + 358, + 185 + ] + ], + [ + [ + 374, + 197 + ], + [ + 370, + 197 + ], + [ + 348, + 192 + ], + [ + 400, + 202 + ], + [ + 351, + 191 + ], + [ + 405, + 201 + ] + ] + ] + }, + { + "task_id": "0002747", + "id": "RH20T_cfg1_task_0005_user_0015_scene_0007_cfg_0001_819112071037#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002747/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 102 + ], + [ + 363, + 40 + ], + [ + 290, + 94 + ], + [ + 399, + 110 + ], + [ + 277, + 155 + ], + [ + 384, + 165 + ] + ], + [ + [ + 329, + 99 + ], + [ + 359, + 41 + ], + [ + 370, + 123 + ], + [ + 279, + 71 + ], + [ + 342, + 175 + ], + [ + 247, + 133 + ] + ], + [ + [ + 362, + 142 + ], + [ + 333, + 186 + ], + [ + 395, + 172 + ], + [ + 326, + 109 + ], + [ + 422, + 132 + ], + [ + 356, + 66 + ] + ], + [ + [ + 348, + 84 + ], + [ + 387, + 135 + ], + [ + 304, + 127 + ], + [ + 397, + 37 + ], + [ + 270, + 80 + ], + [ + 357, + 0 + ] + ] + ] + }, + { + "task_id": "0002748", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_104122061850#217", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002748/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 355, + 181 + ], + [ + 356, + 165 + ], + [ + 349, + 207 + ], + [ + 361, + 156 + ], + [ + 349, + 220 + ], + [ + 361, + 171 + ] + ], + [ + [ + 360, + 178 + ], + [ + 390, + 180 + ], + [ + 358, + 206 + ], + [ + 362, + 151 + ], + [ + 327, + 204 + ], + [ + 333, + 149 + ] + ], + [ + [ + 334, + 196 + ], + [ + 309, + 189 + ], + [ + 349, + 176 + ], + [ + 319, + 219 + ], + [ + 375, + 183 + ], + [ + 346, + 228 + ] + ], + [ + [ + 370, + 161 + ], + [ + 388, + 182 + ], + [ + 346, + 169 + ], + [ + 394, + 153 + ], + [ + 330, + 148 + ], + [ + 377, + 134 + ] + ] + ] + }, + { + "task_id": "0002749", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_104122061850#344", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002749/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 166 + ], + [ + 346, + 154 + ], + [ + 323, + 180 + ], + [ + 371, + 153 + ], + [ + 325, + 191 + ], + [ + 370, + 165 + ] + ], + [ + [ + 332, + 152 + ], + [ + 314, + 132 + ], + [ + 343, + 144 + ], + [ + 320, + 162 + ], + [ + 361, + 163 + ], + [ + 340, + 183 + ] + ], + [ + [ + 348, + 189 + ], + [ + 365, + 203 + ], + [ + 325, + 201 + ], + [ + 369, + 177 + ], + [ + 309, + 188 + ], + [ + 352, + 165 + ] + ], + [ + [ + 344, + 167 + ], + [ + 372, + 166 + ], + [ + 349, + 190 + ], + [ + 340, + 146 + ], + [ + 320, + 191 + ], + [ + 313, + 147 + ] + ] + ] + }, + { + "task_id": "0002750", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0006_cfg_0005_104122061850#424", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002750/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 376, + 115 + ], + [ + 375, + 91 + ], + [ + 350, + 116 + ], + [ + 403, + 113 + ], + [ + 350, + 140 + ], + [ + 404, + 138 + ] + ], + [ + [ + 394, + 108 + ], + [ + 388, + 80 + ], + [ + 419, + 101 + ], + [ + 369, + 114 + ], + [ + 425, + 128 + ], + [ + 375, + 141 + ] + ], + [ + [ + 364, + 147 + ], + [ + 388, + 139 + ], + [ + 355, + 122 + ], + [ + 373, + 171 + ], + [ + 331, + 130 + ], + [ + 348, + 179 + ] + ], + [ + [ + 384, + 120 + ], + [ + 384, + 105 + ], + [ + 369, + 140 + ], + [ + 399, + 100 + ], + [ + 370, + 153 + ], + [ + 398, + 114 + ] + ] + ] + }, + { + "task_id": "0002751", + "id": "RH20T_cfg1_task_0038_user_0014_scene_0004_cfg_0001_039422060546#79", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002751/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 289, + 55 + ], + [ + 286, + 30 + ], + [ + 290, + 77 + ], + [ + 288, + 35 + ], + [ + 292, + 98 + ], + [ + 290, + 58 + ] + ], + [ + [ + 295, + 42 + ], + [ + 274, + 64 + ], + [ + 317, + 60 + ], + [ + 271, + 22 + ], + [ + 338, + 38 + ], + [ + 293, + 0 + ] + ], + [ + [ + 266, + 65 + ], + [ + 293, + 57 + ], + [ + 269, + 62 + ], + [ + 263, + 69 + ], + [ + 244, + 69 + ], + [ + 235, + 77 + ] + ], + [ + [ + 279, + 45 + ], + [ + 254, + 60 + ], + [ + 274, + 25 + ], + [ + 283, + 62 + ], + [ + 299, + 10 + ], + [ + 306, + 48 + ] + ] + ] + }, + { + "task_id": "0002753", + "id": "RH20T_cfg4_task_0014_user_0010_scene_0005_cfg_0004_038522062288#124", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the object from the scale'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002753/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 240, + 209 + ], + [ + 233, + 194 + ], + [ + 219, + 190 + ], + [ + 261, + 228 + ], + [ + 227, + 203 + ], + [ + 266, + 239 + ] + ], + [ + [ + 242, + 188 + ], + [ + 229, + 215 + ], + [ + 271, + 200 + ], + [ + 213, + 175 + ], + [ + 283, + 174 + ], + [ + 226, + 149 + ] + ], + [ + [ + 210, + 220 + ], + [ + 186, + 241 + ], + [ + 197, + 210 + ], + [ + 224, + 231 + ], + [ + 219, + 191 + ], + [ + 247, + 210 + ] + ], + [ + [ + 260, + 233 + ], + [ + 279, + 214 + ], + [ + 266, + 254 + ], + [ + 253, + 210 + ], + [ + 246, + 275 + ], + [ + 230, + 231 + ] + ] + ] + }, + { + "task_id": "0002754", + "id": "RH20T_cfg1_task_0105_user_0010_scene_0005_cfg_0001_038522062547#120", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002754/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 474, + 119 + ], + [ + 483, + 102 + ], + [ + 485, + 145 + ], + [ + 463, + 93 + ], + [ + 476, + 159 + ], + [ + 456, + 109 + ] + ], + [ + [ + 493, + 94 + ], + [ + 484, + 102 + ], + [ + 525, + 94 + ], + [ + 460, + 95 + ], + [ + 537, + 85 + ], + [ + 468, + 87 + ] + ], + [ + [ + 484, + 116 + ], + [ + 452, + 127 + ], + [ + 486, + 99 + ], + [ + 482, + 130 + ], + [ + 522, + 86 + ], + [ + 513, + 120 + ] + ], + [ + [ + 491, + 154 + ], + [ + 486, + 156 + ], + [ + 506, + 180 + ], + [ + 475, + 128 + ], + [ + 510, + 176 + ], + [ + 481, + 127 + ] + ] + ] + }, + { + "task_id": "0002757", + "id": "RH20T_cfg1_task_0064_user_0015_scene_0010_cfg_0001_035622060973#145", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002757/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 249, + 215 + ], + [ + 241, + 202 + ], + [ + 230, + 211 + ], + [ + 270, + 219 + ], + [ + 238, + 222 + ], + [ + 277, + 230 + ] + ], + [ + [ + 248, + 197 + ], + [ + 245, + 178 + ], + [ + 225, + 199 + ], + [ + 270, + 194 + ], + [ + 228, + 218 + ], + [ + 274, + 213 + ] + ], + [ + [ + 255, + 220 + ], + [ + 262, + 233 + ], + [ + 270, + 231 + ], + [ + 240, + 209 + ], + [ + 263, + 219 + ], + [ + 234, + 197 + ] + ], + [ + [ + 237, + 233 + ], + [ + 221, + 216 + ], + [ + 243, + 231 + ], + [ + 232, + 235 + ], + [ + 260, + 248 + ], + [ + 248, + 251 + ] + ] + ] + }, + { + "task_id": "0002758", + "id": "RH20T_cfg4_task_0013_user_0016_scene_0007_cfg_0004_038522062288#120", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002758/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 210, + 304 + ], + [ + 214, + 336 + ], + [ + 194, + 315 + ], + [ + 227, + 292 + ], + [ + 191, + 286 + ], + [ + 223, + 261 + ] + ], + [ + [ + 218, + 300 + ], + [ + 210, + 288 + ], + [ + 200, + 275 + ], + [ + 236, + 326 + ], + [ + 208, + 287 + ], + [ + 242, + 335 + ] + ], + [ + [ + 210, + 283 + ], + [ + 181, + 303 + ], + [ + 215, + 289 + ], + [ + 205, + 278 + ], + [ + 245, + 269 + ], + [ + 231, + 261 + ] + ], + [ + [ + 220, + 286 + ], + [ + 248, + 266 + ], + [ + 201, + 257 + ], + [ + 239, + 314 + ], + [ + 170, + 278 + ], + [ + 210, + 336 + ] + ] + ] + }, + { + "task_id": "0002761", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0005_cfg_0005_037522062165#217", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002761/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 233, + 228 + ], + [ + 218, + 247 + ], + [ + 231, + 208 + ], + [ + 235, + 248 + ], + [ + 246, + 191 + ], + [ + 248, + 231 + ] + ], + [ + [ + 211, + 202 + ], + [ + 206, + 188 + ], + [ + 188, + 189 + ], + [ + 235, + 216 + ], + [ + 194, + 202 + ], + [ + 238, + 228 + ] + ], + [ + [ + 201, + 199 + ], + [ + 184, + 221 + ], + [ + 223, + 215 + ], + [ + 177, + 182 + ], + [ + 240, + 193 + ], + [ + 194, + 160 + ] + ], + [ + [ + 199, + 175 + ], + [ + 201, + 163 + ], + [ + 223, + 160 + ], + [ + 175, + 189 + ], + [ + 222, + 174 + ], + [ + 171, + 204 + ] + ] + ] + }, + { + "task_id": "0002765", + "id": "RH20T_cfg6_task_0225_user_0014_scene_0006_cfg_0006_104122060811#197", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002765/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 392, + 114 + ], + [ + 388, + 118 + ], + [ + 406, + 122 + ], + [ + 377, + 105 + ], + [ + 409, + 119 + ], + [ + 381, + 102 + ] + ], + [ + [ + 402, + 123 + ], + [ + 409, + 111 + ], + [ + 414, + 122 + ], + [ + 388, + 124 + ], + [ + 408, + 133 + ], + [ + 382, + 135 + ] + ], + [ + [ + 389, + 106 + ], + [ + 389, + 123 + ], + [ + 404, + 108 + ], + [ + 374, + 105 + ], + [ + 405, + 92 + ], + [ + 375, + 89 + ] + ], + [ + [ + 393, + 114 + ], + [ + 396, + 110 + ], + [ + 380, + 122 + ], + [ + 406, + 106 + ], + [ + 376, + 127 + ], + [ + 402, + 110 + ] + ] + ] + }, + { + "task_id": "0002766", + "id": "RH20T_cfg6_task_0225_user_0014_scene_0006_cfg_0006_104122060811#439", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002766/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 415, + 56 + ], + [ + 432, + 51 + ], + [ + 421, + 60 + ], + [ + 409, + 52 + ], + [ + 405, + 66 + ], + [ + 393, + 57 + ] + ], + [ + [ + 414, + 44 + ], + [ + 407, + 31 + ], + [ + 426, + 46 + ], + [ + 402, + 42 + ], + [ + 433, + 59 + ], + [ + 409, + 55 + ] + ], + [ + [ + 422, + 40 + ], + [ + 424, + 57 + ], + [ + 439, + 37 + ], + [ + 406, + 43 + ], + [ + 437, + 19 + ], + [ + 404, + 25 + ] + ], + [ + [ + 399, + 46 + ], + [ + 416, + 48 + ], + [ + 402, + 29 + ], + [ + 396, + 64 + ], + [ + 386, + 28 + ], + [ + 380, + 62 + ] + ] + ] + }, + { + "task_id": "0002767", + "id": "RH20T_cfg6_task_0225_user_0014_scene_0006_cfg_0006_104122060811#777", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002767/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 320, + 122 + ], + [ + 310, + 110 + ], + [ + 329, + 124 + ], + [ + 311, + 120 + ], + [ + 339, + 136 + ], + [ + 322, + 132 + ] + ], + [ + [ + 314, + 133 + ], + [ + 312, + 119 + ], + [ + 298, + 138 + ], + [ + 330, + 128 + ], + [ + 300, + 153 + ], + [ + 333, + 143 + ] + ], + [ + [ + 313, + 120 + ], + [ + 319, + 103 + ], + [ + 301, + 114 + ], + [ + 327, + 127 + ], + [ + 295, + 131 + ], + [ + 321, + 144 + ] + ], + [ + [ + 319, + 139 + ], + [ + 324, + 131 + ], + [ + 335, + 137 + ], + [ + 303, + 141 + ], + [ + 330, + 145 + ], + [ + 299, + 149 + ] + ] + ] + }, + { + "task_id": "0002768", + "id": "RH20T_cfg7_task_0029_user_0020_scene_0009_cfg_0007_104122061330#207", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002768/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 373, + 92 + ], + [ + 366, + 94 + ], + [ + 366, + 115 + ], + [ + 381, + 68 + ], + [ + 373, + 112 + ], + [ + 388, + 67 + ] + ], + [ + [ + 379, + 90 + ], + [ + 363, + 101 + ], + [ + 371, + 68 + ], + [ + 388, + 113 + ], + [ + 388, + 56 + ], + [ + 405, + 103 + ] + ], + [ + [ + 366, + 73 + ], + [ + 365, + 73 + ], + [ + 389, + 70 + ], + [ + 342, + 76 + ], + [ + 389, + 70 + ], + [ + 344, + 76 + ] + ], + [ + [ + 360, + 78 + ], + [ + 360, + 58 + ], + [ + 346, + 66 + ], + [ + 374, + 90 + ], + [ + 345, + 87 + ], + [ + 374, + 110 + ] + ] + ] + }, + { + "task_id": "0002769", + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_039422060546#128", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002769/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 340, + 277 + ], + [ + 338, + 259 + ], + [ + 375, + 287 + ], + [ + 307, + 268 + ], + [ + 375, + 303 + ], + [ + 310, + 284 + ] + ], + [ + [ + 307, + 278 + ], + [ + 311, + 245 + ], + [ + 322, + 294 + ], + [ + 293, + 264 + ], + [ + 318, + 328 + ], + [ + 290, + 294 + ] + ], + [ + [ + 312, + 282 + ], + [ + 332, + 312 + ], + [ + 282, + 305 + ], + [ + 341, + 260 + ], + [ + 264, + 277 + ], + [ + 321, + 235 + ] + ], + [ + [ + 330, + 247 + ], + [ + 306, + 231 + ], + [ + 303, + 267 + ], + [ + 357, + 226 + ], + [ + 326, + 281 + ], + [ + 378, + 242 + ] + ] + ] + }, + { + "task_id": "0002770", + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_039422060546#252", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002770/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 345, + 259 + ], + [ + 343, + 240 + ], + [ + 381, + 267 + ], + [ + 311, + 250 + ], + [ + 380, + 284 + ], + [ + 315, + 267 + ] + ], + [ + [ + 388, + 227 + ], + [ + 403, + 200 + ], + [ + 361, + 204 + ], + [ + 414, + 250 + ], + [ + 344, + 233 + ], + [ + 400, + 280 + ] + ], + [ + [ + 334, + 273 + ], + [ + 302, + 255 + ], + [ + 316, + 307 + ], + [ + 351, + 240 + ], + [ + 348, + 324 + ], + [ + 382, + 257 + ] + ], + [ + [ + 326, + 291 + ], + [ + 341, + 263 + ], + [ + 345, + 314 + ], + [ + 306, + 267 + ], + [ + 330, + 346 + ], + [ + 288, + 298 + ] + ] + ] + }, + { + "task_id": "0002775", + "id": "RH20T_cfg4_task_0055_user_0010_scene_0005_cfg_0004_036422060909#80", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002775/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 272, + 173 + ], + [ + 280, + 179 + ], + [ + 302, + 184 + ], + [ + 241, + 163 + ], + [ + 293, + 178 + ], + [ + 236, + 158 + ] + ], + [ + [ + 254, + 175 + ], + [ + 220, + 176 + ], + [ + 254, + 206 + ], + [ + 253, + 145 + ], + [ + 289, + 204 + ], + [ + 286, + 144 + ] + ], + [ + [ + 253, + 157 + ], + [ + 263, + 144 + ], + [ + 275, + 142 + ], + [ + 230, + 172 + ], + [ + 265, + 153 + ], + [ + 222, + 183 + ] + ], + [ + [ + 278, + 130 + ], + [ + 285, + 161 + ], + [ + 250, + 137 + ], + [ + 308, + 123 + ], + [ + 243, + 107 + ], + [ + 301, + 92 + ] + ] + ] + }, + { + "task_id": "0002776", + "id": "RH20T_cfg4_task_0055_user_0010_scene_0005_cfg_0004_036422060909#166", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002776/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 376, + 143 + ], + [ + 358, + 123 + ], + [ + 365, + 142 + ], + [ + 387, + 144 + ], + [ + 385, + 164 + ], + [ + 405, + 165 + ] + ], + [ + [ + 334, + 132 + ], + [ + 341, + 107 + ], + [ + 324, + 143 + ], + [ + 345, + 120 + ], + [ + 317, + 168 + ], + [ + 337, + 148 + ] + ], + [ + [ + 384, + 127 + ], + [ + 357, + 113 + ], + [ + 387, + 130 + ], + [ + 382, + 124 + ], + [ + 416, + 145 + ], + [ + 407, + 137 + ] + ], + [ + [ + 350, + 135 + ], + [ + 351, + 112 + ], + [ + 365, + 118 + ], + [ + 333, + 153 + ], + [ + 363, + 139 + ], + [ + 332, + 174 + ] + ] + ] + }, + { + "task_id": "0002777", + "id": "RH20T_cfg4_task_0055_user_0010_scene_0005_cfg_0004_036422060909#209", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002777/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 367, + 133 + ], + [ + 340, + 133 + ], + [ + 357, + 147 + ], + [ + 379, + 119 + ], + [ + 382, + 146 + ], + [ + 407, + 119 + ] + ], + [ + [ + 358, + 128 + ], + [ + 336, + 128 + ], + [ + 341, + 128 + ], + [ + 377, + 129 + ], + [ + 360, + 128 + ], + [ + 397, + 129 + ] + ], + [ + [ + 353, + 148 + ], + [ + 357, + 130 + ], + [ + 372, + 135 + ], + [ + 334, + 163 + ], + [ + 368, + 152 + ], + [ + 332, + 179 + ] + ], + [ + [ + 342, + 140 + ], + [ + 334, + 127 + ], + [ + 337, + 116 + ], + [ + 348, + 165 + ], + [ + 344, + 129 + ], + [ + 356, + 176 + ] + ] + ] + }, + { + "task_id": "0002778", + "id": "RH20T_cfg4_task_0055_user_0010_scene_0005_cfg_0004_036422060909#352", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002778/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 368, + 93 + ], + [ + 375, + 68 + ], + [ + 365, + 102 + ], + [ + 371, + 85 + ], + [ + 358, + 127 + ], + [ + 365, + 108 + ] + ], + [ + [ + 372, + 75 + ], + [ + 384, + 55 + ], + [ + 378, + 61 + ], + [ + 366, + 90 + ], + [ + 367, + 79 + ], + [ + 355, + 108 + ] + ], + [ + [ + 387, + 92 + ], + [ + 379, + 73 + ], + [ + 379, + 111 + ], + [ + 394, + 75 + ], + [ + 387, + 128 + ], + [ + 401, + 92 + ] + ], + [ + [ + 344, + 72 + ], + [ + 354, + 70 + ], + [ + 325, + 56 + ], + [ + 362, + 87 + ], + [ + 317, + 59 + ], + [ + 353, + 89 + ] + ] + ] + }, + { + "task_id": "0002781", + "id": "RH20T_cfg6_task_0225_user_0014_scene_0006_cfg_0006_104122061602#179", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002781/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 493, + 136 + ], + [ + 472, + 148 + ], + [ + 470, + 140 + ], + [ + 520, + 130 + ], + [ + 490, + 129 + ], + [ + 541, + 118 + ] + ], + [ + [ + 528, + 152 + ], + [ + 512, + 179 + ], + [ + 498, + 138 + ], + [ + 560, + 166 + ], + [ + 514, + 111 + ], + [ + 577, + 138 + ] + ], + [ + [ + 504, + 153 + ], + [ + 478, + 144 + ], + [ + 514, + 166 + ], + [ + 495, + 142 + ], + [ + 540, + 175 + ], + [ + 519, + 150 + ] + ], + [ + [ + 517, + 162 + ], + [ + 532, + 142 + ], + [ + 540, + 185 + ], + [ + 496, + 140 + ], + [ + 525, + 202 + ], + [ + 484, + 158 + ] + ] + ] + }, + { + "task_id": "0002782", + "id": "RH20T_cfg6_task_0225_user_0014_scene_0006_cfg_0006_104122061602#279", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002782/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 531, + 101 + ], + [ + 503, + 109 + ], + [ + 529, + 130 + ], + [ + 532, + 74 + ], + [ + 561, + 123 + ], + [ + 563, + 64 + ] + ], + [ + [ + 561, + 76 + ], + [ + 527, + 66 + ], + [ + 577, + 51 + ], + [ + 546, + 98 + ], + [ + 615, + 63 + ], + [ + 579, + 109 + ] + ], + [ + [ + 560, + 97 + ], + [ + 588, + 94 + ], + [ + 581, + 73 + ], + [ + 542, + 118 + ], + [ + 550, + 75 + ], + [ + 514, + 121 + ] + ], + [ + [ + 536, + 104 + ], + [ + 553, + 89 + ], + [ + 551, + 133 + ], + [ + 522, + 75 + ], + [ + 535, + 144 + ], + [ + 509, + 90 + ] + ] + ] + }, + { + "task_id": "0002784", + "id": "RH20T_cfg6_task_0225_user_0014_scene_0006_cfg_0006_104122061602#446", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002784/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 587, + 55 + ], + [ + 559, + 76 + ], + [ + 570, + 25 + ], + [ + 603, + 85 + ], + [ + 598, + 4 + ], + [ + 630, + 64 + ] + ], + [ + [ + 603, + 62 + ], + [ + 630, + 59 + ], + [ + 613, + 95 + ], + [ + 592, + 28 + ], + [ + 589, + 95 + ], + [ + 569, + 32 + ] + ], + [ + [ + 609, + 76 + ], + [ + 600, + 100 + ], + [ + 596, + 96 + ], + [ + 624, + 52 + ], + [ + 605, + 74 + ], + [ + 634, + 28 + ] + ], + [ + [ + 640, + 53 + ], + [ + 640, + 88 + ], + [ + 640, + 56 + ], + [ + 634, + 50 + ], + [ + 640, + 23 + ], + [ + 624, + 15 + ] + ] + ] + }, + { + "task_id": "0002785", + "id": "RH20T_cfg6_task_0225_user_0014_scene_0006_cfg_0006_104122061602#785", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002785/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 385, + 36 + ], + [ + 392, + 63 + ], + [ + 359, + 43 + ], + [ + 413, + 29 + ], + [ + 352, + 17 + ], + [ + 406, + 1 + ] + ], + [ + [ + 373, + 34 + ], + [ + 377, + 12 + ], + [ + 386, + 51 + ], + [ + 361, + 19 + ], + [ + 381, + 71 + ], + [ + 358, + 40 + ] + ], + [ + [ + 339, + 45 + ], + [ + 354, + 57 + ], + [ + 332, + 27 + ], + [ + 345, + 63 + ], + [ + 317, + 16 + ], + [ + 330, + 52 + ] + ], + [ + [ + 384, + 20 + ], + [ + 406, + 10 + ], + [ + 399, + 29 + ], + [ + 369, + 11 + ], + [ + 379, + 39 + ], + [ + 349, + 21 + ] + ] + ] + }, + { + "task_id": "0002786", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0005_cfg_0001_035622060973#170", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002786/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 228, + 123 + ], + [ + 214, + 110 + ], + [ + 225, + 138 + ], + [ + 231, + 108 + ], + [ + 239, + 150 + ], + [ + 244, + 120 + ] + ], + [ + [ + 246, + 131 + ], + [ + 261, + 138 + ], + [ + 258, + 115 + ], + [ + 235, + 147 + ], + [ + 241, + 106 + ], + [ + 218, + 140 + ] + ], + [ + [ + 214, + 136 + ], + [ + 230, + 131 + ], + [ + 222, + 155 + ], + [ + 207, + 116 + ], + [ + 206, + 161 + ], + [ + 190, + 121 + ] + ], + [ + [ + 221, + 133 + ], + [ + 220, + 120 + ], + [ + 209, + 120 + ], + [ + 233, + 147 + ], + [ + 211, + 132 + ], + [ + 234, + 158 + ] + ] + ] + }, + { + "task_id": "0002788", + "id": "RH20T_cfg5_task_0096_user_0007_scene_0003_cfg_0005_036422060215#267", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002788/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 301, + 150 + ], + [ + 283, + 164 + ], + [ + 315, + 170 + ], + [ + 288, + 131 + ], + [ + 334, + 156 + ], + [ + 307, + 117 + ] + ], + [ + [ + 282, + 151 + ], + [ + 289, + 139 + ], + [ + 286, + 170 + ], + [ + 278, + 132 + ], + [ + 279, + 183 + ], + [ + 270, + 144 + ] + ], + [ + [ + 255, + 145 + ], + [ + 273, + 137 + ], + [ + 244, + 124 + ], + [ + 266, + 167 + ], + [ + 224, + 132 + ], + [ + 248, + 176 + ] + ], + [ + [ + 284, + 138 + ], + [ + 287, + 119 + ], + [ + 302, + 150 + ], + [ + 267, + 126 + ], + [ + 298, + 168 + ], + [ + 265, + 144 + ] + ] + ] + }, + { + "task_id": "0002789", + "id": "RH20T_cfg5_task_0096_user_0007_scene_0003_cfg_0005_036422060215#441", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002789/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 300, + 149 + ], + [ + 280, + 152 + ], + [ + 304, + 172 + ], + [ + 295, + 126 + ], + [ + 324, + 170 + ], + [ + 316, + 123 + ] + ], + [ + [ + 272, + 154 + ], + [ + 288, + 169 + ], + [ + 260, + 171 + ], + [ + 284, + 137 + ], + [ + 244, + 156 + ], + [ + 267, + 122 + ] + ], + [ + [ + 292, + 151 + ], + [ + 298, + 132 + ], + [ + 311, + 146 + ], + [ + 272, + 156 + ], + [ + 306, + 163 + ], + [ + 267, + 174 + ] + ], + [ + [ + 278, + 159 + ], + [ + 259, + 169 + ], + [ + 263, + 151 + ], + [ + 292, + 167 + ], + [ + 284, + 140 + ], + [ + 312, + 157 + ] + ] + ] + }, + { + "task_id": "0002792", + "id": "RH20T_cfg4_task_0066_user_0015_scene_0001_cfg_0004_038522062288#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002792/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 291, + 256 + ], + [ + 291, + 238 + ], + [ + 268, + 237 + ], + [ + 316, + 277 + ], + [ + 270, + 254 + ], + [ + 315, + 293 + ] + ], + [ + [ + 302, + 229 + ], + [ + 315, + 248 + ], + [ + 272, + 241 + ], + [ + 332, + 217 + ], + [ + 257, + 221 + ], + [ + 319, + 195 + ] + ], + [ + [ + 278, + 277 + ], + [ + 259, + 301 + ], + [ + 301, + 299 + ], + [ + 254, + 256 + ], + [ + 317, + 276 + ], + [ + 272, + 235 + ] + ], + [ + [ + 296, + 276 + ], + [ + 266, + 278 + ], + [ + 281, + 251 + ], + [ + 312, + 303 + ], + [ + 307, + 250 + ], + [ + 339, + 300 + ] + ] + ] + }, + { + "task_id": "0002794", + "id": "RH20T_cfg4_task_0092_user_0014_scene_0007_cfg_0004_104122062295#112", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the hands of a clock'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002794/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 245, + 192 + ], + [ + 247, + 158 + ], + [ + 206, + 177 + ], + [ + 284, + 208 + ], + [ + 206, + 207 + ], + [ + 280, + 238 + ] + ], + [ + [ + 201, + 203 + ], + [ + 226, + 225 + ], + [ + 182, + 181 + ], + [ + 222, + 228 + ], + [ + 154, + 156 + ], + [ + 192, + 203 + ] + ], + [ + [ + 221, + 194 + ], + [ + 238, + 196 + ], + [ + 206, + 238 + ], + [ + 235, + 150 + ], + [ + 184, + 241 + ], + [ + 216, + 143 + ] + ], + [ + [ + 281, + 216 + ], + [ + 284, + 214 + ], + [ + 278, + 259 + ], + [ + 283, + 174 + ], + [ + 276, + 257 + ], + [ + 280, + 179 + ] + ] + ] + }, + { + "task_id": "0002795", + "id": "RH20T_cfg4_task_0092_user_0014_scene_0007_cfg_0004_104122062295#136", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the hands of a clock'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002795/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 225, + 171 + ], + [ + 240, + 142 + ], + [ + 187, + 172 + ], + [ + 258, + 170 + ], + [ + 174, + 200 + ], + [ + 243, + 195 + ] + ], + [ + [ + 271, + 136 + ], + [ + 260, + 126 + ], + [ + 300, + 108 + ], + [ + 241, + 165 + ], + [ + 306, + 119 + ], + [ + 252, + 171 + ] + ], + [ + [ + 243, + 157 + ], + [ + 244, + 123 + ], + [ + 204, + 148 + ], + [ + 282, + 166 + ], + [ + 205, + 179 + ], + [ + 278, + 196 + ] + ], + [ + [ + 267, + 192 + ], + [ + 227, + 189 + ], + [ + 264, + 225 + ], + [ + 269, + 162 + ], + [ + 306, + 228 + ], + [ + 306, + 165 + ] + ] + ] + }, + { + "task_id": "0002797", + "id": "RH20T_cfg4_task_0026_user_0016_scene_0001_cfg_0004_038522062288#144", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push and close the drawer'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002797/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 276, + 201 + ], + [ + 272, + 175 + ], + [ + 262, + 177 + ], + [ + 291, + 228 + ], + [ + 266, + 201 + ], + [ + 293, + 250 + ] + ], + [ + [ + 271, + 199 + ], + [ + 238, + 197 + ], + [ + 280, + 203 + ], + [ + 264, + 196 + ], + [ + 314, + 205 + ], + [ + 294, + 198 + ] + ], + [ + [ + 247, + 197 + ], + [ + 232, + 231 + ], + [ + 262, + 196 + ], + [ + 234, + 197 + ], + [ + 278, + 160 + ], + [ + 248, + 166 + ] + ], + [ + [ + 304, + 217 + ], + [ + 280, + 215 + ], + [ + 279, + 205 + ], + [ + 331, + 229 + ], + [ + 301, + 207 + ], + [ + 352, + 230 + ] + ] + ] + }, + { + "task_id": "0002799", + "id": "RH20T_cfg2_task_0096_user_0007_scene_0005_cfg_0002_104422070044#184", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002799/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 289, + 188 + ], + [ + 272, + 183 + ], + [ + 284, + 189 + ], + [ + 293, + 186 + ], + [ + 300, + 194 + ], + [ + 310, + 191 + ] + ], + [ + [ + 263, + 183 + ], + [ + 276, + 177 + ], + [ + 272, + 181 + ], + [ + 252, + 185 + ], + [ + 260, + 187 + ], + [ + 239, + 191 + ] + ], + [ + [ + 276, + 200 + ], + [ + 261, + 193 + ], + [ + 284, + 184 + ], + [ + 268, + 215 + ], + [ + 298, + 191 + ], + [ + 283, + 220 + ] + ], + [ + [ + 280, + 184 + ], + [ + 283, + 174 + ], + [ + 277, + 198 + ], + [ + 283, + 171 + ], + [ + 274, + 207 + ], + [ + 281, + 181 + ] + ] + ] + }, + { + "task_id": "0002802", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_819112070044#403", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002802/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 312, + 203 + ], + [ + 290, + 195 + ], + [ + 312, + 195 + ], + [ + 311, + 212 + ], + [ + 333, + 203 + ], + [ + 334, + 220 + ] + ], + [ + [ + 291, + 211 + ], + [ + 292, + 194 + ], + [ + 311, + 203 + ], + [ + 269, + 219 + ], + [ + 309, + 218 + ], + [ + 268, + 234 + ] + ], + [ + [ + 316, + 224 + ], + [ + 321, + 245 + ], + [ + 334, + 215 + ], + [ + 299, + 232 + ], + [ + 329, + 194 + ], + [ + 295, + 212 + ] + ], + [ + [ + 279, + 212 + ], + [ + 299, + 202 + ], + [ + 289, + 222 + ], + [ + 268, + 202 + ], + [ + 269, + 232 + ], + [ + 247, + 212 + ] + ] + ] + }, + { + "task_id": "0002806", + "id": "RH20T_cfg5_task_0041_user_0007_scene_0006_cfg_0005_037522062165#248", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the tabletop with a sponge'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002806/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 273, + 312 + ], + [ + 298, + 320 + ], + [ + 268, + 340 + ], + [ + 279, + 284 + ], + [ + 245, + 332 + ], + [ + 256, + 277 + ] + ], + [ + [ + 290, + 321 + ], + [ + 270, + 328 + ], + [ + 270, + 336 + ], + [ + 310, + 306 + ], + [ + 289, + 329 + ], + [ + 327, + 300 + ] + ], + [ + [ + 300, + 300 + ], + [ + 300, + 290 + ], + [ + 279, + 280 + ], + [ + 322, + 320 + ], + [ + 280, + 289 + ], + [ + 321, + 327 + ] + ], + [ + [ + 312, + 335 + ], + [ + 296, + 322 + ], + [ + 294, + 324 + ], + [ + 329, + 345 + ], + [ + 311, + 338 + ], + [ + 347, + 358 + ] + ] + ] + }, + { + "task_id": "0002812", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0006_cfg_0001_750612070853#307", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002812/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 295, + 191 + ], + [ + 316, + 155 + ], + [ + 326, + 197 + ], + [ + 259, + 185 + ], + [ + 306, + 228 + ], + [ + 239, + 221 + ] + ], + [ + [ + 360, + 209 + ], + [ + 320, + 224 + ], + [ + 350, + 176 + ], + [ + 372, + 246 + ], + [ + 390, + 160 + ], + [ + 417, + 229 + ] + ], + [ + [ + 356, + 154 + ], + [ + 327, + 146 + ], + [ + 321, + 146 + ], + [ + 386, + 161 + ], + [ + 353, + 155 + ], + [ + 419, + 170 + ] + ], + [ + [ + 318, + 192 + ], + [ + 318, + 155 + ], + [ + 272, + 192 + ], + [ + 363, + 192 + ], + [ + 275, + 224 + ], + [ + 360, + 225 + ] + ] + ] + }, + { + "task_id": "0002813", + "id": "RH20T_cfg7_task_0015_user_0014_scene_0008_cfg_0007_104122063633#143", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002813/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 171, + 150 + ], + [ + 181, + 142 + ], + [ + 186, + 139 + ], + [ + 155, + 161 + ], + [ + 177, + 146 + ], + [ + 146, + 167 + ] + ], + [ + [ + 184, + 150 + ], + [ + 177, + 131 + ], + [ + 189, + 140 + ], + [ + 178, + 161 + ], + [ + 195, + 159 + ], + [ + 185, + 180 + ] + ], + [ + [ + 159, + 148 + ], + [ + 145, + 131 + ], + [ + 148, + 162 + ], + [ + 170, + 136 + ], + [ + 162, + 179 + ], + [ + 182, + 152 + ] + ], + [ + [ + 198, + 151 + ], + [ + 209, + 170 + ], + [ + 218, + 140 + ], + [ + 178, + 162 + ], + [ + 206, + 120 + ], + [ + 166, + 142 + ] + ] + ] + }, + { + "task_id": "0002814", + "id": "RH20T_cfg3_task_0069_user_0010_scene_0005_cfg_0003_104122062823#264", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Assemble one piece of a puzzle'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002814/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 136, + 102 + ], + [ + 130, + 75 + ], + [ + 164, + 110 + ], + [ + 109, + 95 + ], + [ + 168, + 135 + ], + [ + 116, + 120 + ] + ], + [ + [ + 92, + 107 + ], + [ + 68, + 125 + ], + [ + 64, + 85 + ], + [ + 119, + 128 + ], + [ + 87, + 69 + ], + [ + 139, + 111 + ] + ], + [ + [ + 120, + 87 + ], + [ + 95, + 99 + ], + [ + 146, + 109 + ], + [ + 94, + 65 + ], + [ + 167, + 98 + ], + [ + 118, + 57 + ] + ], + [ + [ + 143, + 109 + ], + [ + 134, + 99 + ], + [ + 168, + 92 + ], + [ + 117, + 126 + ], + [ + 178, + 101 + ], + [ + 125, + 138 + ] + ] + ] + }, + { + "task_id": "0002816", + "id": "RH20T_cfg3_task_0069_user_0010_scene_0005_cfg_0003_104122062823#525", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Assemble one piece of a puzzle'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002816/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 275, + 76 + ], + [ + 246, + 80 + ], + [ + 276, + 100 + ], + [ + 274, + 49 + ], + [ + 303, + 96 + ], + [ + 303, + 46 + ] + ], + [ + [ + 259, + 87 + ], + [ + 240, + 105 + ], + [ + 253, + 88 + ], + [ + 265, + 86 + ], + [ + 274, + 68 + ], + [ + 284, + 69 + ] + ], + [ + [ + 282, + 76 + ], + [ + 288, + 52 + ], + [ + 308, + 88 + ], + [ + 256, + 65 + ], + [ + 301, + 109 + ], + [ + 252, + 87 + ] + ], + [ + [ + 292, + 63 + ], + [ + 305, + 47 + ], + [ + 318, + 78 + ], + [ + 266, + 48 + ], + [ + 306, + 95 + ], + [ + 251, + 64 + ] + ] + ] + }, + { + "task_id": "0002817", + "id": "RH20T_cfg3_task_0069_user_0010_scene_0005_cfg_0003_104122062823#571", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Assemble one piece of a puzzle'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002817/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 272, + 72 + ], + [ + 265, + 51 + ], + [ + 299, + 73 + ], + [ + 244, + 70 + ], + [ + 307, + 95 + ], + [ + 250, + 93 + ] + ], + [ + [ + 297, + 65 + ], + [ + 288, + 61 + ], + [ + 309, + 37 + ], + [ + 286, + 93 + ], + [ + 320, + 40 + ], + [ + 296, + 100 + ] + ], + [ + [ + 293, + 90 + ], + [ + 298, + 64 + ], + [ + 322, + 98 + ], + [ + 265, + 82 + ], + [ + 316, + 122 + ], + [ + 262, + 106 + ] + ], + [ + [ + 314, + 103 + ], + [ + 321, + 91 + ], + [ + 287, + 94 + ], + [ + 340, + 113 + ], + [ + 278, + 106 + ], + [ + 333, + 126 + ] + ] + ] + }, + { + "task_id": "0002823", + "id": "RH20T_cfg3_task_0069_user_0010_scene_0005_cfg_0003_104122062823#771", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Assemble one piece of a puzzle'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002823/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 281, + 60 + ], + [ + 275, + 45 + ], + [ + 302, + 39 + ], + [ + 260, + 81 + ], + [ + 310, + 54 + ], + [ + 265, + 98 + ] + ], + [ + [ + 308, + 98 + ], + [ + 325, + 110 + ], + [ + 290, + 118 + ], + [ + 327, + 79 + ], + [ + 271, + 106 + ], + [ + 309, + 64 + ] + ], + [ + [ + 289, + 77 + ], + [ + 293, + 51 + ], + [ + 316, + 84 + ], + [ + 261, + 69 + ], + [ + 311, + 107 + ], + [ + 258, + 93 + ] + ], + [ + [ + 312, + 86 + ], + [ + 318, + 72 + ], + [ + 300, + 60 + ], + [ + 323, + 110 + ], + [ + 292, + 74 + ], + [ + 317, + 126 + ] + ] + ] + }, + { + "task_id": "0002824", + "id": "RH20T_cfg3_task_0069_user_0010_scene_0005_cfg_0003_104122062823#810", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Assemble one piece of a puzzle'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002824/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 297, + 63 + ], + [ + 305, + 68 + ], + [ + 288, + 89 + ], + [ + 306, + 36 + ], + [ + 279, + 85 + ], + [ + 298, + 29 + ] + ], + [ + [ + 280, + 71 + ], + [ + 285, + 46 + ], + [ + 306, + 80 + ], + [ + 253, + 62 + ], + [ + 300, + 103 + ], + [ + 248, + 85 + ] + ], + [ + [ + 274, + 81 + ], + [ + 283, + 69 + ], + [ + 300, + 84 + ], + [ + 248, + 78 + ], + [ + 291, + 97 + ], + [ + 237, + 91 + ] + ], + [ + [ + 257, + 58 + ], + [ + 253, + 37 + ], + [ + 284, + 51 + ], + [ + 230, + 65 + ], + [ + 289, + 73 + ], + [ + 233, + 87 + ] + ] + ] + }, + { + "task_id": "0002825", + "id": "RH20T_cfg3_task_0069_user_0010_scene_0005_cfg_0003_104122062823#828", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Assemble one piece of a puzzle'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002825/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 287, + 97 + ], + [ + 300, + 102 + ], + [ + 309, + 109 + ], + [ + 264, + 85 + ], + [ + 297, + 105 + ], + [ + 254, + 82 + ] + ], + [ + [ + 264, + 72 + ], + [ + 246, + 83 + ], + [ + 243, + 70 + ], + [ + 286, + 75 + ], + [ + 260, + 61 + ], + [ + 302, + 65 + ] + ], + [ + [ + 273, + 71 + ], + [ + 280, + 46 + ], + [ + 300, + 82 + ], + [ + 247, + 61 + ], + [ + 292, + 105 + ], + [ + 241, + 85 + ] + ], + [ + [ + 258, + 34 + ], + [ + 230, + 40 + ], + [ + 268, + 50 + ], + [ + 249, + 19 + ], + [ + 295, + 44 + ], + [ + 275, + 13 + ] + ] + ] + }, + { + "task_id": "0002826", + "id": "RH20T_cfg5_task_0001_user_0007_scene_0001_cfg_0005_037522062165#142", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002826/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 267, + 254 + ], + [ + 266, + 247 + ], + [ + 242, + 237 + ], + [ + 293, + 271 + ], + [ + 244, + 244 + ], + [ + 292, + 276 + ] + ], + [ + [ + 247, + 278 + ], + [ + 250, + 250 + ], + [ + 278, + 278 + ], + [ + 216, + 279 + ], + [ + 276, + 309 + ], + [ + 212, + 309 + ] + ], + [ + [ + 272, + 271 + ], + [ + 303, + 279 + ], + [ + 275, + 294 + ], + [ + 269, + 246 + ], + [ + 247, + 287 + ], + [ + 239, + 239 + ] + ], + [ + [ + 271, + 259 + ], + [ + 284, + 237 + ], + [ + 299, + 266 + ], + [ + 244, + 252 + ], + [ + 286, + 291 + ], + [ + 229, + 275 + ] + ] + ] + }, + { + "task_id": "0002827", + "id": "RH20T_cfg3_task_0007_user_0016_scene_0006_cfg_0003_104122062823#214", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002827/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 329, + 90 + ], + [ + 328, + 66 + ], + [ + 357, + 95 + ], + [ + 301, + 84 + ], + [ + 356, + 117 + ], + [ + 303, + 106 + ] + ], + [ + [ + 341, + 74 + ], + [ + 322, + 97 + ], + [ + 329, + 76 + ], + [ + 354, + 71 + ], + [ + 347, + 54 + ], + [ + 375, + 47 + ] + ], + [ + [ + 323, + 78 + ], + [ + 316, + 101 + ], + [ + 298, + 86 + ], + [ + 350, + 69 + ], + [ + 305, + 65 + ], + [ + 356, + 47 + ] + ], + [ + [ + 330, + 119 + ], + [ + 320, + 132 + ], + [ + 352, + 109 + ], + [ + 306, + 129 + ], + [ + 364, + 94 + ], + [ + 316, + 114 + ] + ] + ] + }, + { + "task_id": "0002828", + "id": "RH20T_cfg3_task_0007_user_0016_scene_0006_cfg_0003_104122062823#260", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002828/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 294, + 62 + ], + [ + 301, + 80 + ], + [ + 268, + 53 + ], + [ + 319, + 69 + ], + [ + 262, + 36 + ], + [ + 312, + 53 + ] + ], + [ + [ + 291, + 81 + ], + [ + 276, + 94 + ], + [ + 297, + 103 + ], + [ + 285, + 56 + ], + [ + 310, + 91 + ], + [ + 300, + 44 + ] + ], + [ + [ + 241, + 61 + ], + [ + 233, + 42 + ], + [ + 264, + 43 + ], + [ + 218, + 79 + ], + [ + 273, + 64 + ], + [ + 225, + 100 + ] + ], + [ + [ + 275, + 76 + ], + [ + 273, + 53 + ], + [ + 300, + 85 + ], + [ + 250, + 67 + ], + [ + 301, + 105 + ], + [ + 253, + 88 + ] + ] + ] + }, + { + "task_id": "0002829", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_039422060546#171", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002829/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 190, + 360 + ], + [ + 173, + 360 + ], + [ + 216, + 360 + ], + [ + 160, + 360 + ], + [ + 233, + 344 + ], + [ + 179, + 353 + ] + ], + [ + [ + 183, + 306 + ], + [ + 214, + 315 + ], + [ + 161, + 336 + ], + [ + 203, + 278 + ], + [ + 131, + 326 + ], + [ + 175, + 270 + ] + ], + [ + [ + 214, + 337 + ], + [ + 230, + 316 + ], + [ + 208, + 360 + ], + [ + 219, + 301 + ], + [ + 190, + 360 + ], + [ + 201, + 322 + ] + ], + [ + [ + 169, + 339 + ], + [ + 153, + 323 + ], + [ + 184, + 360 + ], + [ + 155, + 309 + ], + [ + 198, + 360 + ], + [ + 169, + 323 + ] + ] + ] + }, + { + "task_id": "0002832", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_750612070851#88", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002832/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 129, + 292 + ], + [ + 100, + 315 + ], + [ + 96, + 270 + ], + [ + 163, + 314 + ], + [ + 123, + 253 + ], + [ + 185, + 293 + ] + ], + [ + [ + 154, + 251 + ], + [ + 185, + 275 + ], + [ + 125, + 275 + ], + [ + 180, + 230 + ], + [ + 94, + 250 + ], + [ + 151, + 209 + ] + ], + [ + [ + 129, + 294 + ], + [ + 120, + 274 + ], + [ + 84, + 308 + ], + [ + 172, + 280 + ], + [ + 96, + 325 + ], + [ + 177, + 298 + ] + ], + [ + [ + 124, + 300 + ], + [ + 110, + 330 + ], + [ + 141, + 276 + ], + [ + 108, + 323 + ], + [ + 152, + 250 + ], + [ + 121, + 296 + ] + ] + ] + }, + { + "task_id": "0002834", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_750612070851#301", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002834/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 255, + 337 + ], + [ + 267, + 302 + ], + [ + 217, + 325 + ], + [ + 292, + 349 + ], + [ + 201, + 360 + ], + [ + 280, + 360 + ] + ], + [ + [ + 221, + 308 + ], + [ + 218, + 290 + ], + [ + 180, + 307 + ], + [ + 262, + 309 + ], + [ + 185, + 323 + ], + [ + 262, + 324 + ] + ], + [ + [ + 216, + 302 + ], + [ + 234, + 277 + ], + [ + 251, + 303 + ], + [ + 182, + 301 + ], + [ + 232, + 333 + ], + [ + 159, + 330 + ] + ], + [ + [ + 237, + 347 + ], + [ + 242, + 308 + ], + [ + 239, + 357 + ], + [ + 235, + 334 + ], + [ + 233, + 360 + ], + [ + 229, + 360 + ] + ] + ] + }, + { + "task_id": "0002835", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_750612070851#414", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002835/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 248, + 300 + ], + [ + 246, + 291 + ], + [ + 281, + 315 + ], + [ + 215, + 284 + ], + [ + 286, + 326 + ], + [ + 214, + 293 + ] + ], + [ + [ + 262, + 307 + ], + [ + 262, + 289 + ], + [ + 221, + 307 + ], + [ + 302, + 308 + ], + [ + 224, + 323 + ], + [ + 299, + 324 + ] + ], + [ + [ + 264, + 285 + ], + [ + 226, + 281 + ], + [ + 258, + 253 + ], + [ + 270, + 314 + ], + [ + 300, + 257 + ], + [ + 309, + 319 + ] + ], + [ + [ + 227, + 302 + ], + [ + 227, + 264 + ], + [ + 228, + 292 + ], + [ + 224, + 314 + ], + [ + 228, + 327 + ], + [ + 224, + 356 + ] + ] + ] + }, + { + "task_id": "0002836", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_750612070851#549", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002836/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 264 + ], + [ + 283, + 286 + ], + [ + 312, + 232 + ], + [ + 307, + 292 + ], + [ + 336, + 214 + ], + [ + 329, + 272 + ] + ], + [ + [ + 327, + 324 + ], + [ + 347, + 338 + ], + [ + 300, + 358 + ], + [ + 353, + 291 + ], + [ + 285, + 343 + ], + [ + 335, + 282 + ] + ], + [ + [ + 282, + 270 + ], + [ + 264, + 302 + ], + [ + 313, + 270 + ], + [ + 254, + 271 + ], + [ + 329, + 240 + ], + [ + 271, + 244 + ] + ], + [ + [ + 303, + 313 + ], + [ + 305, + 293 + ], + [ + 262, + 314 + ], + [ + 342, + 312 + ], + [ + 262, + 331 + ], + [ + 338, + 329 + ] + ] + ] + }, + { + "task_id": "0002837", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_036422060215#263", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002837/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 275 + ], + [ + 337, + 291 + ], + [ + 347, + 263 + ], + [ + 330, + 287 + ], + [ + 348, + 249 + ], + [ + 332, + 272 + ] + ], + [ + [ + 338, + 255 + ], + [ + 346, + 242 + ], + [ + 355, + 255 + ], + [ + 320, + 254 + ], + [ + 347, + 267 + ], + [ + 314, + 266 + ] + ], + [ + [ + 318, + 240 + ], + [ + 334, + 250 + ], + [ + 326, + 225 + ], + [ + 311, + 256 + ], + [ + 311, + 215 + ], + [ + 295, + 246 + ] + ], + [ + [ + 323, + 230 + ], + [ + 311, + 238 + ], + [ + 319, + 213 + ], + [ + 327, + 248 + ], + [ + 331, + 204 + ], + [ + 339, + 240 + ] + ] + ] + }, + { + "task_id": "0002838", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_036422060215#410", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002838/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 303, + 236 + ], + [ + 320, + 244 + ], + [ + 311, + 240 + ], + [ + 295, + 233 + ], + [ + 295, + 232 + ], + [ + 278, + 225 + ] + ], + [ + [ + 328, + 254 + ], + [ + 346, + 253 + ], + [ + 317, + 252 + ], + [ + 339, + 255 + ], + [ + 299, + 252 + ], + [ + 322, + 255 + ] + ], + [ + [ + 317, + 250 + ], + [ + 323, + 237 + ], + [ + 334, + 248 + ], + [ + 299, + 251 + ], + [ + 327, + 261 + ], + [ + 293, + 263 + ] + ], + [ + [ + 309, + 253 + ], + [ + 313, + 269 + ], + [ + 304, + 267 + ], + [ + 314, + 239 + ], + [ + 301, + 252 + ], + [ + 310, + 225 + ] + ] + ] + }, + { + "task_id": "0002839", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_036422060215#619", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002839/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 268 + ], + [ + 276, + 287 + ], + [ + 282, + 256 + ], + [ + 286, + 280 + ], + [ + 290, + 238 + ], + [ + 293, + 263 + ] + ], + [ + [ + 282, + 287 + ], + [ + 298, + 277 + ], + [ + 281, + 271 + ], + [ + 282, + 301 + ], + [ + 263, + 282 + ], + [ + 265, + 312 + ] + ], + [ + [ + 286, + 269 + ], + [ + 292, + 258 + ], + [ + 288, + 287 + ], + [ + 283, + 251 + ], + [ + 282, + 297 + ], + [ + 277, + 262 + ] + ], + [ + [ + 286, + 259 + ], + [ + 289, + 245 + ], + [ + 299, + 273 + ], + [ + 273, + 246 + ], + [ + 296, + 286 + ], + [ + 271, + 260 + ] + ] + ] + }, + { + "task_id": "0002841", + "id": "RH20T_cfg6_task_0089_user_0020_scene_0009_cfg_0006_104122061330#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push an object with obstacles'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002841/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 332, + 95 + ], + [ + 308, + 91 + ], + [ + 338, + 104 + ], + [ + 325, + 85 + ], + [ + 361, + 107 + ], + [ + 351, + 89 + ] + ], + [ + [ + 338, + 93 + ], + [ + 361, + 88 + ], + [ + 337, + 116 + ], + [ + 338, + 70 + ], + [ + 314, + 120 + ], + [ + 316, + 76 + ] + ], + [ + [ + 297, + 104 + ], + [ + 298, + 119 + ], + [ + 292, + 85 + ], + [ + 303, + 124 + ], + [ + 290, + 67 + ], + [ + 302, + 107 + ] + ], + [ + [ + 316, + 104 + ], + [ + 314, + 84 + ], + [ + 305, + 119 + ], + [ + 327, + 91 + ], + [ + 307, + 138 + ], + [ + 328, + 110 + ] + ] + ] + }, + { + "task_id": "0002843", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0005_cfg_0005_104122061850#193", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002843/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 186 + ], + [ + 294, + 178 + ], + [ + 329, + 183 + ], + [ + 307, + 189 + ], + [ + 354, + 190 + ], + [ + 334, + 197 + ] + ], + [ + [ + 318, + 170 + ], + [ + 316, + 156 + ], + [ + 305, + 193 + ], + [ + 331, + 148 + ], + [ + 308, + 205 + ], + [ + 333, + 162 + ] + ], + [ + [ + 336, + 170 + ], + [ + 356, + 190 + ], + [ + 345, + 171 + ], + [ + 326, + 168 + ], + [ + 326, + 153 + ], + [ + 306, + 147 + ] + ], + [ + [ + 327, + 183 + ], + [ + 343, + 184 + ], + [ + 349, + 178 + ], + [ + 303, + 188 + ], + [ + 334, + 177 + ], + [ + 289, + 187 + ] + ] + ] + }, + { + "task_id": "0002844", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0005_cfg_0005_104122061850#233", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002844/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 329, + 124 + ], + [ + 326, + 106 + ], + [ + 315, + 145 + ], + [ + 343, + 103 + ], + [ + 318, + 160 + ], + [ + 344, + 120 + ] + ], + [ + [ + 342, + 120 + ], + [ + 313, + 122 + ], + [ + 341, + 91 + ], + [ + 342, + 147 + ], + [ + 370, + 90 + ], + [ + 369, + 145 + ] + ], + [ + [ + 301, + 130 + ], + [ + 325, + 141 + ], + [ + 312, + 103 + ], + [ + 291, + 156 + ], + [ + 288, + 93 + ], + [ + 269, + 145 + ] + ], + [ + [ + 340, + 148 + ], + [ + 321, + 170 + ], + [ + 352, + 162 + ], + [ + 328, + 135 + ], + [ + 373, + 138 + ], + [ + 346, + 114 + ] + ] + ] + }, + { + "task_id": "0002846", + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_104422070044#58", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002846/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 360, + 225 + ], + [ + 376, + 218 + ], + [ + 362, + 223 + ], + [ + 357, + 227 + ], + [ + 345, + 230 + ], + [ + 341, + 234 + ] + ], + [ + [ + 331, + 206 + ], + [ + 340, + 194 + ], + [ + 316, + 196 + ], + [ + 346, + 216 + ], + [ + 307, + 208 + ], + [ + 337, + 229 + ] + ], + [ + [ + 335, + 211 + ], + [ + 327, + 213 + ], + [ + 345, + 197 + ], + [ + 326, + 225 + ], + [ + 354, + 194 + ], + [ + 334, + 223 + ] + ], + [ + [ + 343, + 223 + ], + [ + 341, + 210 + ], + [ + 360, + 228 + ], + [ + 327, + 218 + ], + [ + 362, + 240 + ], + [ + 330, + 230 + ] + ] + ] + }, + { + "task_id": "0002847", + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_104422070044#195", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002847/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 374, + 228 + ], + [ + 373, + 217 + ], + [ + 381, + 242 + ], + [ + 366, + 214 + ], + [ + 381, + 253 + ], + [ + 367, + 225 + ] + ], + [ + [ + 371, + 222 + ], + [ + 355, + 216 + ], + [ + 378, + 205 + ], + [ + 363, + 239 + ], + [ + 394, + 211 + ], + [ + 380, + 245 + ] + ], + [ + [ + 358, + 234 + ], + [ + 343, + 234 + ], + [ + 367, + 242 + ], + [ + 349, + 226 + ], + [ + 383, + 242 + ], + [ + 365, + 226 + ] + ], + [ + [ + 393, + 235 + ], + [ + 386, + 253 + ], + [ + 409, + 243 + ], + [ + 378, + 228 + ], + [ + 416, + 226 + ], + [ + 385, + 210 + ] + ] + ] + }, + { + "task_id": "0002849", + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_104422070044#470", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002849/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 411, + 193 + ], + [ + 414, + 175 + ], + [ + 393, + 190 + ], + [ + 429, + 196 + ], + [ + 390, + 208 + ], + [ + 426, + 213 + ] + ], + [ + [ + 406, + 205 + ], + [ + 406, + 193 + ], + [ + 401, + 219 + ], + [ + 411, + 193 + ], + [ + 401, + 230 + ], + [ + 411, + 205 + ] + ], + [ + [ + 397, + 220 + ], + [ + 402, + 206 + ], + [ + 410, + 232 + ], + [ + 386, + 209 + ], + [ + 405, + 245 + ], + [ + 382, + 222 + ] + ], + [ + [ + 411, + 212 + ], + [ + 394, + 204 + ], + [ + 403, + 227 + ], + [ + 419, + 196 + ], + [ + 420, + 235 + ], + [ + 436, + 204 + ] + ] + ] + }, + { + "task_id": "0002853", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122061018#298", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002853/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 300, + 74 + ], + [ + 323, + 67 + ], + [ + 288, + 86 + ], + [ + 313, + 60 + ], + [ + 265, + 93 + ], + [ + 288, + 67 + ] + ], + [ + [ + 294, + 101 + ], + [ + 313, + 116 + ], + [ + 291, + 121 + ], + [ + 298, + 79 + ], + [ + 273, + 107 + ], + [ + 279, + 64 + ] + ], + [ + [ + 266, + 81 + ], + [ + 267, + 57 + ], + [ + 255, + 65 + ], + [ + 277, + 98 + ], + [ + 255, + 87 + ], + [ + 276, + 120 + ] + ], + [ + [ + 287, + 80 + ], + [ + 268, + 72 + ], + [ + 276, + 106 + ], + [ + 299, + 53 + ], + [ + 294, + 112 + ], + [ + 316, + 61 + ] + ] + ] + }, + { + "task_id": "0002854", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122061018#744", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002854/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 360, + 153 + ], + [ + 362, + 126 + ], + [ + 345, + 139 + ], + [ + 378, + 168 + ], + [ + 344, + 164 + ], + [ + 375, + 194 + ] + ], + [ + [ + 358, + 189 + ], + [ + 337, + 211 + ], + [ + 342, + 183 + ], + [ + 375, + 196 + ], + [ + 361, + 162 + ], + [ + 396, + 173 + ] + ], + [ + [ + 381, + 165 + ], + [ + 356, + 180 + ], + [ + 386, + 187 + ], + [ + 374, + 141 + ], + [ + 409, + 173 + ], + [ + 399, + 127 + ] + ], + [ + [ + 345, + 141 + ], + [ + 343, + 132 + ], + [ + 332, + 170 + ], + [ + 358, + 113 + ], + [ + 333, + 177 + ], + [ + 358, + 123 + ] + ] + ] + }, + { + "task_id": "0002856", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0008_cfg_0006_104122061018#1528", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002856/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 367, + 223 + ], + [ + 332, + 213 + ], + [ + 355, + 244 + ], + [ + 381, + 199 + ], + [ + 388, + 253 + ], + [ + 418, + 210 + ] + ], + [ + [ + 356, + 235 + ], + [ + 332, + 227 + ], + [ + 379, + 222 + ], + [ + 335, + 247 + ], + [ + 401, + 231 + ], + [ + 358, + 254 + ] + ], + [ + [ + 368, + 238 + ], + [ + 373, + 213 + ], + [ + 349, + 215 + ], + [ + 388, + 263 + ], + [ + 346, + 237 + ], + [ + 383, + 285 + ] + ], + [ + [ + 363, + 221 + ], + [ + 387, + 200 + ], + [ + 357, + 198 + ], + [ + 368, + 241 + ], + [ + 330, + 222 + ], + [ + 344, + 263 + ] + ] + ] + }, + { + "task_id": "0002857", + "id": "RH20T_cfg7_task_0044_user_0014_scene_0005_cfg_0007_037522061512#86", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002857/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 269, + 84 + ], + [ + 265, + 107 + ], + [ + 274, + 76 + ], + [ + 263, + 91 + ], + [ + 278, + 53 + ], + [ + 267, + 70 + ] + ], + [ + [ + 288, + 114 + ], + [ + 309, + 109 + ], + [ + 286, + 93 + ], + [ + 291, + 137 + ], + [ + 267, + 98 + ], + [ + 271, + 141 + ] + ], + [ + [ + 291, + 67 + ], + [ + 268, + 71 + ], + [ + 291, + 46 + ], + [ + 291, + 86 + ], + [ + 314, + 42 + ], + [ + 313, + 82 + ] + ], + [ + [ + 274, + 86 + ], + [ + 268, + 71 + ], + [ + 283, + 68 + ], + [ + 265, + 105 + ], + [ + 289, + 83 + ], + [ + 271, + 118 + ] + ] + ] + }, + { + "task_id": "0002858", + "id": "RH20T_cfg6_task_0007_user_0014_scene_0007_cfg_0006_104122064161#99", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002858/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 329, + 254 + ], + [ + 354, + 250 + ], + [ + 327, + 263 + ], + [ + 332, + 245 + ], + [ + 302, + 268 + ], + [ + 305, + 250 + ] + ], + [ + [ + 340, + 264 + ], + [ + 321, + 260 + ], + [ + 343, + 292 + ], + [ + 336, + 235 + ], + [ + 364, + 297 + ], + [ + 357, + 237 + ] + ], + [ + [ + 355, + 260 + ], + [ + 357, + 252 + ], + [ + 355, + 288 + ], + [ + 355, + 234 + ], + [ + 353, + 295 + ], + [ + 354, + 243 + ] + ], + [ + [ + 331, + 243 + ], + [ + 342, + 258 + ], + [ + 331, + 220 + ], + [ + 332, + 264 + ], + [ + 320, + 207 + ], + [ + 322, + 250 + ] + ] + ] + }, + { + "task_id": "0002863", + "id": "RH20T_cfg3_task_0007_user_0016_scene_0006_cfg_0003_104122062295#257", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002863/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 124, + 264 + ], + [ + 125, + 288 + ], + [ + 134, + 233 + ], + [ + 113, + 299 + ], + [ + 133, + 207 + ], + [ + 110, + 273 + ] + ], + [ + [ + 170, + 233 + ], + [ + 192, + 250 + ], + [ + 147, + 200 + ], + [ + 190, + 262 + ], + [ + 127, + 186 + ], + [ + 170, + 246 + ] + ], + [ + [ + 124, + 222 + ], + [ + 88, + 200 + ], + [ + 116, + 254 + ], + [ + 132, + 194 + ], + [ + 151, + 273 + ], + [ + 163, + 214 + ] + ], + [ + [ + 103, + 253 + ], + [ + 107, + 218 + ], + [ + 76, + 227 + ], + [ + 133, + 280 + ], + [ + 74, + 259 + ], + [ + 128, + 310 + ] + ] + ] + }, + { + "task_id": "0002865", + "id": "RH20T_cfg6_task_0125_user_0020_scene_0004_cfg_0006_104122061602#555", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull up the lever'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002865/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 506, + 76 + ], + [ + 501, + 99 + ], + [ + 482, + 74 + ], + [ + 530, + 78 + ], + [ + 486, + 51 + ], + [ + 535, + 54 + ] + ], + [ + [ + 479, + 80 + ], + [ + 483, + 64 + ], + [ + 496, + 93 + ], + [ + 463, + 68 + ], + [ + 492, + 107 + ], + [ + 461, + 83 + ] + ], + [ + [ + 480, + 54 + ], + [ + 482, + 63 + ], + [ + 457, + 59 + ], + [ + 504, + 50 + ], + [ + 456, + 51 + ], + [ + 501, + 42 + ] + ], + [ + [ + 508, + 72 + ], + [ + 494, + 77 + ], + [ + 530, + 74 + ], + [ + 488, + 70 + ], + [ + 544, + 69 + ], + [ + 501, + 66 + ] + ] + ] + }, + { + "task_id": "0002871", + "id": "RH20T_cfg2_task_0011_user_0013_scene_0001_cfg_0002_036422060215#295", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002871/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 190, + 191 + ], + [ + 213, + 191 + ], + [ + 199, + 206 + ], + [ + 182, + 176 + ], + [ + 173, + 206 + ], + [ + 157, + 175 + ] + ], + [ + [ + 193, + 155 + ], + [ + 173, + 141 + ], + [ + 210, + 135 + ], + [ + 176, + 176 + ], + [ + 230, + 149 + ], + [ + 197, + 191 + ] + ], + [ + [ + 168, + 171 + ], + [ + 167, + 151 + ], + [ + 157, + 189 + ], + [ + 177, + 154 + ], + [ + 158, + 207 + ], + [ + 177, + 172 + ] + ], + [ + [ + 169, + 164 + ], + [ + 174, + 178 + ], + [ + 147, + 157 + ], + [ + 192, + 172 + ], + [ + 141, + 141 + ], + [ + 188, + 158 + ] + ] + ] + }, + { + "task_id": "0002878", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122064161#437", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002878/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 332, + 165 + ], + [ + 332, + 168 + ], + [ + 329, + 186 + ], + [ + 334, + 143 + ], + [ + 329, + 183 + ], + [ + 334, + 141 + ] + ], + [ + [ + 351, + 168 + ], + [ + 371, + 160 + ], + [ + 362, + 187 + ], + [ + 340, + 150 + ], + [ + 344, + 194 + ], + [ + 322, + 158 + ] + ], + [ + [ + 338, + 155 + ], + [ + 335, + 132 + ], + [ + 357, + 151 + ], + [ + 317, + 159 + ], + [ + 360, + 174 + ], + [ + 320, + 182 + ] + ], + [ + [ + 348, + 163 + ], + [ + 361, + 163 + ], + [ + 332, + 160 + ], + [ + 365, + 167 + ], + [ + 318, + 160 + ], + [ + 351, + 167 + ] + ] + ] + }, + { + "task_id": "0002879", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122064161#478", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002879/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 318, + 183 + ], + [ + 338, + 178 + ], + [ + 320, + 201 + ], + [ + 315, + 165 + ], + [ + 300, + 206 + ], + [ + 293, + 170 + ] + ], + [ + [ + 315, + 161 + ], + [ + 314, + 158 + ], + [ + 317, + 183 + ], + [ + 313, + 140 + ], + [ + 318, + 186 + ], + [ + 315, + 144 + ] + ], + [ + [ + 335, + 150 + ], + [ + 314, + 157 + ], + [ + 341, + 165 + ], + [ + 329, + 133 + ], + [ + 361, + 158 + ], + [ + 352, + 125 + ] + ], + [ + [ + 311, + 153 + ], + [ + 317, + 145 + ], + [ + 331, + 164 + ], + [ + 292, + 141 + ], + [ + 325, + 170 + ], + [ + 288, + 148 + ] + ] + ] + }, + { + "task_id": "0002881", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122064161#676", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002881/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 374, + 137 + ], + [ + 379, + 129 + ], + [ + 378, + 158 + ], + [ + 369, + 117 + ], + [ + 373, + 165 + ], + [ + 365, + 126 + ] + ], + [ + [ + 381, + 129 + ], + [ + 389, + 131 + ], + [ + 360, + 139 + ], + [ + 401, + 120 + ], + [ + 354, + 136 + ], + [ + 392, + 119 + ] + ], + [ + [ + 388, + 126 + ], + [ + 376, + 110 + ], + [ + 401, + 110 + ], + [ + 375, + 142 + ], + [ + 412, + 125 + ], + [ + 387, + 157 + ] + ], + [ + [ + 371, + 140 + ], + [ + 392, + 139 + ], + [ + 366, + 133 + ], + [ + 376, + 148 + ], + [ + 346, + 134 + ], + [ + 354, + 149 + ] + ] + ] + }, + { + "task_id": "0002885", + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_038522063145#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002885/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 568, + 204 + ], + [ + 554, + 217 + ], + [ + 548, + 224 + ], + [ + 591, + 181 + ], + [ + 562, + 212 + ], + [ + 604, + 169 + ] + ], + [ + [ + 526, + 209 + ], + [ + 549, + 203 + ], + [ + 531, + 182 + ], + [ + 522, + 238 + ], + [ + 510, + 189 + ], + [ + 501, + 242 + ] + ], + [ + [ + 528, + 202 + ], + [ + 559, + 202 + ], + [ + 530, + 171 + ], + [ + 526, + 232 + ], + [ + 499, + 171 + ], + [ + 496, + 232 + ] + ], + [ + [ + 524, + 197 + ], + [ + 538, + 174 + ], + [ + 555, + 200 + ], + [ + 493, + 194 + ], + [ + 540, + 221 + ], + [ + 481, + 215 + ] + ] + ] + }, + { + "task_id": "0002887", + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_038522063145#386", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002887/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 587, + 177 + ], + [ + 566, + 172 + ], + [ + 613, + 178 + ], + [ + 561, + 176 + ], + [ + 638, + 184 + ], + [ + 584, + 182 + ] + ], + [ + [ + 562, + 183 + ], + [ + 567, + 199 + ], + [ + 544, + 205 + ], + [ + 581, + 159 + ], + [ + 541, + 191 + ], + [ + 576, + 145 + ] + ], + [ + [ + 568, + 169 + ], + [ + 583, + 146 + ], + [ + 597, + 171 + ], + [ + 539, + 167 + ], + [ + 582, + 191 + ], + [ + 526, + 188 + ] + ], + [ + [ + 563, + 138 + ], + [ + 562, + 168 + ], + [ + 536, + 135 + ], + [ + 593, + 142 + ], + [ + 537, + 106 + ], + [ + 596, + 109 + ] + ] + ] + }, + { + "task_id": "0002889", + "id": "RH20T_cfg5_task_0123_user_0007_scene_0007_cfg_0005_104122063678#255", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the knife on the knife rack'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002889/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 268, + 153 + ], + [ + 255, + 139 + ], + [ + 289, + 150 + ], + [ + 249, + 155 + ], + [ + 301, + 164 + ], + [ + 263, + 168 + ] + ], + [ + [ + 242, + 147 + ], + [ + 258, + 155 + ], + [ + 224, + 158 + ], + [ + 260, + 136 + ], + [ + 206, + 150 + ], + [ + 243, + 127 + ] + ], + [ + [ + 242, + 195 + ], + [ + 236, + 209 + ], + [ + 264, + 204 + ], + [ + 220, + 186 + ], + [ + 268, + 192 + ], + [ + 226, + 174 + ] + ], + [ + [ + 257, + 178 + ], + [ + 263, + 162 + ], + [ + 256, + 197 + ], + [ + 258, + 161 + ], + [ + 251, + 212 + ], + [ + 253, + 176 + ] + ] + ] + }, + { + "task_id": "0002891", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0005_cfg_0001_819112070044#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002891/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 401, + 170 + ], + [ + 423, + 168 + ], + [ + 398, + 151 + ], + [ + 404, + 188 + ], + [ + 377, + 153 + ], + [ + 384, + 189 + ] + ], + [ + [ + 412, + 151 + ], + [ + 393, + 150 + ], + [ + 408, + 172 + ], + [ + 416, + 131 + ], + [ + 428, + 174 + ], + [ + 436, + 132 + ] + ], + [ + [ + 415, + 167 + ], + [ + 420, + 151 + ], + [ + 430, + 160 + ], + [ + 400, + 175 + ], + [ + 425, + 176 + ], + [ + 396, + 190 + ] + ], + [ + [ + 417, + 145 + ], + [ + 415, + 124 + ], + [ + 402, + 149 + ], + [ + 431, + 141 + ], + [ + 404, + 170 + ], + [ + 433, + 162 + ] + ] + ] + }, + { + "task_id": "0002892", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0005_cfg_0001_819112070044#491", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002892/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 383, + 154 + ], + [ + 364, + 154 + ], + [ + 378, + 172 + ], + [ + 388, + 137 + ], + [ + 399, + 172 + ], + [ + 408, + 136 + ] + ], + [ + [ + 398, + 173 + ], + [ + 402, + 158 + ], + [ + 415, + 167 + ], + [ + 380, + 178 + ], + [ + 411, + 181 + ], + [ + 377, + 193 + ] + ], + [ + [ + 405, + 181 + ], + [ + 392, + 168 + ], + [ + 398, + 195 + ], + [ + 411, + 166 + ], + [ + 411, + 208 + ], + [ + 425, + 180 + ] + ], + [ + [ + 401, + 169 + ], + [ + 415, + 181 + ], + [ + 410, + 152 + ], + [ + 392, + 185 + ], + [ + 396, + 141 + ], + [ + 380, + 174 + ] + ] + ] + }, + { + "task_id": "0002893", + "id": "RH20T_cfg4_task_0013_user_0016_scene_0007_cfg_0004_104122062823#144", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002893/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 281, + 46 + ], + [ + 305, + 65 + ], + [ + 297, + 25 + ], + [ + 267, + 66 + ], + [ + 272, + 5 + ], + [ + 244, + 48 + ] + ], + [ + [ + 245, + 36 + ], + [ + 269, + 42 + ], + [ + 238, + 17 + ], + [ + 253, + 57 + ], + [ + 212, + 10 + ], + [ + 226, + 51 + ] + ], + [ + [ + 270, + 61 + ], + [ + 297, + 60 + ], + [ + 270, + 82 + ], + [ + 270, + 37 + ], + [ + 243, + 83 + ], + [ + 241, + 38 + ] + ], + [ + [ + 260, + 43 + ], + [ + 260, + 16 + ], + [ + 282, + 53 + ], + [ + 240, + 34 + ], + [ + 281, + 79 + ], + [ + 242, + 59 + ] + ] + ] + }, + { + "task_id": "0002895", + "id": "RH20T_cfg1_task_0015_user_0008_scene_0008_cfg_0001_038522062547#141", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002895/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 449, + 208 + ], + [ + 420, + 207 + ], + [ + 452, + 205 + ], + [ + 447, + 212 + ], + [ + 481, + 206 + ], + [ + 480, + 213 + ] + ], + [ + [ + 458, + 253 + ], + [ + 451, + 234 + ], + [ + 439, + 240 + ], + [ + 476, + 266 + ], + [ + 446, + 262 + ], + [ + 484, + 288 + ] + ], + [ + [ + 490, + 261 + ], + [ + 512, + 286 + ], + [ + 466, + 280 + ], + [ + 514, + 241 + ], + [ + 445, + 256 + ], + [ + 492, + 217 + ] + ], + [ + [ + 471, + 236 + ], + [ + 477, + 226 + ], + [ + 501, + 245 + ], + [ + 443, + 226 + ], + [ + 494, + 254 + ], + [ + 439, + 236 + ] + ] + ] + }, + { + "task_id": "0002898", + "id": "RH20T_cfg4_task_0125_user_0007_scene_0004_cfg_0004_036422060909#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull up the lever'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002898/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 143 + ], + [ + 295, + 128 + ], + [ + 299, + 127 + ], + [ + 317, + 158 + ], + [ + 313, + 143 + ], + [ + 331, + 174 + ] + ], + [ + [ + 293, + 117 + ], + [ + 294, + 97 + ], + [ + 308, + 101 + ], + [ + 277, + 133 + ], + [ + 306, + 120 + ], + [ + 276, + 151 + ] + ], + [ + [ + 282, + 117 + ], + [ + 265, + 115 + ], + [ + 260, + 114 + ], + [ + 303, + 120 + ], + [ + 279, + 116 + ], + [ + 322, + 122 + ] + ], + [ + [ + 289, + 150 + ], + [ + 311, + 166 + ], + [ + 279, + 153 + ], + [ + 298, + 147 + ], + [ + 257, + 136 + ], + [ + 278, + 132 + ] + ] + ] + }, + { + "task_id": "0002901", + "id": "RH20T_cfg6_task_0077_user_0014_scene_0004_cfg_0006_037522061512#193", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002901/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 226, + 149 + ], + [ + 238, + 166 + ], + [ + 212, + 161 + ], + [ + 239, + 138 + ], + [ + 199, + 144 + ], + [ + 226, + 121 + ] + ], + [ + [ + 218, + 132 + ], + [ + 202, + 138 + ], + [ + 206, + 136 + ], + [ + 231, + 128 + ], + [ + 221, + 130 + ], + [ + 246, + 122 + ] + ], + [ + [ + 214, + 142 + ], + [ + 205, + 129 + ], + [ + 224, + 125 + ], + [ + 202, + 159 + ], + [ + 232, + 137 + ], + [ + 211, + 171 + ] + ], + [ + [ + 211, + 143 + ], + [ + 200, + 131 + ], + [ + 227, + 143 + ], + [ + 194, + 143 + ], + [ + 238, + 156 + ], + [ + 205, + 157 + ] + ] + ] + }, + { + "task_id": "0002904", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0006_cfg_0003_104122062823#126", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002904/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 141, + 63 + ], + [ + 133, + 36 + ], + [ + 162, + 50 + ], + [ + 118, + 76 + ], + [ + 168, + 74 + ], + [ + 126, + 100 + ] + ], + [ + [ + 175, + 68 + ], + [ + 149, + 52 + ], + [ + 170, + 71 + ], + [ + 182, + 65 + ], + [ + 193, + 85 + ], + [ + 208, + 81 + ] + ], + [ + [ + 160, + 80 + ], + [ + 185, + 89 + ], + [ + 160, + 53 + ], + [ + 161, + 106 + ], + [ + 135, + 44 + ], + [ + 139, + 97 + ] + ], + [ + [ + 155, + 65 + ], + [ + 137, + 82 + ], + [ + 128, + 54 + ], + [ + 178, + 75 + ], + [ + 147, + 35 + ], + [ + 197, + 59 + ] + ] + ] + }, + { + "task_id": "0002905", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0006_cfg_0003_104122062823#167", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002905/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 123, + 58 + ], + [ + 113, + 32 + ], + [ + 144, + 44 + ], + [ + 102, + 72 + ], + [ + 152, + 68 + ], + [ + 112, + 96 + ] + ], + [ + [ + 152, + 44 + ], + [ + 176, + 37 + ], + [ + 137, + 25 + ], + [ + 168, + 63 + ], + [ + 110, + 32 + ], + [ + 143, + 71 + ] + ], + [ + [ + 101, + 50 + ], + [ + 70, + 35 + ], + [ + 100, + 45 + ], + [ + 103, + 54 + ], + [ + 127, + 58 + ], + [ + 133, + 68 + ] + ], + [ + [ + 154, + 82 + ], + [ + 176, + 87 + ], + [ + 154, + 54 + ], + [ + 153, + 109 + ], + [ + 132, + 50 + ], + [ + 132, + 104 + ] + ] + ] + }, + { + "task_id": "0002906", + "id": "RH20T_cfg7_task_0038_user_0020_scene_0006_cfg_0007_104122061602#191", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002906/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 463, + 162 + ], + [ + 448, + 176 + ], + [ + 447, + 141 + ], + [ + 479, + 183 + ], + [ + 461, + 127 + ], + [ + 493, + 168 + ] + ], + [ + [ + 478, + 146 + ], + [ + 460, + 164 + ], + [ + 489, + 151 + ], + [ + 468, + 142 + ], + [ + 509, + 132 + ], + [ + 486, + 125 + ] + ], + [ + [ + 465, + 131 + ], + [ + 469, + 107 + ], + [ + 445, + 136 + ], + [ + 485, + 127 + ], + [ + 441, + 159 + ], + [ + 480, + 150 + ] + ], + [ + [ + 468, + 153 + ], + [ + 476, + 134 + ], + [ + 484, + 170 + ], + [ + 453, + 137 + ], + [ + 476, + 187 + ], + [ + 447, + 154 + ] + ] + ] + }, + { + "task_id": "0002907", + "id": "RH20T_cfg6_task_0046_user_0020_scene_0004_cfg_0006_104122061018#370", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002907/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 304, + 57 + ], + [ + 299, + 53 + ], + [ + 328, + 76 + ], + [ + 278, + 36 + ], + [ + 336, + 81 + ], + [ + 282, + 39 + ] + ], + [ + [ + 321, + 86 + ], + [ + 306, + 63 + ], + [ + 323, + 100 + ], + [ + 319, + 71 + ], + [ + 338, + 123 + ], + [ + 336, + 96 + ] + ], + [ + [ + 310, + 88 + ], + [ + 296, + 84 + ], + [ + 319, + 119 + ], + [ + 301, + 56 + ], + [ + 335, + 126 + ], + [ + 317, + 58 + ] + ], + [ + [ + 295, + 108 + ], + [ + 270, + 117 + ], + [ + 312, + 135 + ], + [ + 277, + 79 + ], + [ + 339, + 127 + ], + [ + 305, + 68 + ] + ] + ] + }, + { + "task_id": "0002909", + "id": "RH20T_cfg1_task_0045_user_0015_scene_0002_cfg_0001_035622060973#136", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the brush on the pen rack'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002909/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 238, + 207 + ], + [ + 218, + 197 + ], + [ + 228, + 227 + ], + [ + 248, + 187 + ], + [ + 249, + 237 + ], + [ + 268, + 197 + ] + ], + [ + [ + 230, + 246 + ], + [ + 217, + 244 + ], + [ + 237, + 265 + ], + [ + 223, + 227 + ], + [ + 248, + 266 + ], + [ + 234, + 230 + ] + ], + [ + [ + 251, + 230 + ], + [ + 249, + 222 + ], + [ + 230, + 227 + ], + [ + 273, + 234 + ], + [ + 233, + 235 + ], + [ + 274, + 242 + ] + ], + [ + [ + 248, + 222 + ], + [ + 227, + 215 + ], + [ + 255, + 224 + ], + [ + 241, + 220 + ], + [ + 275, + 230 + ], + [ + 261, + 226 + ] + ] + ] + }, + { + "task_id": "0002913", + "id": "RH20T_cfg1_task_0038_user_0016_scene_0009_cfg_0001_039422060546#110", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002913/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 287, + 164 + ], + [ + 284, + 163 + ], + [ + 282, + 128 + ], + [ + 292, + 200 + ], + [ + 285, + 126 + ], + [ + 295, + 204 + ] + ], + [ + [ + 324, + 184 + ], + [ + 346, + 210 + ], + [ + 347, + 178 + ], + [ + 303, + 188 + ], + [ + 322, + 148 + ], + [ + 281, + 162 + ] + ], + [ + [ + 310, + 150 + ], + [ + 294, + 128 + ], + [ + 310, + 172 + ], + [ + 310, + 126 + ], + [ + 326, + 196 + ], + [ + 327, + 151 + ] + ], + [ + [ + 297, + 171 + ], + [ + 289, + 146 + ], + [ + 290, + 200 + ], + [ + 302, + 145 + ], + [ + 298, + 222 + ], + [ + 309, + 168 + ] + ] + ] + }, + { + "task_id": "0002915", + "id": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_036422060909#181", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002915/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 196, + 185 + ], + [ + 199, + 189 + ], + [ + 178, + 208 + ], + [ + 215, + 163 + ], + [ + 177, + 203 + ], + [ + 211, + 160 + ] + ], + [ + [ + 182, + 224 + ], + [ + 188, + 231 + ], + [ + 189, + 250 + ], + [ + 174, + 197 + ], + [ + 183, + 242 + ], + [ + 169, + 192 + ] + ], + [ + [ + 185, + 208 + ], + [ + 212, + 204 + ], + [ + 190, + 229 + ], + [ + 180, + 185 + ], + [ + 163, + 233 + ], + [ + 152, + 188 + ] + ], + [ + [ + 206, + 209 + ], + [ + 203, + 191 + ], + [ + 205, + 187 + ], + [ + 207, + 233 + ], + [ + 207, + 204 + ], + [ + 209, + 249 + ] + ] + ] + }, + { + "task_id": "0002918", + "id": "RH20T_cfg4_task_0064_user_0016_scene_0003_cfg_0004_104122063550#75", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002918/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 381, + 287 + ], + [ + 366, + 315 + ], + [ + 368, + 272 + ], + [ + 396, + 304 + ], + [ + 382, + 245 + ], + [ + 412, + 273 + ] + ], + [ + [ + 349, + 294 + ], + [ + 351, + 293 + ], + [ + 381, + 288 + ], + [ + 318, + 301 + ], + [ + 377, + 290 + ], + [ + 318, + 302 + ] + ], + [ + [ + 332, + 279 + ], + [ + 324, + 254 + ], + [ + 360, + 266 + ], + [ + 305, + 292 + ], + [ + 371, + 293 + ], + [ + 313, + 320 + ] + ], + [ + [ + 354, + 280 + ], + [ + 349, + 312 + ], + [ + 327, + 275 + ], + [ + 380, + 285 + ], + [ + 332, + 244 + ], + [ + 385, + 255 + ] + ] + ] + }, + { + "task_id": "0002921", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0008_cfg_0006_104122061602#141", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002921/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 455, + 166 + ], + [ + 446, + 159 + ], + [ + 462, + 140 + ], + [ + 449, + 192 + ], + [ + 470, + 146 + ], + [ + 458, + 197 + ] + ], + [ + [ + 459, + 182 + ], + [ + 462, + 198 + ], + [ + 476, + 164 + ], + [ + 442, + 201 + ], + [ + 474, + 146 + ], + [ + 438, + 184 + ] + ], + [ + [ + 428, + 193 + ], + [ + 404, + 207 + ], + [ + 411, + 173 + ], + [ + 446, + 215 + ], + [ + 435, + 159 + ], + [ + 471, + 200 + ] + ], + [ + [ + 449, + 179 + ], + [ + 470, + 185 + ], + [ + 451, + 206 + ], + [ + 447, + 152 + ], + [ + 432, + 199 + ], + [ + 427, + 148 + ] + ] + ] + }, + { + "task_id": "0002926", + "id": "RH20T_cfg7_task_0038_user_0020_scene_0006_cfg_0007_104122063633#180", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002926/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 150, + 200 + ], + [ + 174, + 206 + ], + [ + 145, + 199 + ], + [ + 154, + 201 + ], + [ + 121, + 193 + ], + [ + 127, + 194 + ] + ], + [ + [ + 171, + 181 + ], + [ + 170, + 161 + ], + [ + 177, + 166 + ], + [ + 165, + 197 + ], + [ + 178, + 184 + ], + [ + 166, + 216 + ] + ], + [ + [ + 169, + 204 + ], + [ + 170, + 179 + ], + [ + 176, + 202 + ], + [ + 161, + 207 + ], + [ + 175, + 226 + ], + [ + 161, + 233 + ] + ], + [ + [ + 169, + 195 + ], + [ + 176, + 216 + ], + [ + 147, + 205 + ], + [ + 189, + 185 + ], + [ + 138, + 182 + ], + [ + 181, + 163 + ] + ] + ] + }, + { + "task_id": "0002928", + "id": "RH20T_cfg3_task_0043_user_0010_scene_0006_cfg_0003_104122063550#101", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002928/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 390, + 195 + ], + [ + 393, + 200 + ], + [ + 398, + 226 + ], + [ + 382, + 164 + ], + [ + 395, + 220 + ], + [ + 380, + 162 + ] + ], + [ + [ + 369, + 192 + ], + [ + 350, + 176 + ], + [ + 349, + 211 + ], + [ + 389, + 174 + ], + [ + 369, + 229 + ], + [ + 410, + 191 + ] + ], + [ + [ + 394, + 192 + ], + [ + 390, + 161 + ], + [ + 399, + 201 + ], + [ + 389, + 184 + ], + [ + 403, + 233 + ], + [ + 393, + 213 + ] + ], + [ + [ + 391, + 200 + ], + [ + 399, + 184 + ], + [ + 420, + 201 + ], + [ + 361, + 199 + ], + [ + 412, + 215 + ], + [ + 356, + 213 + ] + ] + ] + }, + { + "task_id": "0002929", + "id": "RH20T_cfg3_task_0091_user_0010_scene_0004_cfg_0003_104122062823#217", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002929/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 358, + 79 + ], + [ + 364, + 64 + ], + [ + 347, + 45 + ], + [ + 369, + 110 + ], + [ + 339, + 60 + ], + [ + 363, + 127 + ] + ], + [ + [ + 369, + 79 + ], + [ + 372, + 47 + ], + [ + 402, + 83 + ], + [ + 336, + 74 + ], + [ + 397, + 113 + ], + [ + 334, + 104 + ] + ], + [ + [ + 390, + 84 + ], + [ + 357, + 84 + ], + [ + 392, + 119 + ], + [ + 387, + 49 + ], + [ + 424, + 118 + ], + [ + 419, + 49 + ] + ], + [ + [ + 356, + 93 + ], + [ + 355, + 105 + ], + [ + 388, + 76 + ], + [ + 326, + 109 + ], + [ + 387, + 66 + ], + [ + 328, + 98 + ] + ] + ] + }, + { + "task_id": "0002931", + "id": "RH20T_cfg3_task_0091_user_0010_scene_0004_cfg_0003_104122062823#502", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002931/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 317, + 120 + ], + [ + 320, + 108 + ], + [ + 296, + 140 + ], + [ + 337, + 101 + ], + [ + 294, + 150 + ], + [ + 333, + 113 + ] + ], + [ + [ + 325, + 116 + ], + [ + 355, + 111 + ], + [ + 326, + 145 + ], + [ + 324, + 84 + ], + [ + 296, + 150 + ], + [ + 293, + 88 + ] + ], + [ + [ + 317, + 95 + ], + [ + 317, + 69 + ], + [ + 332, + 79 + ], + [ + 300, + 112 + ], + [ + 331, + 103 + ], + [ + 301, + 136 + ] + ], + [ + [ + 341, + 83 + ], + [ + 325, + 101 + ], + [ + 343, + 106 + ], + [ + 339, + 61 + ], + [ + 361, + 87 + ], + [ + 356, + 41 + ] + ] + ] + }, + { + "task_id": "0002934", + "id": "RH20T_cfg3_task_0038_user_0010_scene_0003_cfg_0003_104122063550#93", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002934/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 238, + 204 + ], + [ + 269, + 203 + ], + [ + 235, + 233 + ], + [ + 242, + 176 + ], + [ + 204, + 233 + ], + [ + 212, + 176 + ] + ], + [ + [ + 221, + 238 + ], + [ + 197, + 238 + ], + [ + 233, + 217 + ], + [ + 208, + 261 + ], + [ + 258, + 217 + ], + [ + 234, + 262 + ] + ], + [ + [ + 224, + 210 + ], + [ + 209, + 183 + ], + [ + 250, + 196 + ], + [ + 200, + 223 + ], + [ + 265, + 223 + ], + [ + 215, + 250 + ] + ], + [ + [ + 232, + 226 + ], + [ + 229, + 217 + ], + [ + 236, + 256 + ], + [ + 228, + 197 + ], + [ + 238, + 262 + ], + [ + 230, + 207 + ] + ] + ] + }, + { + "task_id": "0002937", + "id": "RH20T_cfg6_task_0008_user_0014_scene_0007_cfg_0006_104122061330#166", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002937/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 295, + 185 + ], + [ + 292, + 213 + ], + [ + 304, + 184 + ], + [ + 285, + 187 + ], + [ + 307, + 157 + ], + [ + 289, + 157 + ] + ], + [ + [ + 275, + 194 + ], + [ + 264, + 196 + ], + [ + 252, + 206 + ], + [ + 298, + 182 + ], + [ + 264, + 204 + ], + [ + 311, + 179 + ] + ], + [ + [ + 293, + 197 + ], + [ + 287, + 181 + ], + [ + 280, + 220 + ], + [ + 305, + 175 + ], + [ + 287, + 233 + ], + [ + 311, + 189 + ] + ], + [ + [ + 272, + 183 + ], + [ + 242, + 188 + ], + [ + 268, + 162 + ], + [ + 275, + 206 + ], + [ + 296, + 159 + ], + [ + 305, + 202 + ] + ] + ] + }, + { + "task_id": "0002938", + "id": "RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_104122061018#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002938/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 325, + 99 + ], + [ + 328, + 79 + ], + [ + 313, + 81 + ], + [ + 337, + 118 + ], + [ + 310, + 100 + ], + [ + 333, + 136 + ] + ], + [ + [ + 353, + 78 + ], + [ + 349, + 50 + ], + [ + 376, + 76 + ], + [ + 328, + 80 + ], + [ + 379, + 104 + ], + [ + 331, + 109 + ] + ], + [ + [ + 335, + 102 + ], + [ + 316, + 119 + ], + [ + 332, + 100 + ], + [ + 337, + 105 + ], + [ + 350, + 84 + ], + [ + 357, + 86 + ] + ], + [ + [ + 301, + 108 + ], + [ + 305, + 119 + ], + [ + 300, + 132 + ], + [ + 302, + 85 + ], + [ + 297, + 120 + ], + [ + 299, + 71 + ] + ] + ] + }, + { + "task_id": "0002939", + "id": "RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_104122061018#164", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002939/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 312, + 97 + ], + [ + 291, + 105 + ], + [ + 298, + 74 + ], + [ + 325, + 120 + ], + [ + 321, + 64 + ], + [ + 348, + 112 + ] + ], + [ + [ + 317, + 91 + ], + [ + 337, + 79 + ], + [ + 300, + 71 + ], + [ + 333, + 109 + ], + [ + 281, + 84 + ], + [ + 314, + 119 + ] + ], + [ + [ + 299, + 140 + ], + [ + 276, + 156 + ], + [ + 283, + 119 + ], + [ + 315, + 162 + ], + [ + 306, + 104 + ], + [ + 337, + 146 + ] + ], + [ + [ + 305, + 106 + ], + [ + 308, + 86 + ], + [ + 294, + 88 + ], + [ + 317, + 126 + ], + [ + 292, + 106 + ], + [ + 313, + 144 + ] + ] + ] + }, + { + "task_id": "0002943", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_104122063678#293", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002943/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 330, + 233 + ], + [ + 306, + 240 + ], + [ + 319, + 213 + ], + [ + 341, + 253 + ], + [ + 344, + 206 + ], + [ + 366, + 247 + ] + ], + [ + [ + 307, + 231 + ], + [ + 307, + 219 + ], + [ + 331, + 226 + ], + [ + 282, + 237 + ], + [ + 329, + 237 + ], + [ + 283, + 249 + ] + ], + [ + [ + 306, + 247 + ], + [ + 308, + 232 + ], + [ + 309, + 271 + ], + [ + 304, + 225 + ], + [ + 306, + 284 + ], + [ + 302, + 239 + ] + ], + [ + [ + 330, + 259 + ], + [ + 357, + 256 + ], + [ + 338, + 270 + ], + [ + 323, + 250 + ], + [ + 308, + 275 + ], + [ + 297, + 253 + ] + ] + ] + }, + { + "task_id": "0002945", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_104122063678#716", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002945/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 355, + 224 + ], + [ + 342, + 231 + ], + [ + 374, + 239 + ], + [ + 336, + 210 + ], + [ + 390, + 232 + ], + [ + 350, + 202 + ] + ], + [ + [ + 372, + 209 + ], + [ + 378, + 194 + ], + [ + 377, + 229 + ], + [ + 367, + 191 + ], + [ + 371, + 243 + ], + [ + 361, + 205 + ] + ], + [ + [ + 399, + 210 + ], + [ + 374, + 211 + ], + [ + 398, + 192 + ], + [ + 401, + 227 + ], + [ + 424, + 190 + ], + [ + 425, + 226 + ] + ], + [ + [ + 359, + 224 + ], + [ + 372, + 234 + ], + [ + 343, + 220 + ], + [ + 377, + 227 + ], + [ + 329, + 209 + ], + [ + 363, + 216 + ] + ] + ] + }, + { + "task_id": "0002946", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_104122063678#1006", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002946/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 297, + 205 + ], + [ + 320, + 207 + ], + [ + 305, + 192 + ], + [ + 290, + 217 + ], + [ + 280, + 190 + ], + [ + 267, + 215 + ] + ], + [ + [ + 307, + 213 + ], + [ + 292, + 212 + ], + [ + 288, + 221 + ], + [ + 325, + 207 + ], + [ + 304, + 222 + ], + [ + 342, + 208 + ] + ], + [ + [ + 304, + 205 + ], + [ + 308, + 189 + ], + [ + 304, + 225 + ], + [ + 304, + 186 + ], + [ + 300, + 240 + ], + [ + 301, + 202 + ] + ], + [ + [ + 320, + 210 + ], + [ + 314, + 186 + ], + [ + 346, + 204 + ], + [ + 294, + 215 + ], + [ + 353, + 229 + ], + [ + 300, + 241 + ] + ] + ] + }, + { + "task_id": "0002950", + "id": "RH20T_cfg4_task_0065_user_0015_scene_0012_cfg_0004_104122062823#208", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002950/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 240, + 69 + ], + [ + 247, + 69 + ], + [ + 230, + 44 + ], + [ + 248, + 94 + ], + [ + 224, + 45 + ], + [ + 241, + 92 + ] + ], + [ + [ + 236, + 61 + ], + [ + 210, + 54 + ], + [ + 234, + 87 + ], + [ + 238, + 36 + ], + [ + 259, + 93 + ], + [ + 263, + 43 + ] + ], + [ + [ + 228, + 43 + ], + [ + 254, + 51 + ], + [ + 238, + 29 + ], + [ + 220, + 55 + ], + [ + 209, + 19 + ], + [ + 194, + 47 + ] + ], + [ + [ + 239, + 45 + ], + [ + 234, + 20 + ], + [ + 257, + 54 + ], + [ + 222, + 36 + ], + [ + 261, + 76 + ], + [ + 227, + 59 + ] + ] + ] + }, + { + "task_id": "0002951", + "id": "RH20T_cfg4_task_0065_user_0015_scene_0012_cfg_0004_104122062823#578", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002951/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 172, + 76 + ], + [ + 193, + 69 + ], + [ + 163, + 99 + ], + [ + 182, + 51 + ], + [ + 141, + 107 + ], + [ + 159, + 56 + ] + ], + [ + [ + 195, + 78 + ], + [ + 198, + 77 + ], + [ + 189, + 46 + ], + [ + 200, + 108 + ], + [ + 186, + 48 + ], + [ + 197, + 107 + ] + ], + [ + [ + 209, + 80 + ], + [ + 202, + 53 + ], + [ + 233, + 88 + ], + [ + 187, + 72 + ], + [ + 238, + 112 + ], + [ + 194, + 96 + ] + ], + [ + [ + 229, + 100 + ], + [ + 247, + 84 + ], + [ + 238, + 82 + ], + [ + 221, + 115 + ], + [ + 218, + 100 + ], + [ + 202, + 132 + ] + ] + ] + }, + { + "task_id": "0002953", + "id": "RH20T_cfg4_task_0033_user_0016_scene_0008_cfg_0004_104422070011#129", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002953/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 201, + 195 + ], + [ + 215, + 219 + ], + [ + 192, + 209 + ], + [ + 211, + 179 + ], + [ + 179, + 186 + ], + [ + 197, + 155 + ] + ], + [ + [ + 211, + 170 + ], + [ + 233, + 188 + ], + [ + 234, + 148 + ], + [ + 186, + 193 + ], + [ + 212, + 132 + ], + [ + 165, + 175 + ] + ], + [ + [ + 171, + 185 + ], + [ + 169, + 175 + ], + [ + 151, + 211 + ], + [ + 190, + 161 + ], + [ + 154, + 218 + ], + [ + 191, + 171 + ] + ], + [ + [ + 186, + 207 + ], + [ + 205, + 210 + ], + [ + 160, + 189 + ], + [ + 209, + 223 + ], + [ + 143, + 187 + ], + [ + 191, + 220 + ] + ] + ] + }, + { + "task_id": "0002958", + "id": "RH20T_cfg5_task_0078_user_0007_scene_0010_cfg_0005_104122061850#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close the microwave door'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002958/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 375, + 116 + ], + [ + 357, + 118 + ], + [ + 397, + 141 + ], + [ + 352, + 90 + ], + [ + 418, + 141 + ], + [ + 371, + 86 + ] + ], + [ + [ + 375, + 154 + ], + [ + 375, + 190 + ], + [ + 355, + 169 + ], + [ + 398, + 137 + ], + [ + 355, + 137 + ], + [ + 397, + 101 + ] + ], + [ + [ + 387, + 154 + ], + [ + 384, + 133 + ], + [ + 362, + 179 + ], + [ + 412, + 130 + ], + [ + 366, + 196 + ], + [ + 413, + 149 + ] + ], + [ + [ + 426, + 139 + ], + [ + 418, + 115 + ], + [ + 453, + 113 + ], + [ + 399, + 166 + ], + [ + 459, + 136 + ], + [ + 408, + 187 + ] + ] + ] + }, + { + "task_id": "0002961", + "id": "RH20T_cfg1_task_0111_user_0010_scene_0006_cfg_0001_039422060546#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object with the corresponding shape into the corresponding hole'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002961/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 402, + 209 + ], + [ + 378, + 198 + ], + [ + 381, + 229 + ], + [ + 421, + 191 + ], + [ + 407, + 243 + ], + [ + 447, + 202 + ] + ], + [ + [ + 389, + 195 + ], + [ + 392, + 175 + ], + [ + 375, + 219 + ], + [ + 401, + 173 + ], + [ + 373, + 237 + ], + [ + 398, + 192 + ] + ], + [ + [ + 419, + 212 + ], + [ + 446, + 231 + ], + [ + 426, + 203 + ], + [ + 412, + 222 + ], + [ + 401, + 186 + ], + [ + 385, + 202 + ] + ], + [ + [ + 347, + 220 + ], + [ + 354, + 223 + ], + [ + 374, + 236 + ], + [ + 321, + 204 + ], + [ + 367, + 233 + ], + [ + 312, + 199 + ] + ] + ] + }, + { + "task_id": "0002962", + "id": "RH20T_cfg1_task_0065_user_0015_scene_0001_cfg_0001_750612070851#302", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002962/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 535, + 225 + ], + [ + 570, + 234 + ], + [ + 530, + 255 + ], + [ + 540, + 197 + ], + [ + 496, + 246 + ], + [ + 507, + 189 + ] + ], + [ + [ + 527, + 217 + ], + [ + 541, + 206 + ], + [ + 499, + 226 + ], + [ + 558, + 207 + ], + [ + 483, + 238 + ], + [ + 542, + 219 + ] + ], + [ + [ + 541, + 235 + ], + [ + 555, + 216 + ], + [ + 534, + 263 + ], + [ + 548, + 210 + ], + [ + 521, + 279 + ], + [ + 535, + 228 + ] + ], + [ + [ + 537, + 222 + ], + [ + 515, + 238 + ], + [ + 527, + 192 + ], + [ + 547, + 254 + ], + [ + 550, + 172 + ], + [ + 573, + 237 + ] + ] + ] + }, + { + "task_id": "0002963", + "id": "RH20T_cfg1_task_0065_user_0015_scene_0001_cfg_0001_750612070851#515", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002963/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 367, + 210 + ], + [ + 338, + 197 + ], + [ + 370, + 205 + ], + [ + 362, + 217 + ], + [ + 398, + 218 + ], + [ + 393, + 232 + ] + ], + [ + [ + 383, + 226 + ], + [ + 354, + 220 + ], + [ + 379, + 260 + ], + [ + 386, + 193 + ], + [ + 407, + 264 + ], + [ + 413, + 199 + ] + ], + [ + [ + 385, + 217 + ], + [ + 389, + 197 + ], + [ + 381, + 246 + ], + [ + 389, + 190 + ], + [ + 378, + 263 + ], + [ + 385, + 208 + ] + ], + [ + [ + 406, + 207 + ], + [ + 442, + 214 + ], + [ + 403, + 238 + ], + [ + 409, + 180 + ], + [ + 366, + 231 + ], + [ + 375, + 174 + ] + ] + ] + }, + { + "task_id": "0002966", + "id": "RH20T_cfg2_task_0011_user_0013_scene_0001_cfg_0002_105422061350#300", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002966/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 234, + 238 + ], + [ + 199, + 245 + ], + [ + 238, + 275 + ], + [ + 230, + 203 + ], + [ + 274, + 269 + ], + [ + 265, + 197 + ] + ], + [ + [ + 202, + 238 + ], + [ + 178, + 250 + ], + [ + 174, + 253 + ], + [ + 231, + 223 + ], + [ + 195, + 242 + ], + [ + 249, + 214 + ] + ], + [ + [ + 217, + 212 + ], + [ + 214, + 195 + ], + [ + 230, + 242 + ], + [ + 204, + 183 + ], + [ + 231, + 255 + ], + [ + 207, + 199 + ] + ], + [ + [ + 187, + 182 + ], + [ + 162, + 175 + ], + [ + 207, + 199 + ], + [ + 164, + 162 + ], + [ + 234, + 208 + ], + [ + 192, + 169 + ] + ] + ] + }, + { + "task_id": "0002970", + "id": "RH20T_cfg1_task_0061_user_0015_scene_0001_cfg_0001_819112071037#990", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002970/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 514, + 141 + ], + [ + 533, + 91 + ], + [ + 488, + 149 + ], + [ + 537, + 134 + ], + [ + 470, + 198 + ], + [ + 519, + 177 + ] + ], + [ + [ + 489, + 112 + ], + [ + 458, + 87 + ], + [ + 530, + 85 + ], + [ + 448, + 139 + ], + [ + 569, + 111 + ], + [ + 482, + 170 + ] + ], + [ + [ + 554, + 106 + ], + [ + 613, + 127 + ], + [ + 549, + 146 + ], + [ + 561, + 58 + ], + [ + 503, + 127 + ], + [ + 507, + 42 + ] + ], + [ + [ + 556, + 159 + ], + [ + 599, + 146 + ], + [ + 554, + 106 + ], + [ + 559, + 209 + ], + [ + 507, + 119 + ], + [ + 515, + 223 + ] + ] + ] + }, + { + "task_id": "0002973", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_037522062165#168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002973/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 204, + 202 + ], + [ + 224, + 181 + ], + [ + 212, + 189 + ], + [ + 197, + 214 + ], + [ + 190, + 212 + ], + [ + 177, + 236 + ] + ], + [ + [ + 175, + 200 + ], + [ + 200, + 202 + ], + [ + 190, + 227 + ], + [ + 161, + 174 + ], + [ + 162, + 226 + ], + [ + 133, + 170 + ] + ], + [ + [ + 231, + 214 + ], + [ + 212, + 234 + ], + [ + 209, + 207 + ], + [ + 255, + 220 + ], + [ + 227, + 189 + ], + [ + 272, + 200 + ] + ], + [ + [ + 213, + 227 + ], + [ + 209, + 211 + ], + [ + 197, + 203 + ], + [ + 229, + 253 + ], + [ + 200, + 217 + ], + [ + 231, + 265 + ] + ] + ] + }, + { + "task_id": "0002976", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_750612070853#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002976/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 358, + 150 + ], + [ + 393, + 137 + ], + [ + 371, + 172 + ], + [ + 343, + 124 + ], + [ + 339, + 183 + ], + [ + 308, + 137 + ] + ], + [ + [ + 387, + 151 + ], + [ + 351, + 153 + ], + [ + 364, + 140 + ], + [ + 406, + 161 + ], + [ + 405, + 137 + ], + [ + 443, + 159 + ] + ], + [ + [ + 362, + 139 + ], + [ + 359, + 102 + ], + [ + 324, + 150 + ], + [ + 398, + 128 + ], + [ + 329, + 184 + ], + [ + 399, + 162 + ] + ], + [ + [ + 373, + 84 + ], + [ + 370, + 53 + ], + [ + 398, + 62 + ], + [ + 351, + 104 + ], + [ + 402, + 98 + ], + [ + 353, + 136 + ] + ] + ] + }, + { + "task_id": "0002978", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_750612070853#459", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002978/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 567, + 331 + ], + [ + 487, + 319 + ], + [ + 574, + 360 + ], + [ + 561, + 263 + ], + [ + 640, + 360 + ], + [ + 640, + 271 + ] + ], + [ + [ + 612, + 320 + ], + [ + 640, + 257 + ], + [ + 581, + 304 + ], + [ + 638, + 333 + ], + [ + 544, + 360 + ], + [ + 607, + 360 + ] + ], + [ + [ + 616, + 295 + ], + [ + 618, + 267 + ], + [ + 568, + 237 + ], + [ + 640, + 360 + ], + [ + 571, + 264 + ], + [ + 640, + 360 + ] + ], + [ + [ + 608, + 360 + ], + [ + 624, + 318 + ], + [ + 532, + 332 + ], + [ + 640, + 360 + ], + [ + 526, + 360 + ], + [ + 640, + 360 + ] + ] + ] + }, + { + "task_id": "0002983", + "id": "RH20T_cfg7_task_0047_user_0020_scene_0004_cfg_0007_104122061602#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002983/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 482, + 104 + ], + [ + 482, + 79 + ], + [ + 478, + 114 + ], + [ + 486, + 96 + ], + [ + 478, + 139 + ], + [ + 486, + 119 + ] + ], + [ + [ + 510, + 127 + ], + [ + 528, + 134 + ], + [ + 489, + 145 + ], + [ + 532, + 109 + ], + [ + 475, + 138 + ], + [ + 515, + 104 + ] + ], + [ + [ + 493, + 129 + ], + [ + 512, + 138 + ], + [ + 500, + 148 + ], + [ + 485, + 108 + ], + [ + 483, + 139 + ], + [ + 468, + 100 + ] + ], + [ + [ + 490, + 150 + ], + [ + 516, + 147 + ], + [ + 496, + 176 + ], + [ + 485, + 125 + ], + [ + 469, + 180 + ], + [ + 459, + 128 + ] + ] + ] + }, + { + "task_id": "0002984", + "id": "RH20T_cfg5_task_0029_user_0010_scene_0006_cfg_0005_036422060215#147", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002984/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 246, + 100 + ], + [ + 232, + 119 + ], + [ + 242, + 103 + ], + [ + 251, + 98 + ], + [ + 258, + 83 + ], + [ + 265, + 80 + ] + ], + [ + [ + 254, + 132 + ], + [ + 279, + 127 + ], + [ + 254, + 139 + ], + [ + 254, + 125 + ], + [ + 228, + 145 + ], + [ + 231, + 130 + ] + ], + [ + [ + 231, + 132 + ], + [ + 244, + 117 + ], + [ + 233, + 116 + ], + [ + 229, + 146 + ], + [ + 219, + 132 + ], + [ + 216, + 162 + ] + ], + [ + [ + 257, + 121 + ], + [ + 264, + 100 + ], + [ + 281, + 128 + ], + [ + 233, + 114 + ], + [ + 274, + 147 + ], + [ + 227, + 134 + ] + ] + ] + }, + { + "task_id": "0002985", + "id": "RH20T_cfg5_task_0029_user_0010_scene_0006_cfg_0005_036422060215#196", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002985/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 263, + 139 + ], + [ + 270, + 120 + ], + [ + 286, + 146 + ], + [ + 240, + 132 + ], + [ + 280, + 163 + ], + [ + 234, + 150 + ] + ], + [ + [ + 241, + 142 + ], + [ + 242, + 158 + ], + [ + 223, + 155 + ], + [ + 259, + 128 + ], + [ + 223, + 140 + ], + [ + 258, + 114 + ] + ], + [ + [ + 248, + 129 + ], + [ + 265, + 121 + ], + [ + 235, + 139 + ], + [ + 259, + 119 + ], + [ + 219, + 147 + ], + [ + 243, + 126 + ] + ], + [ + [ + 276, + 144 + ], + [ + 293, + 158 + ], + [ + 266, + 135 + ], + [ + 285, + 152 + ], + [ + 250, + 122 + ], + [ + 269, + 140 + ] + ] + ] + }, + { + "task_id": "0002986", + "id": "RH20T_cfg5_task_0029_user_0010_scene_0006_cfg_0005_036422060215#241", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002986/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 268, + 128 + ], + [ + 269, + 114 + ], + [ + 257, + 144 + ], + [ + 280, + 111 + ], + [ + 255, + 158 + ], + [ + 280, + 125 + ] + ], + [ + [ + 271, + 136 + ], + [ + 247, + 129 + ], + [ + 278, + 114 + ], + [ + 264, + 158 + ], + [ + 303, + 120 + ], + [ + 289, + 164 + ] + ], + [ + [ + 270, + 117 + ], + [ + 270, + 108 + ], + [ + 287, + 102 + ], + [ + 253, + 132 + ], + [ + 287, + 111 + ], + [ + 251, + 142 + ] + ], + [ + [ + 278, + 135 + ], + [ + 284, + 116 + ], + [ + 301, + 140 + ], + [ + 254, + 130 + ], + [ + 295, + 158 + ], + [ + 250, + 148 + ] + ] + ] + }, + { + "task_id": "0002990", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0003_cfg_0001_819112070044#505", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002990/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 321, + 208 + ], + [ + 321, + 221 + ], + [ + 341, + 215 + ], + [ + 300, + 201 + ], + [ + 340, + 203 + ], + [ + 301, + 189 + ] + ], + [ + [ + 329, + 202 + ], + [ + 331, + 187 + ], + [ + 331, + 187 + ], + [ + 328, + 218 + ], + [ + 329, + 201 + ], + [ + 326, + 232 + ] + ], + [ + [ + 347, + 205 + ], + [ + 354, + 189 + ], + [ + 357, + 220 + ], + [ + 339, + 191 + ], + [ + 350, + 235 + ], + [ + 333, + 206 + ] + ], + [ + [ + 339, + 221 + ], + [ + 336, + 234 + ], + [ + 341, + 206 + ], + [ + 337, + 237 + ], + [ + 345, + 192 + ], + [ + 341, + 224 + ] + ] + ] + }, + { + "task_id": "0002995", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0003_cfg_0001_819112070044#1254", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0002995/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 257, + 189 + ], + [ + 246, + 206 + ], + [ + 258, + 175 + ], + [ + 257, + 202 + ], + [ + 269, + 160 + ], + [ + 267, + 187 + ] + ], + [ + [ + 274, + 200 + ], + [ + 292, + 205 + ], + [ + 277, + 180 + ], + [ + 271, + 221 + ], + [ + 258, + 176 + ], + [ + 252, + 217 + ] + ], + [ + [ + 247, + 178 + ], + [ + 241, + 159 + ], + [ + 229, + 188 + ], + [ + 265, + 170 + ], + [ + 235, + 207 + ], + [ + 270, + 188 + ] + ], + [ + [ + 262, + 195 + ], + [ + 264, + 182 + ], + [ + 265, + 178 + ], + [ + 259, + 213 + ], + [ + 264, + 190 + ], + [ + 258, + 224 + ] + ] + ] + }, + { + "task_id": "0003000", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_038522063145#303", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003000/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 494, + 157 + ], + [ + 501, + 140 + ], + [ + 514, + 170 + ], + [ + 475, + 145 + ], + [ + 506, + 186 + ], + [ + 470, + 161 + ] + ], + [ + [ + 504, + 160 + ], + [ + 504, + 149 + ], + [ + 485, + 171 + ], + [ + 525, + 150 + ], + [ + 484, + 183 + ], + [ + 525, + 162 + ] + ], + [ + [ + 507, + 168 + ], + [ + 523, + 182 + ], + [ + 492, + 185 + ], + [ + 523, + 151 + ], + [ + 476, + 171 + ], + [ + 507, + 136 + ] + ], + [ + [ + 494, + 162 + ], + [ + 516, + 158 + ], + [ + 505, + 176 + ], + [ + 482, + 147 + ], + [ + 485, + 179 + ], + [ + 461, + 151 + ] + ] + ] + }, + { + "task_id": "0003001", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_038522063145#416", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003001/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 471, + 162 + ], + [ + 452, + 154 + ], + [ + 460, + 182 + ], + [ + 483, + 142 + ], + [ + 478, + 190 + ], + [ + 502, + 151 + ] + ], + [ + [ + 488, + 146 + ], + [ + 470, + 140 + ], + [ + 480, + 133 + ], + [ + 495, + 158 + ], + [ + 499, + 138 + ], + [ + 514, + 164 + ] + ], + [ + [ + 474, + 147 + ], + [ + 480, + 129 + ], + [ + 493, + 159 + ], + [ + 456, + 135 + ], + [ + 487, + 175 + ], + [ + 451, + 152 + ] + ], + [ + [ + 469, + 150 + ], + [ + 453, + 134 + ], + [ + 456, + 168 + ], + [ + 482, + 133 + ], + [ + 471, + 183 + ], + [ + 498, + 148 + ] + ] + ] + }, + { + "task_id": "0003002", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_038522063145#556", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003002/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 463, + 121 + ], + [ + 469, + 103 + ], + [ + 483, + 132 + ], + [ + 445, + 111 + ], + [ + 477, + 148 + ], + [ + 440, + 127 + ] + ], + [ + [ + 469, + 112 + ], + [ + 448, + 117 + ], + [ + 477, + 99 + ], + [ + 462, + 124 + ], + [ + 499, + 93 + ], + [ + 483, + 119 + ] + ], + [ + [ + 463, + 132 + ], + [ + 450, + 145 + ], + [ + 460, + 146 + ], + [ + 467, + 117 + ], + [ + 472, + 133 + ], + [ + 481, + 103 + ] + ], + [ + [ + 445, + 104 + ], + [ + 446, + 124 + ], + [ + 421, + 106 + ], + [ + 469, + 101 + ], + [ + 420, + 86 + ], + [ + 468, + 80 + ] + ] + ] + }, + { + "task_id": "0003007", + "id": "RH20T_cfg6_task_0011_user_0014_scene_0007_cfg_0006_104122061018#129", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003007/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 327, + 62 + ], + [ + 331, + 33 + ], + [ + 314, + 46 + ], + [ + 341, + 79 + ], + [ + 312, + 72 + ], + [ + 337, + 105 + ] + ], + [ + [ + 314, + 42 + ], + [ + 343, + 48 + ], + [ + 303, + 48 + ], + [ + 328, + 34 + ], + [ + 275, + 42 + ], + [ + 296, + 26 + ] + ], + [ + [ + 312, + 68 + ], + [ + 307, + 89 + ], + [ + 292, + 83 + ], + [ + 331, + 54 + ], + [ + 297, + 59 + ], + [ + 337, + 31 + ] + ], + [ + [ + 316, + 55 + ], + [ + 322, + 48 + ], + [ + 314, + 84 + ], + [ + 319, + 24 + ], + [ + 308, + 94 + ], + [ + 313, + 30 + ] + ] + ] + }, + { + "task_id": "0003008", + "id": "RH20T_cfg4_task_0058_user_0014_scene_0008_cfg_0004_104122062295#142", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003008/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 115 + ], + [ + 346, + 90 + ], + [ + 323, + 98 + ], + [ + 352, + 131 + ], + [ + 314, + 126 + ], + [ + 344, + 157 + ] + ], + [ + [ + 343, + 127 + ], + [ + 346, + 104 + ], + [ + 312, + 122 + ], + [ + 375, + 133 + ], + [ + 311, + 143 + ], + [ + 371, + 154 + ] + ], + [ + [ + 337, + 112 + ], + [ + 349, + 86 + ], + [ + 327, + 121 + ], + [ + 347, + 101 + ], + [ + 315, + 147 + ], + [ + 333, + 130 + ] + ], + [ + [ + 364, + 143 + ], + [ + 340, + 155 + ], + [ + 345, + 134 + ], + [ + 382, + 150 + ], + [ + 372, + 119 + ], + [ + 407, + 138 + ] + ] + ] + }, + { + "task_id": "0003009", + "id": "RH20T_cfg4_task_0019_user_0016_scene_0005_cfg_0004_104122062823#470", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the first move as black on the 3-4 point in the upper right corner of the Go board'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003009/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 339, + 0 + ], + [ + 341, + 0 + ], + [ + 360, + 0 + ], + [ + 316, + 0 + ], + [ + 357, + 0 + ], + [ + 315, + 0 + ] + ], + [ + [ + 356, + 20 + ], + [ + 362, + 0 + ], + [ + 382, + 26 + ], + [ + 329, + 14 + ], + [ + 376, + 51 + ], + [ + 325, + 38 + ] + ], + [ + [ + 369, + 38 + ], + [ + 357, + 62 + ], + [ + 383, + 39 + ], + [ + 357, + 37 + ], + [ + 397, + 13 + ], + [ + 370, + 14 + ] + ], + [ + [ + 350, + 27 + ], + [ + 349, + 33 + ], + [ + 375, + 33 + ], + [ + 324, + 20 + ], + [ + 373, + 27 + ], + [ + 325, + 14 + ] + ] + ] + }, + { + "task_id": "0003011", + "id": "RH20T_cfg4_task_0018_user_0016_scene_0001_cfg_0004_104122062823#170", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003011/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 282, + 64 + ], + [ + 263, + 79 + ], + [ + 306, + 82 + ], + [ + 257, + 44 + ], + [ + 328, + 66 + ], + [ + 277, + 26 + ] + ], + [ + [ + 293, + 69 + ], + [ + 265, + 61 + ], + [ + 277, + 62 + ], + [ + 306, + 74 + ], + [ + 308, + 71 + ], + [ + 334, + 82 + ] + ], + [ + [ + 345, + 59 + ], + [ + 376, + 54 + ], + [ + 343, + 51 + ], + [ + 346, + 67 + ], + [ + 310, + 57 + ], + [ + 318, + 72 + ] + ], + [ + [ + 306, + 66 + ], + [ + 305, + 37 + ], + [ + 321, + 51 + ], + [ + 290, + 82 + ], + [ + 322, + 78 + ], + [ + 292, + 109 + ] + ] + ] + }, + { + "task_id": "0003012", + "id": "RH20T_cfg5_task_0010_user_0010_scene_0005_cfg_0005_105422061350#348", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003012/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 269, + 83 + ], + [ + 267, + 107 + ], + [ + 266, + 81 + ], + [ + 273, + 84 + ], + [ + 269, + 57 + ], + [ + 276, + 58 + ] + ], + [ + [ + 278, + 50 + ], + [ + 252, + 54 + ], + [ + 273, + 53 + ], + [ + 283, + 48 + ], + [ + 300, + 49 + ], + [ + 307, + 45 + ] + ], + [ + [ + 310, + 85 + ], + [ + 300, + 61 + ], + [ + 334, + 76 + ], + [ + 285, + 95 + ], + [ + 343, + 100 + ], + [ + 295, + 119 + ] + ], + [ + [ + 290, + 77 + ], + [ + 299, + 61 + ], + [ + 302, + 62 + ], + [ + 277, + 93 + ], + [ + 293, + 76 + ], + [ + 269, + 107 + ] + ] + ] + }, + { + "task_id": "0003014", + "id": "RH20T_cfg7_task_0065_user_0014_scene_0003_cfg_0007_104122064161#298", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003014/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 419, + 223 + ], + [ + 396, + 214 + ], + [ + 416, + 218 + ], + [ + 421, + 227 + ], + [ + 440, + 228 + ], + [ + 444, + 237 + ] + ], + [ + [ + 449, + 210 + ], + [ + 456, + 193 + ], + [ + 463, + 226 + ], + [ + 436, + 194 + ], + [ + 456, + 244 + ], + [ + 428, + 212 + ] + ], + [ + [ + 433, + 230 + ], + [ + 455, + 230 + ], + [ + 429, + 252 + ], + [ + 437, + 206 + ], + [ + 407, + 252 + ], + [ + 414, + 205 + ] + ], + [ + [ + 428, + 217 + ], + [ + 433, + 202 + ], + [ + 432, + 237 + ], + [ + 424, + 199 + ], + [ + 427, + 251 + ], + [ + 420, + 213 + ] + ] + ] + }, + { + "task_id": "0003016", + "id": "RH20T_cfg7_task_0063_user_0020_scene_0008_cfg_0007_104122064161#145", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a shovel to scoop up an object'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003016/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 132 + ], + [ + 377, + 121 + ], + [ + 378, + 143 + ], + [ + 346, + 121 + ], + [ + 364, + 152 + ], + [ + 333, + 131 + ] + ], + [ + [ + 355, + 135 + ], + [ + 353, + 142 + ], + [ + 358, + 155 + ], + [ + 352, + 115 + ], + [ + 359, + 148 + ], + [ + 353, + 110 + ] + ], + [ + [ + 359, + 115 + ], + [ + 349, + 102 + ], + [ + 340, + 121 + ], + [ + 379, + 109 + ], + [ + 349, + 132 + ], + [ + 387, + 122 + ] + ], + [ + [ + 360, + 131 + ], + [ + 361, + 120 + ], + [ + 362, + 149 + ], + [ + 359, + 114 + ], + [ + 362, + 159 + ], + [ + 358, + 125 + ] + ] + ] + }, + { + "task_id": "0003019", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0004_cfg_0004_104422070011#230", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003019/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 261, + 247 + ], + [ + 251, + 272 + ], + [ + 289, + 255 + ], + [ + 231, + 239 + ], + [ + 300, + 230 + ], + [ + 241, + 213 + ] + ], + [ + [ + 258, + 269 + ], + [ + 261, + 243 + ], + [ + 229, + 258 + ], + [ + 285, + 280 + ], + [ + 224, + 287 + ], + [ + 283, + 309 + ] + ], + [ + [ + 283, + 268 + ], + [ + 290, + 260 + ], + [ + 266, + 295 + ], + [ + 299, + 243 + ], + [ + 261, + 301 + ], + [ + 293, + 252 + ] + ], + [ + [ + 318, + 267 + ], + [ + 306, + 244 + ], + [ + 331, + 245 + ], + [ + 305, + 288 + ], + [ + 345, + 271 + ], + [ + 317, + 313 + ] + ] + ] + }, + { + "task_id": "0003023", + "id": "RH20T_cfg1_task_0004_user_0001_scene_0005_cfg_0001_035622060973#500", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003023/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 258, + 107 + ], + [ + 243, + 128 + ], + [ + 238, + 94 + ], + [ + 278, + 120 + ], + [ + 253, + 74 + ], + [ + 293, + 100 + ] + ], + [ + [ + 265, + 118 + ], + [ + 279, + 138 + ], + [ + 251, + 132 + ], + [ + 280, + 104 + ], + [ + 238, + 113 + ], + [ + 266, + 84 + ] + ], + [ + [ + 264, + 137 + ], + [ + 250, + 123 + ], + [ + 280, + 140 + ], + [ + 246, + 135 + ], + [ + 294, + 154 + ], + [ + 261, + 150 + ] + ], + [ + [ + 267, + 107 + ], + [ + 272, + 93 + ], + [ + 267, + 87 + ], + [ + 267, + 129 + ], + [ + 262, + 101 + ], + [ + 261, + 141 + ] + ] + ] + }, + { + "task_id": "0003025", + "id": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_038522062547#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003025/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 405, + 47 + ], + [ + 379, + 63 + ], + [ + 416, + 57 + ], + [ + 394, + 38 + ], + [ + 444, + 40 + ], + [ + 419, + 23 + ] + ], + [ + [ + 373, + 44 + ], + [ + 376, + 26 + ], + [ + 400, + 42 + ], + [ + 345, + 47 + ], + [ + 396, + 59 + ], + [ + 344, + 63 + ] + ], + [ + [ + 349, + 46 + ], + [ + 366, + 33 + ], + [ + 365, + 31 + ], + [ + 333, + 63 + ], + [ + 350, + 43 + ], + [ + 319, + 74 + ] + ], + [ + [ + 395, + 53 + ], + [ + 376, + 34 + ], + [ + 405, + 32 + ], + [ + 385, + 75 + ], + [ + 423, + 51 + ], + [ + 404, + 93 + ] + ] + ] + }, + { + "task_id": "0003026", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0001_cfg_0004_038522062288#155", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003026/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 198, + 181 + ], + [ + 192, + 206 + ], + [ + 231, + 186 + ], + [ + 163, + 175 + ], + [ + 239, + 160 + ], + [ + 168, + 147 + ] + ], + [ + [ + 191, + 133 + ], + [ + 200, + 124 + ], + [ + 163, + 148 + ], + [ + 220, + 118 + ], + [ + 151, + 159 + ], + [ + 211, + 127 + ] + ], + [ + [ + 207, + 159 + ], + [ + 199, + 134 + ], + [ + 172, + 159 + ], + [ + 242, + 159 + ], + [ + 182, + 180 + ], + [ + 247, + 180 + ] + ], + [ + [ + 188, + 173 + ], + [ + 214, + 188 + ], + [ + 166, + 199 + ], + [ + 209, + 146 + ], + [ + 136, + 183 + ], + [ + 181, + 128 + ] + ] + ] + }, + { + "task_id": "0003028", + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122063633#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003028/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 215, + 97 + ], + [ + 201, + 80 + ], + [ + 217, + 105 + ], + [ + 213, + 87 + ], + [ + 230, + 121 + ], + [ + 228, + 104 + ] + ], + [ + [ + 173, + 135 + ], + [ + 167, + 120 + ], + [ + 181, + 118 + ], + [ + 165, + 153 + ], + [ + 187, + 132 + ], + [ + 172, + 166 + ] + ], + [ + [ + 191, + 123 + ], + [ + 189, + 102 + ], + [ + 196, + 112 + ], + [ + 185, + 133 + ], + [ + 198, + 131 + ], + [ + 187, + 153 + ] + ], + [ + [ + 189, + 137 + ], + [ + 171, + 126 + ], + [ + 177, + 156 + ], + [ + 201, + 119 + ], + [ + 196, + 167 + ], + [ + 219, + 130 + ] + ] + ] + }, + { + "task_id": "0003031", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_035622060973#101", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003031/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 213, + 80 + ], + [ + 222, + 63 + ], + [ + 202, + 73 + ], + [ + 223, + 86 + ], + [ + 193, + 91 + ], + [ + 214, + 103 + ] + ], + [ + [ + 202, + 83 + ], + [ + 198, + 69 + ], + [ + 188, + 73 + ], + [ + 216, + 93 + ], + [ + 192, + 86 + ], + [ + 219, + 106 + ] + ], + [ + [ + 198, + 91 + ], + [ + 209, + 76 + ], + [ + 181, + 80 + ], + [ + 215, + 102 + ], + [ + 170, + 96 + ], + [ + 204, + 117 + ] + ], + [ + [ + 215, + 104 + ], + [ + 196, + 113 + ], + [ + 206, + 87 + ], + [ + 223, + 121 + ], + [ + 225, + 79 + ], + [ + 240, + 113 + ] + ] + ] + }, + { + "task_id": "0003032", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_035622060973#251", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003032/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 239, + 159 + ], + [ + 236, + 145 + ], + [ + 231, + 143 + ], + [ + 247, + 175 + ], + [ + 234, + 156 + ], + [ + 249, + 188 + ] + ], + [ + [ + 249, + 140 + ], + [ + 264, + 125 + ], + [ + 245, + 129 + ], + [ + 253, + 152 + ], + [ + 231, + 143 + ], + [ + 237, + 166 + ] + ], + [ + [ + 240, + 166 + ], + [ + 220, + 174 + ], + [ + 244, + 181 + ], + [ + 237, + 151 + ], + [ + 263, + 173 + ], + [ + 258, + 143 + ] + ], + [ + [ + 232, + 158 + ], + [ + 227, + 137 + ], + [ + 241, + 150 + ], + [ + 222, + 168 + ], + [ + 246, + 170 + ], + [ + 228, + 189 + ] + ] + ] + }, + { + "task_id": "0003034", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_038522062547#103", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003034/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 361, + 360 + ], + [ + 377, + 333 + ], + [ + 376, + 360 + ], + [ + 348, + 354 + ], + [ + 358, + 360 + ], + [ + 332, + 360 + ] + ], + [ + [ + 354, + 325 + ], + [ + 353, + 323 + ], + [ + 381, + 342 + ], + [ + 329, + 309 + ], + [ + 380, + 343 + ], + [ + 331, + 312 + ] + ], + [ + [ + 333, + 307 + ], + [ + 331, + 308 + ], + [ + 363, + 308 + ], + [ + 303, + 306 + ], + [ + 363, + 307 + ], + [ + 307, + 306 + ] + ], + [ + [ + 320, + 334 + ], + [ + 346, + 324 + ], + [ + 334, + 353 + ], + [ + 306, + 314 + ], + [ + 310, + 360 + ], + [ + 281, + 325 + ] + ] + ] + }, + { + "task_id": "0003036", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_038522062547#222", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003036/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 379, + 310 + ], + [ + 378, + 305 + ], + [ + 405, + 327 + ], + [ + 355, + 294 + ], + [ + 405, + 331 + ], + [ + 357, + 299 + ] + ], + [ + [ + 372, + 293 + ], + [ + 373, + 299 + ], + [ + 396, + 276 + ], + [ + 349, + 309 + ], + [ + 393, + 272 + ], + [ + 350, + 303 + ] + ], + [ + [ + 412, + 313 + ], + [ + 428, + 309 + ], + [ + 435, + 303 + ], + [ + 387, + 324 + ], + [ + 420, + 307 + ], + [ + 374, + 327 + ] + ], + [ + [ + 372, + 317 + ], + [ + 351, + 319 + ], + [ + 363, + 346 + ], + [ + 380, + 291 + ], + [ + 387, + 345 + ], + [ + 403, + 288 + ] + ] + ] + }, + { + "task_id": "0003038", + "id": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_104122063550#94", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003038/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 279, + 152 + ], + [ + 274, + 155 + ], + [ + 276, + 128 + ], + [ + 281, + 176 + ], + [ + 280, + 127 + ], + [ + 285, + 172 + ] + ], + [ + [ + 277, + 143 + ], + [ + 271, + 169 + ], + [ + 278, + 148 + ], + [ + 276, + 139 + ], + [ + 283, + 124 + ], + [ + 282, + 112 + ] + ], + [ + [ + 276, + 135 + ], + [ + 276, + 123 + ], + [ + 300, + 140 + ], + [ + 252, + 129 + ], + [ + 298, + 150 + ], + [ + 253, + 140 + ] + ], + [ + [ + 298, + 133 + ], + [ + 316, + 139 + ], + [ + 304, + 149 + ], + [ + 291, + 116 + ], + [ + 287, + 143 + ], + [ + 273, + 111 + ] + ] + ] + }, + { + "task_id": "0003040", + "id": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_104122063550#118", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003040/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 301, + 78 + ], + [ + 291, + 92 + ], + [ + 283, + 60 + ], + [ + 317, + 94 + ], + [ + 293, + 48 + ], + [ + 326, + 81 + ] + ], + [ + [ + 258, + 105 + ], + [ + 265, + 84 + ], + [ + 247, + 107 + ], + [ + 270, + 102 + ], + [ + 239, + 128 + ], + [ + 262, + 124 + ] + ], + [ + [ + 292, + 73 + ], + [ + 278, + 74 + ], + [ + 313, + 75 + ], + [ + 272, + 72 + ], + [ + 325, + 74 + ], + [ + 285, + 71 + ] + ], + [ + [ + 283, + 98 + ], + [ + 283, + 87 + ], + [ + 305, + 106 + ], + [ + 260, + 90 + ], + [ + 304, + 116 + ], + [ + 261, + 101 + ] + ] + ] + }, + { + "task_id": "0003041", + "id": "RH20T_cfg3_task_0058_user_0010_scene_0006_cfg_0003_104122063550#147", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003041/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 86 + ], + [ + 324, + 78 + ], + [ + 294, + 86 + ], + [ + 315, + 86 + ], + [ + 275, + 94 + ], + [ + 297, + 93 + ] + ], + [ + [ + 283, + 100 + ], + [ + 304, + 94 + ], + [ + 272, + 87 + ], + [ + 293, + 113 + ], + [ + 252, + 94 + ], + [ + 274, + 119 + ] + ], + [ + [ + 297, + 84 + ], + [ + 274, + 82 + ], + [ + 299, + 59 + ], + [ + 296, + 107 + ], + [ + 323, + 61 + ], + [ + 319, + 109 + ] + ], + [ + [ + 283, + 83 + ], + [ + 285, + 71 + ], + [ + 305, + 89 + ], + [ + 260, + 77 + ], + [ + 303, + 100 + ], + [ + 260, + 89 + ] + ] + ] + }, + { + "task_id": "0003042", + "id": "RH20T_cfg1_task_0053_user_0014_scene_0003_cfg_0001_038522063145#168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a clamp'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003042/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 99 + ], + [ + 339, + 82 + ], + [ + 341, + 110 + ], + [ + 311, + 88 + ], + [ + 328, + 128 + ], + [ + 299, + 105 + ] + ], + [ + [ + 338, + 107 + ], + [ + 333, + 88 + ], + [ + 353, + 99 + ], + [ + 322, + 116 + ], + [ + 357, + 117 + ], + [ + 327, + 135 + ] + ], + [ + [ + 349, + 98 + ], + [ + 363, + 101 + ], + [ + 333, + 105 + ], + [ + 364, + 91 + ], + [ + 320, + 101 + ], + [ + 351, + 88 + ] + ], + [ + [ + 333, + 123 + ], + [ + 334, + 139 + ], + [ + 313, + 128 + ], + [ + 352, + 117 + ], + [ + 311, + 111 + ], + [ + 352, + 99 + ] + ] + ] + }, + { + "task_id": "0003043", + "id": "RH20T_cfg3_task_0067_user_0010_scene_0010_cfg_0003_036422060909#177", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button from top to bottom with obstacles'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003043/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 333, + 288 + ], + [ + 336, + 275 + ], + [ + 365, + 279 + ], + [ + 301, + 298 + ], + [ + 361, + 291 + ], + [ + 301, + 310 + ] + ], + [ + [ + 361, + 264 + ], + [ + 381, + 237 + ], + [ + 336, + 246 + ], + [ + 384, + 281 + ], + [ + 315, + 274 + ], + [ + 365, + 308 + ] + ], + [ + [ + 299, + 284 + ], + [ + 316, + 309 + ], + [ + 274, + 294 + ], + [ + 326, + 273 + ], + [ + 257, + 269 + ], + [ + 308, + 245 + ] + ], + [ + [ + 341, + 262 + ], + [ + 308, + 261 + ], + [ + 341, + 228 + ], + [ + 342, + 296 + ], + [ + 374, + 229 + ], + [ + 376, + 297 + ] + ] + ] + }, + { + "task_id": "0003045", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0009_cfg_0005_037522062165#168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003045/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 203, + 252 + ], + [ + 204, + 222 + ], + [ + 216, + 237 + ], + [ + 191, + 266 + ], + [ + 216, + 270 + ], + [ + 190, + 297 + ] + ], + [ + [ + 220, + 224 + ], + [ + 216, + 209 + ], + [ + 203, + 201 + ], + [ + 239, + 249 + ], + [ + 207, + 216 + ], + [ + 241, + 262 + ] + ], + [ + [ + 230, + 213 + ], + [ + 203, + 228 + ], + [ + 220, + 209 + ], + [ + 241, + 218 + ], + [ + 245, + 195 + ], + [ + 268, + 203 + ] + ], + [ + [ + 202, + 248 + ], + [ + 211, + 274 + ], + [ + 164, + 253 + ], + [ + 237, + 244 + ], + [ + 158, + 229 + ], + [ + 228, + 222 + ] + ] + ] + }, + { + "task_id": "0003046", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0009_cfg_0005_037522062165#404", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003046/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 244 + ], + [ + 372, + 253 + ], + [ + 344, + 245 + ], + [ + 350, + 243 + ], + [ + 317, + 236 + ], + [ + 326, + 234 + ] + ], + [ + [ + 315, + 242 + ], + [ + 294, + 234 + ], + [ + 332, + 225 + ], + [ + 298, + 258 + ], + [ + 350, + 233 + ], + [ + 318, + 264 + ] + ], + [ + [ + 336, + 231 + ], + [ + 340, + 217 + ], + [ + 330, + 207 + ], + [ + 343, + 256 + ], + [ + 327, + 221 + ], + [ + 339, + 267 + ] + ], + [ + [ + 329, + 218 + ], + [ + 346, + 221 + ], + [ + 310, + 226 + ], + [ + 351, + 210 + ], + [ + 292, + 223 + ], + [ + 333, + 206 + ] + ] + ] + }, + { + "task_id": "0003047", + "id": "RH20T_cfg1_task_0023_user_0014_scene_0002_cfg_0001_039422060546#100", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003047/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 273, + 148 + ], + [ + 270, + 128 + ], + [ + 267, + 173 + ], + [ + 279, + 125 + ], + [ + 270, + 190 + ], + [ + 281, + 144 + ] + ], + [ + [ + 300, + 158 + ], + [ + 327, + 151 + ], + [ + 309, + 171 + ], + [ + 290, + 143 + ], + [ + 284, + 177 + ], + [ + 263, + 149 + ] + ], + [ + [ + 304, + 154 + ], + [ + 314, + 157 + ], + [ + 285, + 134 + ], + [ + 324, + 174 + ], + [ + 273, + 129 + ], + [ + 314, + 172 + ] + ], + [ + [ + 291, + 132 + ], + [ + 275, + 104 + ], + [ + 278, + 139 + ], + [ + 303, + 125 + ], + [ + 295, + 168 + ], + [ + 317, + 151 + ] + ] + ] + }, + { + "task_id": "0003049", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122063633#536", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003049/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 192, + 171 + ], + [ + 194, + 160 + ], + [ + 213, + 168 + ], + [ + 170, + 174 + ], + [ + 210, + 178 + ], + [ + 168, + 185 + ] + ], + [ + [ + 177, + 151 + ], + [ + 155, + 160 + ], + [ + 173, + 137 + ], + [ + 180, + 163 + ], + [ + 195, + 128 + ], + [ + 200, + 154 + ] + ], + [ + [ + 178, + 182 + ], + [ + 177, + 175 + ], + [ + 163, + 198 + ], + [ + 194, + 166 + ], + [ + 164, + 208 + ], + [ + 196, + 174 + ] + ], + [ + [ + 191, + 166 + ], + [ + 196, + 144 + ], + [ + 197, + 168 + ], + [ + 186, + 163 + ], + [ + 193, + 188 + ], + [ + 181, + 186 + ] + ] + ] + }, + { + "task_id": "0003051", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_037522061512#295", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003051/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 233, + 142 + ], + [ + 223, + 133 + ], + [ + 214, + 155 + ], + [ + 252, + 129 + ], + [ + 224, + 163 + ], + [ + 260, + 138 + ] + ], + [ + [ + 244, + 142 + ], + [ + 247, + 119 + ], + [ + 259, + 142 + ], + [ + 228, + 142 + ], + [ + 256, + 164 + ], + [ + 226, + 165 + ] + ], + [ + [ + 250, + 126 + ], + [ + 269, + 130 + ], + [ + 261, + 114 + ], + [ + 237, + 138 + ], + [ + 243, + 110 + ], + [ + 218, + 133 + ] + ], + [ + [ + 240, + 146 + ], + [ + 229, + 126 + ], + [ + 219, + 156 + ], + [ + 260, + 138 + ], + [ + 230, + 176 + ], + [ + 270, + 157 + ] + ] + ] + }, + { + "task_id": "0003052", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_037522061512#370", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003052/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 237, + 125 + ], + [ + 240, + 102 + ], + [ + 256, + 125 + ], + [ + 216, + 126 + ], + [ + 253, + 146 + ], + [ + 214, + 149 + ] + ], + [ + [ + 246, + 147 + ], + [ + 249, + 124 + ], + [ + 262, + 148 + ], + [ + 228, + 146 + ], + [ + 259, + 169 + ], + [ + 225, + 169 + ] + ], + [ + [ + 251, + 149 + ], + [ + 254, + 153 + ], + [ + 243, + 127 + ], + [ + 259, + 170 + ], + [ + 241, + 124 + ], + [ + 257, + 165 + ] + ], + [ + [ + 252, + 164 + ], + [ + 267, + 147 + ], + [ + 269, + 179 + ], + [ + 235, + 149 + ], + [ + 255, + 195 + ], + [ + 220, + 166 + ] + ] + ] + }, + { + "task_id": "0003054", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_037522061512#563", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003054/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 144 + ], + [ + 280, + 140 + ], + [ + 259, + 122 + ], + [ + 273, + 166 + ], + [ + 246, + 127 + ], + [ + 260, + 170 + ] + ], + [ + [ + 272, + 136 + ], + [ + 274, + 114 + ], + [ + 286, + 136 + ], + [ + 256, + 137 + ], + [ + 285, + 157 + ], + [ + 255, + 160 + ] + ], + [ + [ + 280, + 119 + ], + [ + 294, + 106 + ], + [ + 269, + 126 + ], + [ + 293, + 111 + ], + [ + 255, + 139 + ], + [ + 279, + 124 + ] + ], + [ + [ + 268, + 137 + ], + [ + 279, + 135 + ], + [ + 288, + 134 + ], + [ + 248, + 141 + ], + [ + 278, + 136 + ], + [ + 239, + 143 + ] + ] + ] + }, + { + "task_id": "0003060", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061602#309", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003060/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 460, + 183 + ], + [ + 479, + 179 + ], + [ + 465, + 208 + ], + [ + 454, + 159 + ], + [ + 447, + 211 + ], + [ + 437, + 164 + ] + ], + [ + [ + 471, + 182 + ], + [ + 481, + 172 + ], + [ + 449, + 168 + ], + [ + 495, + 196 + ], + [ + 438, + 177 + ], + [ + 485, + 207 + ] + ], + [ + [ + 458, + 203 + ], + [ + 463, + 205 + ], + [ + 467, + 226 + ], + [ + 449, + 180 + ], + [ + 462, + 224 + ], + [ + 443, + 176 + ] + ], + [ + [ + 475, + 179 + ], + [ + 482, + 169 + ], + [ + 453, + 163 + ], + [ + 496, + 196 + ], + [ + 445, + 174 + ], + [ + 490, + 208 + ] + ] + ] + }, + { + "task_id": "0003061", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061602#379", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003061/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 442, + 203 + ], + [ + 418, + 212 + ], + [ + 439, + 184 + ], + [ + 445, + 222 + ], + [ + 464, + 175 + ], + [ + 469, + 213 + ] + ], + [ + [ + 453, + 181 + ], + [ + 441, + 199 + ], + [ + 442, + 187 + ], + [ + 464, + 175 + ], + [ + 454, + 167 + ], + [ + 476, + 157 + ] + ], + [ + [ + 471, + 167 + ], + [ + 445, + 165 + ], + [ + 469, + 191 + ], + [ + 472, + 141 + ], + [ + 496, + 194 + ], + [ + 500, + 142 + ] + ], + [ + [ + 450, + 179 + ], + [ + 469, + 175 + ], + [ + 453, + 205 + ], + [ + 447, + 154 + ], + [ + 435, + 207 + ], + [ + 430, + 159 + ] + ] + ] + }, + { + "task_id": "0003063", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0003_cfg_0006_104122061602#560", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003063/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 463, + 185 + ], + [ + 440, + 199 + ], + [ + 447, + 168 + ], + [ + 481, + 202 + ], + [ + 469, + 155 + ], + [ + 506, + 187 + ] + ], + [ + [ + 454, + 201 + ], + [ + 474, + 196 + ], + [ + 460, + 228 + ], + [ + 448, + 175 + ], + [ + 441, + 231 + ], + [ + 430, + 181 + ] + ], + [ + [ + 480, + 186 + ], + [ + 480, + 208 + ], + [ + 505, + 197 + ], + [ + 457, + 176 + ], + [ + 506, + 173 + ], + [ + 457, + 154 + ] + ], + [ + [ + 472, + 219 + ], + [ + 470, + 235 + ], + [ + 477, + 196 + ], + [ + 467, + 240 + ], + [ + 478, + 182 + ], + [ + 468, + 225 + ] + ] + ] + }, + { + "task_id": "0003064", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_038522062288#191", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003064/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 72, + 231 + ], + [ + 62, + 212 + ], + [ + 53, + 215 + ], + [ + 92, + 248 + ], + [ + 64, + 232 + ], + [ + 101, + 264 + ] + ], + [ + [ + 36, + 225 + ], + [ + 16, + 241 + ], + [ + 9, + 237 + ], + [ + 60, + 216 + ], + [ + 31, + 219 + ], + [ + 79, + 200 + ] + ], + [ + [ + 99, + 218 + ], + [ + 86, + 232 + ], + [ + 112, + 192 + ], + [ + 86, + 243 + ], + [ + 122, + 181 + ], + [ + 98, + 229 + ] + ], + [ + [ + 82, + 250 + ], + [ + 70, + 274 + ], + [ + 81, + 267 + ], + [ + 82, + 234 + ], + [ + 94, + 242 + ], + [ + 93, + 211 + ] + ] + ] + }, + { + "task_id": "0003066", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_038522062288#747", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003066/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 104, + 253 + ], + [ + 95, + 237 + ], + [ + 77, + 241 + ], + [ + 131, + 266 + ], + [ + 86, + 257 + ], + [ + 138, + 280 + ] + ], + [ + [ + 116, + 261 + ], + [ + 97, + 240 + ], + [ + 97, + 262 + ], + [ + 134, + 260 + ], + [ + 118, + 284 + ], + [ + 153, + 280 + ] + ], + [ + [ + 114, + 267 + ], + [ + 89, + 291 + ], + [ + 135, + 280 + ], + [ + 94, + 255 + ], + [ + 158, + 256 + ], + [ + 118, + 233 + ] + ], + [ + [ + 89, + 205 + ], + [ + 113, + 212 + ], + [ + 73, + 183 + ], + [ + 103, + 226 + ], + [ + 49, + 177 + ], + [ + 80, + 219 + ] + ] + ] + }, + { + "task_id": "0003067", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_038522062288#918", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003067/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 111, + 243 + ], + [ + 94, + 218 + ], + [ + 85, + 241 + ], + [ + 133, + 245 + ], + [ + 103, + 270 + ], + [ + 149, + 269 + ] + ], + [ + [ + 121, + 270 + ], + [ + 87, + 284 + ], + [ + 117, + 243 + ], + [ + 126, + 299 + ], + [ + 149, + 230 + ], + [ + 161, + 284 + ] + ], + [ + [ + 125, + 256 + ], + [ + 114, + 265 + ], + [ + 91, + 241 + ], + [ + 157, + 270 + ], + [ + 101, + 230 + ], + [ + 169, + 262 + ] + ], + [ + [ + 130, + 272 + ], + [ + 122, + 256 + ], + [ + 113, + 250 + ], + [ + 148, + 295 + ], + [ + 122, + 264 + ], + [ + 155, + 308 + ] + ] + ] + }, + { + "task_id": "0003068", + "id": "RH20T_cfg4_task_0067_user_0015_scene_0006_cfg_0004_104122062823#143", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button from top to bottom with obstacles'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003068/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 107 + ], + [ + 408, + 146 + ], + [ + 409, + 81 + ], + [ + 404, + 129 + ], + [ + 407, + 43 + ], + [ + 402, + 95 + ] + ], + [ + [ + 387, + 98 + ], + [ + 358, + 88 + ], + [ + 363, + 102 + ], + [ + 409, + 94 + ], + [ + 396, + 114 + ], + [ + 441, + 104 + ] + ], + [ + [ + 420, + 109 + ], + [ + 423, + 75 + ], + [ + 463, + 110 + ], + [ + 378, + 107 + ], + [ + 457, + 139 + ], + [ + 378, + 136 + ] + ], + [ + [ + 409, + 135 + ], + [ + 406, + 170 + ], + [ + 451, + 140 + ], + [ + 368, + 130 + ], + [ + 452, + 108 + ], + [ + 372, + 99 + ] + ] + ] + }, + { + "task_id": "0003069", + "id": "RH20T_cfg1_task_0004_user_0014_scene_0004_cfg_0001_039422060546#98", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003069/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 301, + 203 + ], + [ + 297, + 184 + ], + [ + 302, + 232 + ], + [ + 301, + 176 + ], + [ + 305, + 248 + ], + [ + 304, + 194 + ] + ], + [ + [ + 290, + 194 + ], + [ + 290, + 227 + ], + [ + 256, + 194 + ], + [ + 323, + 193 + ], + [ + 256, + 161 + ], + [ + 323, + 160 + ] + ], + [ + [ + 333, + 190 + ], + [ + 319, + 162 + ], + [ + 335, + 197 + ], + [ + 332, + 183 + ], + [ + 350, + 226 + ], + [ + 345, + 209 + ] + ], + [ + [ + 292, + 201 + ], + [ + 272, + 225 + ], + [ + 312, + 224 + ], + [ + 273, + 180 + ], + [ + 334, + 199 + ], + [ + 293, + 157 + ] + ] + ] + }, + { + "task_id": "0003075", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104122062295#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003075/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 63 + ], + [ + 343, + 64 + ], + [ + 360, + 93 + ], + [ + 280, + 35 + ], + [ + 333, + 95 + ], + [ + 247, + 30 + ] + ], + [ + [ + 257, + 81 + ], + [ + 271, + 33 + ], + [ + 222, + 57 + ], + [ + 295, + 107 + ], + [ + 212, + 99 + ], + [ + 280, + 150 + ] + ], + [ + [ + 224, + 38 + ], + [ + 273, + 28 + ], + [ + 232, + 76 + ], + [ + 215, + 0 + ], + [ + 185, + 86 + ], + [ + 158, + 4 + ] + ], + [ + [ + 232, + 50 + ], + [ + 266, + 87 + ], + [ + 273, + 13 + ], + [ + 192, + 86 + ], + [ + 237, + 0 + ], + [ + 150, + 47 + ] + ] + ] + }, + { + "task_id": "0003077", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104122062295#198", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003077/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 255, + 82 + ], + [ + 210, + 97 + ], + [ + 251, + 126 + ], + [ + 260, + 32 + ], + [ + 288, + 111 + ], + [ + 302, + 22 + ] + ], + [ + [ + 244, + 133 + ], + [ + 245, + 152 + ], + [ + 193, + 101 + ], + [ + 291, + 163 + ], + [ + 198, + 89 + ], + [ + 285, + 146 + ] + ], + [ + [ + 306, + 149 + ], + [ + 289, + 190 + ], + [ + 250, + 127 + ], + [ + 361, + 171 + ], + [ + 264, + 76 + ], + [ + 385, + 125 + ] + ], + [ + [ + 258, + 124 + ], + [ + 272, + 76 + ], + [ + 218, + 92 + ], + [ + 301, + 158 + ], + [ + 210, + 133 + ], + [ + 286, + 197 + ] + ] + ] + }, + { + "task_id": "0003082", + "id": "RH20T_cfg4_task_0081_user_0007_scene_0004_cfg_0004_104122062823#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003082/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 345, + 93 + ], + [ + 361, + 91 + ], + [ + 367, + 67 + ], + [ + 325, + 117 + ], + [ + 349, + 67 + ], + [ + 306, + 121 + ] + ], + [ + [ + 384, + 87 + ], + [ + 379, + 109 + ], + [ + 350, + 78 + ], + [ + 417, + 97 + ], + [ + 356, + 58 + ], + [ + 420, + 76 + ] + ], + [ + [ + 373, + 93 + ], + [ + 371, + 65 + ], + [ + 407, + 93 + ], + [ + 341, + 92 + ], + [ + 407, + 118 + ], + [ + 345, + 117 + ] + ], + [ + [ + 399, + 108 + ], + [ + 395, + 131 + ], + [ + 430, + 114 + ], + [ + 368, + 102 + ], + [ + 432, + 93 + ], + [ + 372, + 82 + ] + ] + ] + }, + { + "task_id": "0003083", + "id": "RH20T_cfg4_task_0081_user_0007_scene_0004_cfg_0004_104122062823#349", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003083/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 324, + 93 + ], + [ + 352, + 108 + ], + [ + 342, + 74 + ], + [ + 308, + 109 + ], + [ + 311, + 56 + ], + [ + 279, + 94 + ] + ], + [ + [ + 297, + 143 + ], + [ + 271, + 136 + ], + [ + 300, + 110 + ], + [ + 294, + 176 + ], + [ + 324, + 117 + ], + [ + 319, + 181 + ] + ], + [ + [ + 321, + 94 + ], + [ + 348, + 108 + ], + [ + 294, + 112 + ], + [ + 347, + 78 + ], + [ + 271, + 98 + ], + [ + 323, + 66 + ] + ], + [ + [ + 307, + 121 + ], + [ + 301, + 93 + ], + [ + 332, + 103 + ], + [ + 281, + 139 + ], + [ + 336, + 128 + ], + [ + 288, + 163 + ] + ] + ] + }, + { + "task_id": "0003086", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122063633#675", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003086/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 251, + 167 + ], + [ + 272, + 169 + ], + [ + 250, + 188 + ], + [ + 253, + 147 + ], + [ + 229, + 186 + ], + [ + 232, + 145 + ] + ], + [ + [ + 237, + 153 + ], + [ + 242, + 143 + ], + [ + 238, + 171 + ], + [ + 236, + 135 + ], + [ + 233, + 182 + ], + [ + 232, + 146 + ] + ], + [ + [ + 225, + 147 + ], + [ + 240, + 138 + ], + [ + 236, + 165 + ], + [ + 214, + 129 + ], + [ + 221, + 174 + ], + [ + 200, + 139 + ] + ], + [ + [ + 215, + 176 + ], + [ + 229, + 177 + ], + [ + 201, + 164 + ], + [ + 228, + 186 + ], + [ + 187, + 164 + ], + [ + 214, + 186 + ] + ] + ] + }, + { + "task_id": "0003088", + "id": "RH20T_cfg7_task_0015_user_0020_scene_0007_cfg_0007_104122060811#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003088/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 342, + 141 + ], + [ + 344, + 126 + ], + [ + 327, + 140 + ], + [ + 358, + 143 + ], + [ + 325, + 155 + ], + [ + 356, + 159 + ] + ], + [ + [ + 346, + 132 + ], + [ + 343, + 141 + ], + [ + 354, + 122 + ], + [ + 339, + 143 + ], + [ + 357, + 114 + ], + [ + 343, + 135 + ] + ], + [ + [ + 326, + 133 + ], + [ + 322, + 119 + ], + [ + 324, + 138 + ], + [ + 327, + 129 + ], + [ + 327, + 152 + ], + [ + 331, + 144 + ] + ], + [ + [ + 333, + 144 + ], + [ + 341, + 146 + ], + [ + 320, + 150 + ], + [ + 345, + 139 + ], + [ + 311, + 148 + ], + [ + 337, + 137 + ] + ] + ] + }, + { + "task_id": "0003089", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_104122063678#341", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003089/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 367, + 243 + ], + [ + 379, + 271 + ], + [ + 365, + 247 + ], + [ + 369, + 239 + ], + [ + 353, + 217 + ], + [ + 358, + 212 + ] + ], + [ + [ + 364, + 249 + ], + [ + 337, + 230 + ], + [ + 348, + 268 + ], + [ + 381, + 228 + ], + [ + 374, + 287 + ], + [ + 409, + 247 + ] + ], + [ + [ + 372, + 236 + ], + [ + 401, + 231 + ], + [ + 359, + 213 + ], + [ + 384, + 258 + ], + [ + 332, + 218 + ], + [ + 358, + 262 + ] + ], + [ + [ + 343, + 241 + ], + [ + 352, + 221 + ], + [ + 348, + 267 + ], + [ + 339, + 217 + ], + [ + 339, + 285 + ], + [ + 331, + 235 + ] + ] + ] + }, + { + "task_id": "0003091", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_104122063678#690", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003091/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 336, + 212 + ], + [ + 322, + 222 + ], + [ + 345, + 233 + ], + [ + 328, + 191 + ], + [ + 360, + 223 + ], + [ + 342, + 180 + ] + ], + [ + [ + 356, + 192 + ], + [ + 362, + 177 + ], + [ + 359, + 211 + ], + [ + 353, + 175 + ], + [ + 353, + 225 + ], + [ + 347, + 189 + ] + ], + [ + [ + 342, + 188 + ], + [ + 334, + 190 + ], + [ + 342, + 166 + ], + [ + 342, + 211 + ], + [ + 350, + 165 + ], + [ + 350, + 209 + ] + ], + [ + [ + 364, + 222 + ], + [ + 375, + 231 + ], + [ + 347, + 237 + ], + [ + 381, + 207 + ], + [ + 337, + 228 + ], + [ + 370, + 199 + ] + ] + ] + }, + { + "task_id": "0003092", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_104122063678#974", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003092/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 152 + ], + [ + 330, + 168 + ], + [ + 338, + 138 + ], + [ + 293, + 166 + ], + [ + 322, + 121 + ], + [ + 276, + 150 + ] + ], + [ + [ + 312, + 179 + ], + [ + 317, + 159 + ], + [ + 313, + 200 + ], + [ + 310, + 161 + ], + [ + 307, + 219 + ], + [ + 306, + 180 + ] + ], + [ + [ + 320, + 198 + ], + [ + 308, + 222 + ], + [ + 344, + 206 + ], + [ + 298, + 189 + ], + [ + 356, + 182 + ], + [ + 310, + 167 + ] + ], + [ + [ + 324, + 169 + ], + [ + 297, + 169 + ], + [ + 317, + 158 + ], + [ + 331, + 180 + ], + [ + 342, + 158 + ], + [ + 358, + 179 + ] + ] + ] + }, + { + "task_id": "0003094", + "id": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_038522062547#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003094/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 361, + 136 + ], + [ + 342, + 145 + ], + [ + 351, + 142 + ], + [ + 371, + 130 + ], + [ + 373, + 132 + ], + [ + 391, + 120 + ] + ], + [ + [ + 374, + 129 + ], + [ + 383, + 116 + ], + [ + 396, + 126 + ], + [ + 350, + 132 + ], + [ + 386, + 138 + ], + [ + 342, + 143 + ] + ], + [ + [ + 389, + 127 + ], + [ + 394, + 104 + ], + [ + 383, + 134 + ], + [ + 396, + 119 + ], + [ + 378, + 156 + ], + [ + 391, + 143 + ] + ], + [ + [ + 371, + 122 + ], + [ + 359, + 99 + ], + [ + 390, + 112 + ], + [ + 351, + 132 + ], + [ + 403, + 134 + ], + [ + 365, + 156 + ] + ] + ] + }, + { + "task_id": "0003095", + "id": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_038522062547#628", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003095/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 400, + 98 + ], + [ + 404, + 91 + ], + [ + 415, + 117 + ], + [ + 385, + 79 + ], + [ + 411, + 123 + ], + [ + 382, + 87 + ] + ], + [ + [ + 425, + 110 + ], + [ + 447, + 101 + ], + [ + 416, + 96 + ], + [ + 433, + 124 + ], + [ + 393, + 105 + ], + [ + 411, + 133 + ] + ], + [ + [ + 422, + 107 + ], + [ + 439, + 96 + ], + [ + 420, + 86 + ], + [ + 423, + 126 + ], + [ + 401, + 98 + ], + [ + 406, + 137 + ] + ], + [ + [ + 385, + 103 + ], + [ + 367, + 118 + ], + [ + 375, + 89 + ], + [ + 394, + 116 + ], + [ + 394, + 73 + ], + [ + 412, + 101 + ] + ] + ] + }, + { + "task_id": "0003097", + "id": "RH20T_cfg4_task_0067_user_0015_scene_0006_cfg_0004_104122063550#77", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button from top to bottom with obstacles'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003097/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 298, + 360 + ], + [ + 322, + 356 + ], + [ + 293, + 340 + ], + [ + 305, + 360 + ], + [ + 272, + 356 + ], + [ + 282, + 360 + ] + ], + [ + [ + 346, + 321 + ], + [ + 341, + 355 + ], + [ + 343, + 309 + ], + [ + 349, + 334 + ], + [ + 347, + 277 + ], + [ + 354, + 297 + ] + ], + [ + [ + 308, + 324 + ], + [ + 304, + 324 + ], + [ + 328, + 353 + ], + [ + 289, + 296 + ], + [ + 330, + 351 + ], + [ + 293, + 298 + ] + ], + [ + [ + 336, + 309 + ], + [ + 309, + 328 + ], + [ + 321, + 299 + ], + [ + 349, + 318 + ], + [ + 350, + 279 + ], + [ + 375, + 299 + ] + ] + ] + }, + { + "task_id": "0003099", + "id": "RH20T_cfg3_task_0092_user_0010_scene_0003_cfg_0003_104122062823#187", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the hands of a clock'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003099/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 287, + 150 + ], + [ + 303, + 131 + ], + [ + 315, + 138 + ], + [ + 262, + 162 + ], + [ + 297, + 161 + ], + [ + 244, + 183 + ] + ], + [ + [ + 272, + 143 + ], + [ + 273, + 113 + ], + [ + 305, + 152 + ], + [ + 241, + 133 + ], + [ + 303, + 179 + ], + [ + 242, + 160 + ] + ], + [ + [ + 259, + 105 + ], + [ + 248, + 126 + ], + [ + 291, + 105 + ], + [ + 227, + 106 + ], + [ + 300, + 85 + ], + [ + 238, + 87 + ] + ], + [ + [ + 243, + 146 + ], + [ + 223, + 115 + ], + [ + 213, + 158 + ], + [ + 275, + 134 + ], + [ + 231, + 186 + ], + [ + 292, + 164 + ] + ] + ] + }, + { + "task_id": "0003100", + "id": "RH20T_cfg3_task_0038_user_0010_scene_0010_cfg_0003_104122062295#163", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003100/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 196, + 153 + ], + [ + 187, + 107 + ], + [ + 176, + 124 + ], + [ + 218, + 186 + ], + [ + 184, + 163 + ], + [ + 223, + 225 + ] + ], + [ + [ + 169, + 88 + ], + [ + 184, + 127 + ], + [ + 223, + 80 + ], + [ + 109, + 96 + ], + [ + 207, + 48 + ], + [ + 98, + 60 + ] + ], + [ + [ + 139, + 159 + ], + [ + 85, + 159 + ], + [ + 124, + 182 + ], + [ + 152, + 139 + ], + [ + 183, + 181 + ], + [ + 200, + 139 + ] + ], + [ + [ + 172, + 131 + ], + [ + 142, + 140 + ], + [ + 226, + 139 + ], + [ + 119, + 123 + ], + [ + 244, + 131 + ], + [ + 148, + 117 + ] + ] + ] + }, + { + "task_id": "0003101", + "id": "RH20T_cfg6_task_0082_user_0020_scene_0001_cfg_0006_104122063633#506", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003101/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 224, + 137 + ], + [ + 230, + 124 + ], + [ + 222, + 154 + ], + [ + 227, + 119 + ], + [ + 216, + 168 + ], + [ + 220, + 133 + ] + ], + [ + [ + 208, + 132 + ], + [ + 210, + 126 + ], + [ + 187, + 144 + ], + [ + 229, + 120 + ], + [ + 183, + 151 + ], + [ + 227, + 126 + ] + ], + [ + [ + 203, + 141 + ], + [ + 184, + 150 + ], + [ + 192, + 120 + ], + [ + 214, + 162 + ], + [ + 211, + 110 + ], + [ + 233, + 153 + ] + ], + [ + [ + 237, + 162 + ], + [ + 243, + 142 + ], + [ + 251, + 171 + ], + [ + 222, + 152 + ], + [ + 245, + 191 + ], + [ + 215, + 173 + ] + ] + ] + }, + { + "task_id": "0003102", + "id": "RH20T_cfg6_task_0007_user_0014_scene_0007_cfg_0006_037522061512#163", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003102/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 89 + ], + [ + 303, + 68 + ], + [ + 308, + 86 + ], + [ + 289, + 91 + ], + [ + 303, + 105 + ], + [ + 285, + 112 + ] + ], + [ + [ + 279, + 101 + ], + [ + 263, + 106 + ], + [ + 268, + 84 + ], + [ + 289, + 119 + ], + [ + 284, + 80 + ], + [ + 304, + 113 + ] + ], + [ + [ + 279, + 68 + ], + [ + 296, + 55 + ], + [ + 270, + 60 + ], + [ + 288, + 76 + ], + [ + 254, + 73 + ], + [ + 270, + 90 + ] + ], + [ + [ + 298, + 70 + ], + [ + 285, + 63 + ], + [ + 295, + 90 + ], + [ + 301, + 49 + ], + [ + 309, + 98 + ], + [ + 316, + 56 + ] + ] + ] + }, + { + "task_id": "0003105", + "id": "RH20T_cfg3_task_0032_user_0011_scene_0002_cfg_0003_104122062295#225", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003105/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 378, + 360 + ], + [ + 358, + 352 + ], + [ + 426, + 360 + ], + [ + 327, + 360 + ], + [ + 456, + 360 + ], + [ + 346, + 360 + ] + ], + [ + [ + 360, + 324 + ], + [ + 368, + 313 + ], + [ + 384, + 287 + ], + [ + 335, + 360 + ], + [ + 376, + 297 + ], + [ + 322, + 360 + ] + ], + [ + [ + 401, + 326 + ], + [ + 413, + 299 + ], + [ + 397, + 282 + ], + [ + 407, + 360 + ], + [ + 388, + 306 + ], + [ + 395, + 360 + ] + ], + [ + [ + 362, + 291 + ], + [ + 338, + 267 + ], + [ + 404, + 287 + ], + [ + 324, + 294 + ], + [ + 424, + 309 + ], + [ + 346, + 314 + ] + ] + ] + }, + { + "task_id": "0003106", + "id": "RH20T_cfg7_task_0017_user_0014_scene_0002_cfg_0007_104122063633#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the pen into the pen holder'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003106/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 220, + 233 + ], + [ + 215, + 209 + ], + [ + 240, + 227 + ], + [ + 201, + 239 + ], + [ + 246, + 251 + ], + [ + 206, + 262 + ] + ], + [ + [ + 199, + 227 + ], + [ + 201, + 206 + ], + [ + 205, + 215 + ], + [ + 192, + 241 + ], + [ + 203, + 235 + ], + [ + 190, + 262 + ] + ], + [ + [ + 212, + 203 + ], + [ + 234, + 194 + ], + [ + 200, + 187 + ], + [ + 222, + 217 + ], + [ + 177, + 197 + ], + [ + 201, + 226 + ] + ], + [ + [ + 194, + 241 + ], + [ + 208, + 246 + ], + [ + 190, + 219 + ], + [ + 198, + 265 + ], + [ + 174, + 212 + ], + [ + 182, + 261 + ] + ] + ] + }, + { + "task_id": "0003107", + "id": "RH20T_cfg1_task_0015_user_0008_scene_0008_cfg_0001_819112071037#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003107/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 115, + 63 + ], + [ + 93, + 44 + ], + [ + 76, + 0 + ], + [ + 145, + 160 + ], + [ + 103, + 0 + ], + [ + 178, + 201 + ] + ], + [ + [ + 215, + 116 + ], + [ + 142, + 115 + ], + [ + 163, + 11 + ], + [ + 255, + 195 + ], + [ + 264, + 0 + ], + [ + 339, + 205 + ] + ], + [ + [ + 0, + 106 + ], + [ + 3, + 102 + ], + [ + 24, + 32 + ], + [ + 0, + 206 + ], + [ + 0, + 41 + ], + [ + 0, + 195 + ] + ], + [ + [ + 109, + 169 + ], + [ + 99, + 62 + ], + [ + 8, + 208 + ], + [ + 176, + 144 + ], + [ + 28, + 318 + ], + [ + 179, + 223 + ] + ] + ] + }, + { + "task_id": "0003108", + "id": "RH20T_cfg5_task_0058_user_0007_scene_0004_cfg_0005_104122063678#292", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003108/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 331, + 196 + ], + [ + 349, + 183 + ], + [ + 314, + 187 + ], + [ + 347, + 205 + ], + [ + 297, + 201 + ], + [ + 330, + 217 + ] + ], + [ + [ + 350, + 211 + ], + [ + 354, + 190 + ], + [ + 355, + 225 + ], + [ + 345, + 199 + ], + [ + 351, + 246 + ], + [ + 341, + 219 + ] + ], + [ + [ + 344, + 230 + ], + [ + 321, + 220 + ], + [ + 350, + 209 + ], + [ + 337, + 252 + ], + [ + 372, + 219 + ], + [ + 360, + 262 + ] + ], + [ + [ + 361, + 220 + ], + [ + 359, + 215 + ], + [ + 366, + 243 + ], + [ + 356, + 198 + ], + [ + 368, + 250 + ], + [ + 357, + 202 + ] + ] + ] + }, + { + "task_id": "0003110", + "id": "RH20T_cfg5_task_0058_user_0007_scene_0004_cfg_0005_104122063678#440", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003110/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 310, + 212 + ], + [ + 313, + 190 + ], + [ + 314, + 225 + ], + [ + 306, + 201 + ], + [ + 312, + 247 + ], + [ + 304, + 221 + ] + ], + [ + [ + 338, + 202 + ], + [ + 340, + 185 + ], + [ + 346, + 220 + ], + [ + 329, + 183 + ], + [ + 344, + 238 + ], + [ + 326, + 201 + ] + ], + [ + [ + 293, + 200 + ], + [ + 267, + 201 + ], + [ + 298, + 182 + ], + [ + 288, + 217 + ], + [ + 323, + 182 + ], + [ + 312, + 216 + ] + ], + [ + [ + 294, + 241 + ], + [ + 283, + 262 + ], + [ + 317, + 254 + ], + [ + 272, + 229 + ], + [ + 327, + 234 + ], + [ + 283, + 210 + ] + ] + ] + }, + { + "task_id": "0003111", + "id": "RH20T_cfg5_task_0058_user_0007_scene_0004_cfg_0005_104122063678#523", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003111/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 281, + 202 + ], + [ + 258, + 199 + ], + [ + 284, + 178 + ], + [ + 277, + 227 + ], + [ + 308, + 181 + ], + [ + 300, + 231 + ] + ], + [ + [ + 279, + 205 + ], + [ + 255, + 195 + ], + [ + 288, + 195 + ], + [ + 271, + 213 + ], + [ + 313, + 204 + ], + [ + 293, + 222 + ] + ], + [ + [ + 291, + 212 + ], + [ + 294, + 189 + ], + [ + 295, + 225 + ], + [ + 288, + 200 + ], + [ + 292, + 247 + ], + [ + 285, + 221 + ] + ], + [ + [ + 310, + 228 + ], + [ + 315, + 202 + ], + [ + 303, + 226 + ], + [ + 319, + 230 + ], + [ + 299, + 250 + ], + [ + 314, + 257 + ] + ] + ] + }, + { + "task_id": "0003112", + "id": "RH20T_cfg7_task_0007_user_0014_scene_0005_cfg_0007_037522061512#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003112/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 252, + 114 + ], + [ + 240, + 97 + ], + [ + 243, + 118 + ], + [ + 261, + 110 + ], + [ + 255, + 136 + ], + [ + 272, + 127 + ] + ], + [ + [ + 256, + 127 + ], + [ + 276, + 138 + ], + [ + 244, + 144 + ], + [ + 268, + 111 + ], + [ + 224, + 133 + ], + [ + 250, + 100 + ] + ], + [ + [ + 245, + 105 + ], + [ + 229, + 93 + ], + [ + 231, + 121 + ], + [ + 259, + 89 + ], + [ + 248, + 134 + ], + [ + 276, + 101 + ] + ], + [ + [ + 256, + 122 + ], + [ + 252, + 105 + ], + [ + 267, + 109 + ], + [ + 244, + 135 + ], + [ + 270, + 125 + ], + [ + 247, + 151 + ] + ] + ] + }, + { + "task_id": "0003115", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0003_cfg_0006_037522061512#233", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003115/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 236, + 158 + ], + [ + 245, + 140 + ], + [ + 250, + 177 + ], + [ + 221, + 137 + ], + [ + 241, + 196 + ], + [ + 210, + 156 + ] + ], + [ + [ + 202, + 133 + ], + [ + 178, + 115 + ], + [ + 190, + 139 + ], + [ + 216, + 126 + ], + [ + 212, + 155 + ], + [ + 239, + 143 + ] + ], + [ + [ + 216, + 135 + ], + [ + 201, + 122 + ], + [ + 229, + 112 + ], + [ + 202, + 158 + ], + [ + 241, + 124 + ], + [ + 215, + 169 + ] + ], + [ + [ + 206, + 112 + ], + [ + 197, + 100 + ], + [ + 180, + 112 + ], + [ + 231, + 112 + ], + [ + 189, + 124 + ], + [ + 241, + 124 + ] + ] + ] + }, + { + "task_id": "0003116", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0003_cfg_0006_037522061512#572", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003116/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 275, + 81 + ], + [ + 272, + 62 + ], + [ + 287, + 68 + ], + [ + 263, + 95 + ], + [ + 290, + 85 + ], + [ + 267, + 113 + ] + ], + [ + [ + 271, + 91 + ], + [ + 248, + 89 + ], + [ + 266, + 109 + ], + [ + 277, + 72 + ], + [ + 287, + 112 + ], + [ + 299, + 75 + ] + ], + [ + [ + 271, + 105 + ], + [ + 289, + 90 + ], + [ + 273, + 108 + ], + [ + 269, + 102 + ], + [ + 256, + 123 + ], + [ + 249, + 117 + ] + ], + [ + [ + 291, + 60 + ], + [ + 310, + 61 + ], + [ + 278, + 74 + ], + [ + 303, + 47 + ], + [ + 260, + 73 + ], + [ + 284, + 46 + ] + ] + ] + }, + { + "task_id": "0003117", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0003_cfg_0006_037522061512#756", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003117/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 224, + 111 + ], + [ + 213, + 96 + ], + [ + 238, + 90 + ], + [ + 211, + 132 + ], + [ + 247, + 105 + ], + [ + 222, + 145 + ] + ], + [ + [ + 209, + 107 + ], + [ + 223, + 129 + ], + [ + 229, + 94 + ], + [ + 189, + 121 + ], + [ + 215, + 71 + ], + [ + 175, + 99 + ] + ], + [ + [ + 257, + 116 + ], + [ + 235, + 109 + ], + [ + 262, + 91 + ], + [ + 251, + 140 + ], + [ + 285, + 98 + ], + [ + 274, + 147 + ] + ], + [ + [ + 235, + 131 + ], + [ + 214, + 125 + ], + [ + 249, + 135 + ], + [ + 220, + 127 + ], + [ + 270, + 142 + ], + [ + 242, + 134 + ] + ] + ] + }, + { + "task_id": "0003120", + "id": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122061018#294", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the pen into the pen holder'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003120/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 391, + 153 + ], + [ + 414, + 145 + ], + [ + 377, + 157 + ], + [ + 407, + 149 + ], + [ + 355, + 165 + ], + [ + 382, + 159 + ] + ], + [ + [ + 392, + 168 + ], + [ + 420, + 155 + ], + [ + 380, + 141 + ], + [ + 404, + 195 + ], + [ + 353, + 154 + ], + [ + 376, + 208 + ] + ], + [ + [ + 380, + 143 + ], + [ + 386, + 171 + ], + [ + 408, + 137 + ], + [ + 352, + 150 + ], + [ + 402, + 108 + ], + [ + 345, + 122 + ] + ], + [ + [ + 391, + 141 + ], + [ + 382, + 115 + ], + [ + 378, + 146 + ], + [ + 406, + 136 + ], + [ + 387, + 171 + ], + [ + 416, + 165 + ] + ] + ] + }, + { + "task_id": "0003123", + "id": "RH20T_cfg5_task_0083_user_0007_scene_0005_cfg_0005_104122063678#342", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Swat with a flyswatter'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003123/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 361, + 153 + ], + [ + 342, + 151 + ], + [ + 365, + 138 + ], + [ + 357, + 168 + ], + [ + 384, + 139 + ], + [ + 377, + 169 + ] + ], + [ + [ + 385, + 143 + ], + [ + 396, + 144 + ], + [ + 383, + 122 + ], + [ + 388, + 164 + ], + [ + 372, + 121 + ], + [ + 377, + 163 + ] + ], + [ + [ + 363, + 146 + ], + [ + 377, + 154 + ], + [ + 352, + 138 + ], + [ + 373, + 153 + ], + [ + 340, + 131 + ], + [ + 361, + 146 + ] + ], + [ + [ + 365, + 147 + ], + [ + 370, + 132 + ], + [ + 345, + 146 + ], + [ + 384, + 149 + ], + [ + 340, + 160 + ], + [ + 379, + 163 + ] + ] + ] + }, + { + "task_id": "0003126", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_104422070044#271", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003126/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 421, + 159 + ], + [ + 402, + 160 + ], + [ + 422, + 178 + ], + [ + 421, + 140 + ], + [ + 441, + 178 + ], + [ + 441, + 139 + ] + ], + [ + [ + 402, + 186 + ], + [ + 403, + 181 + ], + [ + 389, + 199 + ], + [ + 416, + 172 + ], + [ + 388, + 205 + ], + [ + 417, + 176 + ] + ], + [ + [ + 392, + 188 + ], + [ + 388, + 186 + ], + [ + 402, + 171 + ], + [ + 382, + 204 + ], + [ + 405, + 173 + ], + [ + 385, + 205 + ] + ], + [ + [ + 409, + 184 + ], + [ + 408, + 173 + ], + [ + 412, + 200 + ], + [ + 406, + 169 + ], + [ + 414, + 210 + ], + [ + 408, + 180 + ] + ] + ] + }, + { + "task_id": "0003128", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0001_cfg_0006_104122064161#181", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003128/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 398, + 127 + ], + [ + 415, + 116 + ], + [ + 391, + 142 + ], + [ + 404, + 113 + ], + [ + 376, + 152 + ], + [ + 388, + 123 + ] + ], + [ + [ + 391, + 132 + ], + [ + 394, + 114 + ], + [ + 392, + 146 + ], + [ + 389, + 119 + ], + [ + 389, + 163 + ], + [ + 386, + 136 + ] + ], + [ + [ + 410, + 132 + ], + [ + 432, + 133 + ], + [ + 409, + 113 + ], + [ + 411, + 151 + ], + [ + 387, + 112 + ], + [ + 389, + 150 + ] + ], + [ + [ + 401, + 142 + ], + [ + 390, + 145 + ], + [ + 412, + 124 + ], + [ + 390, + 159 + ], + [ + 424, + 119 + ], + [ + 402, + 156 + ] + ] + ] + }, + { + "task_id": "0003131", + "id": "RH20T_cfg7_task_0025_user_0020_scene_0008_cfg_0007_037522061512#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold a piece of paper'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003131/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 263, + 172 + ], + [ + 279, + 186 + ], + [ + 247, + 186 + ], + [ + 278, + 159 + ], + [ + 231, + 172 + ], + [ + 263, + 146 + ] + ], + [ + [ + 262, + 151 + ], + [ + 272, + 170 + ], + [ + 272, + 147 + ], + [ + 254, + 154 + ], + [ + 262, + 126 + ], + [ + 245, + 135 + ] + ], + [ + [ + 239, + 147 + ], + [ + 240, + 129 + ], + [ + 216, + 148 + ], + [ + 262, + 145 + ], + [ + 216, + 164 + ], + [ + 260, + 161 + ] + ], + [ + [ + 245, + 153 + ], + [ + 250, + 132 + ], + [ + 258, + 151 + ], + [ + 230, + 155 + ], + [ + 253, + 171 + ], + [ + 225, + 177 + ] + ] + ] + }, + { + "task_id": "0003132", + "id": "RH20T_cfg7_task_0025_user_0020_scene_0008_cfg_0007_037522061512#271", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold a piece of paper'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003132/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 177, + 206 + ], + [ + 163, + 185 + ], + [ + 168, + 203 + ], + [ + 186, + 208 + ], + [ + 183, + 226 + ], + [ + 200, + 228 + ] + ], + [ + [ + 169, + 197 + ], + [ + 189, + 210 + ], + [ + 153, + 208 + ], + [ + 183, + 186 + ], + [ + 133, + 195 + ], + [ + 164, + 174 + ] + ], + [ + [ + 183, + 210 + ], + [ + 202, + 221 + ], + [ + 173, + 232 + ], + [ + 194, + 189 + ], + [ + 155, + 221 + ], + [ + 176, + 179 + ] + ], + [ + [ + 183, + 207 + ], + [ + 184, + 186 + ], + [ + 200, + 197 + ], + [ + 164, + 217 + ], + [ + 199, + 216 + ], + [ + 164, + 237 + ] + ] + ] + }, + { + "task_id": "0003133", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122064161#117", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003133/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 375, + 143 + ], + [ + 377, + 126 + ], + [ + 374, + 160 + ], + [ + 376, + 128 + ], + [ + 372, + 176 + ], + [ + 374, + 144 + ] + ], + [ + [ + 395, + 153 + ], + [ + 397, + 143 + ], + [ + 376, + 142 + ], + [ + 414, + 164 + ], + [ + 375, + 152 + ], + [ + 411, + 174 + ] + ], + [ + [ + 354, + 149 + ], + [ + 364, + 158 + ], + [ + 352, + 169 + ], + [ + 355, + 130 + ], + [ + 341, + 160 + ], + [ + 344, + 119 + ] + ], + [ + [ + 399, + 147 + ], + [ + 423, + 149 + ], + [ + 403, + 158 + ], + [ + 395, + 138 + ], + [ + 378, + 156 + ], + [ + 371, + 137 + ] + ] + ] + }, + { + "task_id": "0003134", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122064161#238", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003134/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 385, + 142 + ], + [ + 403, + 153 + ], + [ + 377, + 161 + ], + [ + 394, + 123 + ], + [ + 360, + 150 + ], + [ + 377, + 113 + ] + ], + [ + [ + 388, + 136 + ], + [ + 386, + 147 + ], + [ + 369, + 144 + ], + [ + 407, + 127 + ], + [ + 371, + 132 + ], + [ + 410, + 115 + ] + ], + [ + [ + 396, + 156 + ], + [ + 374, + 155 + ], + [ + 397, + 137 + ], + [ + 395, + 177 + ], + [ + 420, + 138 + ], + [ + 418, + 179 + ] + ], + [ + [ + 390, + 153 + ], + [ + 392, + 137 + ], + [ + 391, + 170 + ], + [ + 389, + 137 + ], + [ + 389, + 184 + ], + [ + 387, + 152 + ] + ] + ] + }, + { + "task_id": "0003135", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122064161#347", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003135/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 346, + 158 + ], + [ + 325, + 171 + ], + [ + 334, + 138 + ], + [ + 357, + 179 + ], + [ + 355, + 125 + ], + [ + 378, + 167 + ] + ], + [ + [ + 383, + 164 + ], + [ + 401, + 174 + ], + [ + 396, + 168 + ], + [ + 369, + 160 + ], + [ + 380, + 158 + ], + [ + 352, + 150 + ] + ], + [ + [ + 364, + 152 + ], + [ + 364, + 151 + ], + [ + 362, + 175 + ], + [ + 365, + 129 + ], + [ + 362, + 175 + ], + [ + 364, + 131 + ] + ], + [ + [ + 380, + 155 + ], + [ + 357, + 154 + ], + [ + 382, + 161 + ], + [ + 379, + 147 + ], + [ + 405, + 161 + ], + [ + 404, + 148 + ] + ] + ] + }, + { + "task_id": "0003136", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122064161#462", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003136/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 340, + 124 + ], + [ + 329, + 138 + ], + [ + 351, + 142 + ], + [ + 330, + 107 + ], + [ + 364, + 127 + ], + [ + 342, + 92 + ] + ], + [ + [ + 376, + 134 + ], + [ + 356, + 145 + ], + [ + 383, + 153 + ], + [ + 369, + 116 + ], + [ + 404, + 142 + ], + [ + 389, + 105 + ] + ], + [ + [ + 356, + 141 + ], + [ + 356, + 120 + ], + [ + 355, + 149 + ], + [ + 357, + 133 + ], + [ + 355, + 170 + ], + [ + 357, + 153 + ] + ], + [ + [ + 346, + 144 + ], + [ + 365, + 154 + ], + [ + 338, + 164 + ], + [ + 354, + 123 + ], + [ + 320, + 154 + ], + [ + 335, + 113 + ] + ] + ] + }, + { + "task_id": "0003140", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122060811#260", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003140/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 418, + 145 + ], + [ + 432, + 141 + ], + [ + 418, + 146 + ], + [ + 417, + 143 + ], + [ + 404, + 149 + ], + [ + 402, + 147 + ] + ], + [ + [ + 435, + 134 + ], + [ + 430, + 138 + ], + [ + 439, + 148 + ], + [ + 431, + 121 + ], + [ + 444, + 143 + ], + [ + 436, + 116 + ] + ], + [ + [ + 428, + 135 + ], + [ + 434, + 128 + ], + [ + 440, + 131 + ], + [ + 416, + 140 + ], + [ + 435, + 138 + ], + [ + 411, + 147 + ] + ], + [ + [ + 414, + 138 + ], + [ + 420, + 145 + ], + [ + 423, + 144 + ], + [ + 406, + 131 + ], + [ + 416, + 136 + ], + [ + 399, + 123 + ] + ] + ] + }, + { + "task_id": "0003141", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122060811#700", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003141/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 417, + 139 + ], + [ + 420, + 125 + ], + [ + 425, + 139 + ], + [ + 408, + 139 + ], + [ + 422, + 153 + ], + [ + 405, + 152 + ] + ], + [ + [ + 404, + 157 + ], + [ + 401, + 145 + ], + [ + 394, + 164 + ], + [ + 414, + 150 + ], + [ + 396, + 177 + ], + [ + 417, + 163 + ] + ], + [ + [ + 392, + 156 + ], + [ + 390, + 158 + ], + [ + 383, + 166 + ], + [ + 402, + 146 + ], + [ + 384, + 164 + ], + [ + 404, + 143 + ] + ], + [ + [ + 409, + 149 + ], + [ + 422, + 148 + ], + [ + 404, + 157 + ], + [ + 414, + 141 + ], + [ + 391, + 159 + ], + [ + 401, + 142 + ] + ] + ] + }, + { + "task_id": "0003144", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_038522063145#136", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003144/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 173, + 88 + ], + [ + 154, + 74 + ], + [ + 170, + 104 + ], + [ + 175, + 72 + ], + [ + 188, + 117 + ], + [ + 192, + 86 + ] + ], + [ + [ + 163, + 77 + ], + [ + 151, + 58 + ], + [ + 182, + 74 + ], + [ + 145, + 81 + ], + [ + 193, + 91 + ], + [ + 157, + 98 + ] + ], + [ + [ + 164, + 97 + ], + [ + 174, + 84 + ], + [ + 164, + 82 + ], + [ + 164, + 112 + ], + [ + 153, + 96 + ], + [ + 153, + 126 + ] + ], + [ + [ + 179, + 99 + ], + [ + 172, + 120 + ], + [ + 167, + 97 + ], + [ + 190, + 100 + ], + [ + 174, + 74 + ], + [ + 197, + 80 + ] + ] + ] + }, + { + "task_id": "0003147", + "id": "RH20T_cfg3_task_0023_user_0010_scene_0003_cfg_0003_038522062288#135", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003147/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 209, + 228 + ], + [ + 202, + 214 + ], + [ + 183, + 221 + ], + [ + 236, + 236 + ], + [ + 191, + 233 + ], + [ + 242, + 248 + ] + ], + [ + [ + 222, + 250 + ], + [ + 199, + 257 + ], + [ + 243, + 263 + ], + [ + 203, + 238 + ], + [ + 262, + 256 + ], + [ + 223, + 233 + ] + ], + [ + [ + 190, + 253 + ], + [ + 172, + 249 + ], + [ + 166, + 266 + ], + [ + 215, + 239 + ], + [ + 183, + 268 + ], + [ + 229, + 243 + ] + ], + [ + [ + 192, + 197 + ], + [ + 218, + 194 + ], + [ + 192, + 167 + ], + [ + 191, + 226 + ], + [ + 163, + 168 + ], + [ + 163, + 230 + ] + ] + ] + }, + { + "task_id": "0003149", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0006_cfg_0004_104122063550#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003149/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 198, + 195 + ], + [ + 223, + 194 + ], + [ + 206, + 171 + ], + [ + 191, + 218 + ], + [ + 179, + 172 + ], + [ + 164, + 221 + ] + ], + [ + [ + 218, + 182 + ], + [ + 198, + 183 + ], + [ + 202, + 160 + ], + [ + 232, + 203 + ], + [ + 224, + 158 + ], + [ + 253, + 203 + ] + ], + [ + [ + 215, + 197 + ], + [ + 200, + 175 + ], + [ + 236, + 180 + ], + [ + 192, + 215 + ], + [ + 249, + 201 + ], + [ + 207, + 236 + ] + ], + [ + [ + 227, + 214 + ], + [ + 205, + 230 + ], + [ + 227, + 224 + ], + [ + 228, + 202 + ], + [ + 247, + 209 + ], + [ + 250, + 186 + ] + ] + ] + }, + { + "task_id": "0003150", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0006_cfg_0004_104122063550#1803", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003150/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 282 + ], + [ + 345, + 289 + ], + [ + 330, + 253 + ], + [ + 307, + 312 + ], + [ + 306, + 248 + ], + [ + 282, + 304 + ] + ], + [ + [ + 265, + 279 + ], + [ + 282, + 274 + ], + [ + 276, + 248 + ], + [ + 255, + 309 + ], + [ + 258, + 251 + ], + [ + 236, + 316 + ] + ], + [ + [ + 277, + 292 + ], + [ + 264, + 279 + ], + [ + 291, + 314 + ], + [ + 262, + 269 + ], + [ + 306, + 329 + ], + [ + 276, + 283 + ] + ], + [ + [ + 283, + 283 + ], + [ + 279, + 258 + ], + [ + 302, + 297 + ], + [ + 265, + 270 + ], + [ + 305, + 322 + ], + [ + 269, + 293 + ] + ] + ] + }, + { + "task_id": "0003151", + "id": "RH20T_cfg4_task_0009_user_0010_scene_0004_cfg_0004_036422060909#98", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003151/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 318, + 117 + ], + [ + 321, + 97 + ], + [ + 334, + 103 + ], + [ + 302, + 131 + ], + [ + 331, + 122 + ], + [ + 300, + 150 + ] + ], + [ + [ + 319, + 144 + ], + [ + 307, + 144 + ], + [ + 341, + 132 + ], + [ + 299, + 156 + ], + [ + 351, + 133 + ], + [ + 310, + 155 + ] + ], + [ + [ + 309, + 141 + ], + [ + 286, + 129 + ], + [ + 297, + 165 + ], + [ + 321, + 118 + ], + [ + 319, + 177 + ], + [ + 344, + 130 + ] + ], + [ + [ + 318, + 130 + ], + [ + 321, + 104 + ], + [ + 294, + 134 + ], + [ + 341, + 127 + ], + [ + 292, + 159 + ], + [ + 337, + 151 + ] + ] + ] + }, + { + "task_id": "0003152", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122060811#230", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003152/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 351, + 170 + ], + [ + 346, + 157 + ], + [ + 362, + 161 + ], + [ + 341, + 178 + ], + [ + 368, + 174 + ], + [ + 346, + 191 + ] + ], + [ + [ + 377, + 182 + ], + [ + 379, + 182 + ], + [ + 363, + 188 + ], + [ + 391, + 175 + ], + [ + 360, + 189 + ], + [ + 389, + 175 + ] + ], + [ + [ + 363, + 186 + ], + [ + 365, + 183 + ], + [ + 360, + 201 + ], + [ + 367, + 170 + ], + [ + 358, + 204 + ], + [ + 365, + 173 + ] + ], + [ + [ + 364, + 182 + ], + [ + 374, + 174 + ], + [ + 375, + 182 + ], + [ + 353, + 182 + ], + [ + 366, + 190 + ], + [ + 344, + 190 + ] + ] + ] + }, + { + "task_id": "0003153", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122060811#651", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003153/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 313, + 144 + ], + [ + 326, + 145 + ], + [ + 318, + 156 + ], + [ + 309, + 131 + ], + [ + 306, + 155 + ], + [ + 297, + 130 + ] + ], + [ + [ + 311, + 148 + ], + [ + 321, + 141 + ], + [ + 321, + 150 + ], + [ + 300, + 146 + ], + [ + 312, + 158 + ], + [ + 290, + 154 + ] + ], + [ + [ + 309, + 144 + ], + [ + 303, + 139 + ], + [ + 296, + 145 + ], + [ + 322, + 143 + ], + [ + 302, + 150 + ], + [ + 329, + 148 + ] + ], + [ + [ + 304, + 149 + ], + [ + 313, + 137 + ], + [ + 307, + 151 + ], + [ + 302, + 148 + ], + [ + 298, + 163 + ], + [ + 293, + 159 + ] + ] + ] + }, + { + "task_id": "0003154", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122060811#1000", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003154/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 360, + 121 + ], + [ + 364, + 115 + ], + [ + 371, + 130 + ], + [ + 348, + 113 + ], + [ + 367, + 136 + ], + [ + 343, + 118 + ] + ], + [ + [ + 371, + 123 + ], + [ + 362, + 127 + ], + [ + 365, + 111 + ], + [ + 378, + 136 + ], + [ + 373, + 108 + ], + [ + 387, + 132 + ] + ], + [ + [ + 368, + 118 + ], + [ + 373, + 110 + ], + [ + 380, + 116 + ], + [ + 355, + 120 + ], + [ + 375, + 124 + ], + [ + 351, + 127 + ] + ], + [ + [ + 362, + 112 + ], + [ + 376, + 108 + ], + [ + 363, + 120 + ], + [ + 362, + 104 + ], + [ + 349, + 124 + ], + [ + 348, + 107 + ] + ] + ] + }, + { + "task_id": "0003155", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122060811#1279", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003155/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 411, + 149 + ], + [ + 398, + 144 + ], + [ + 411, + 141 + ], + [ + 411, + 157 + ], + [ + 425, + 145 + ], + [ + 424, + 161 + ] + ], + [ + [ + 412, + 127 + ], + [ + 417, + 131 + ], + [ + 413, + 141 + ], + [ + 411, + 112 + ], + [ + 409, + 137 + ], + [ + 407, + 109 + ] + ], + [ + [ + 411, + 133 + ], + [ + 421, + 126 + ], + [ + 422, + 135 + ], + [ + 400, + 132 + ], + [ + 412, + 142 + ], + [ + 391, + 139 + ] + ], + [ + [ + 412, + 134 + ], + [ + 414, + 146 + ], + [ + 425, + 129 + ], + [ + 398, + 138 + ], + [ + 423, + 117 + ], + [ + 397, + 126 + ] + ] + ] + }, + { + "task_id": "0003156", + "id": "RH20T_cfg1_task_0043_user_0016_scene_0002_cfg_0001_819112071037#123", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003156/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 293, + 88 + ], + [ + 276, + 76 + ], + [ + 338, + 97 + ], + [ + 244, + 78 + ], + [ + 361, + 111 + ], + [ + 260, + 92 + ] + ], + [ + [ + 285, + 32 + ], + [ + 341, + 19 + ], + [ + 288, + 52 + ], + [ + 280, + 6 + ], + [ + 237, + 64 + ], + [ + 215, + 21 + ] + ], + [ + [ + 261, + 114 + ], + [ + 310, + 106 + ], + [ + 262, + 70 + ], + [ + 260, + 161 + ], + [ + 217, + 78 + ], + [ + 211, + 167 + ] + ], + [ + [ + 313, + 73 + ], + [ + 313, + 15 + ], + [ + 259, + 74 + ], + [ + 362, + 71 + ], + [ + 261, + 129 + ], + [ + 361, + 122 + ] + ] + ] + }, + { + "task_id": "0003157", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_104122061018#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003157/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 362, + 214 + ], + [ + 358, + 185 + ], + [ + 346, + 200 + ], + [ + 380, + 230 + ], + [ + 350, + 226 + ], + [ + 384, + 258 + ] + ], + [ + [ + 322, + 197 + ], + [ + 304, + 213 + ], + [ + 313, + 165 + ], + [ + 330, + 227 + ], + [ + 330, + 153 + ], + [ + 346, + 212 + ] + ], + [ + [ + 352, + 202 + ], + [ + 351, + 173 + ], + [ + 376, + 216 + ], + [ + 330, + 190 + ], + [ + 376, + 244 + ], + [ + 332, + 216 + ] + ], + [ + [ + 375, + 226 + ], + [ + 343, + 243 + ], + [ + 384, + 224 + ], + [ + 367, + 228 + ], + [ + 418, + 207 + ], + [ + 396, + 213 + ] + ] + ] + }, + { + "task_id": "0003158", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_104122061018#190", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003158/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 380, + 208 + ], + [ + 347, + 193 + ], + [ + 366, + 240 + ], + [ + 396, + 175 + ], + [ + 399, + 255 + ], + [ + 429, + 190 + ] + ], + [ + [ + 408, + 249 + ], + [ + 432, + 226 + ], + [ + 402, + 234 + ], + [ + 415, + 266 + ], + [ + 380, + 256 + ], + [ + 390, + 290 + ] + ], + [ + [ + 378, + 217 + ], + [ + 377, + 190 + ], + [ + 360, + 200 + ], + [ + 398, + 236 + ], + [ + 362, + 223 + ], + [ + 398, + 261 + ] + ], + [ + [ + 390, + 181 + ], + [ + 360, + 197 + ], + [ + 389, + 163 + ], + [ + 390, + 197 + ], + [ + 421, + 147 + ], + [ + 418, + 183 + ] + ] + ] + }, + { + "task_id": "0003159", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_104122061018#292", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003159/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 440, + 236 + ], + [ + 423, + 210 + ], + [ + 463, + 215 + ], + [ + 418, + 255 + ], + [ + 483, + 244 + ], + [ + 435, + 282 + ] + ], + [ + [ + 377, + 223 + ], + [ + 351, + 207 + ], + [ + 393, + 203 + ], + [ + 361, + 245 + ], + [ + 419, + 219 + ], + [ + 388, + 263 + ] + ], + [ + [ + 381, + 209 + ], + [ + 382, + 178 + ], + [ + 347, + 205 + ], + [ + 416, + 213 + ], + [ + 345, + 238 + ], + [ + 417, + 247 + ] + ], + [ + [ + 399, + 222 + ], + [ + 402, + 196 + ], + [ + 380, + 203 + ], + [ + 420, + 242 + ], + [ + 377, + 226 + ], + [ + 415, + 267 + ] + ] + ] + }, + { + "task_id": "0003160", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0009_cfg_0007_104122061018#414", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003160/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 425, + 208 + ], + [ + 456, + 208 + ], + [ + 429, + 241 + ], + [ + 421, + 177 + ], + [ + 396, + 241 + ], + [ + 389, + 178 + ] + ], + [ + [ + 433, + 189 + ], + [ + 457, + 172 + ], + [ + 457, + 196 + ], + [ + 412, + 183 + ], + [ + 430, + 215 + ], + [ + 387, + 199 + ] + ], + [ + [ + 417, + 191 + ], + [ + 400, + 175 + ], + [ + 401, + 175 + ], + [ + 435, + 208 + ], + [ + 417, + 189 + ], + [ + 450, + 222 + ] + ], + [ + [ + 415, + 225 + ], + [ + 415, + 199 + ], + [ + 394, + 207 + ], + [ + 439, + 244 + ], + [ + 394, + 230 + ], + [ + 437, + 267 + ] + ] + ] + }, + { + "task_id": "0003163", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122061602#1197", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003163/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 592, + 320 + ], + [ + 624, + 324 + ], + [ + 605, + 352 + ], + [ + 579, + 287 + ], + [ + 576, + 347 + ], + [ + 551, + 285 + ] + ], + [ + [ + 589, + 321 + ], + [ + 600, + 322 + ], + [ + 585, + 353 + ], + [ + 593, + 289 + ], + [ + 573, + 353 + ], + [ + 582, + 286 + ] + ], + [ + [ + 640, + 345 + ], + [ + 614, + 346 + ], + [ + 637, + 308 + ], + [ + 640, + 360 + ], + [ + 640, + 304 + ], + [ + 640, + 360 + ] + ], + [ + [ + 617, + 331 + ], + [ + 625, + 331 + ], + [ + 631, + 301 + ], + [ + 603, + 360 + ], + [ + 623, + 299 + ], + [ + 593, + 360 + ] + ] + ] + }, + { + "task_id": "0003165", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0001_cfg_0006_104122061602#2061", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003165/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 431, + 102 + ], + [ + 428, + 82 + ], + [ + 440, + 95 + ], + [ + 422, + 108 + ], + [ + 443, + 116 + ], + [ + 423, + 127 + ] + ], + [ + [ + 444, + 117 + ], + [ + 442, + 103 + ], + [ + 436, + 132 + ], + [ + 451, + 101 + ], + [ + 438, + 144 + ], + [ + 452, + 114 + ] + ], + [ + [ + 433, + 106 + ], + [ + 448, + 113 + ], + [ + 432, + 125 + ], + [ + 434, + 86 + ], + [ + 419, + 117 + ], + [ + 420, + 80 + ] + ], + [ + [ + 455, + 108 + ], + [ + 451, + 102 + ], + [ + 434, + 103 + ], + [ + 477, + 112 + ], + [ + 439, + 109 + ], + [ + 481, + 118 + ] + ] + ] + }, + { + "task_id": "0003167", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104422070011#123", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003167/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 362, + 90 + ], + [ + 371, + 62 + ], + [ + 337, + 83 + ], + [ + 388, + 97 + ], + [ + 329, + 109 + ], + [ + 379, + 125 + ] + ], + [ + [ + 356, + 98 + ], + [ + 342, + 120 + ], + [ + 334, + 81 + ], + [ + 378, + 115 + ], + [ + 348, + 57 + ], + [ + 394, + 92 + ] + ], + [ + [ + 370, + 126 + ], + [ + 356, + 102 + ], + [ + 355, + 124 + ], + [ + 384, + 126 + ], + [ + 370, + 150 + ], + [ + 398, + 150 + ] + ], + [ + [ + 353, + 106 + ], + [ + 364, + 89 + ], + [ + 326, + 111 + ], + [ + 378, + 100 + ], + [ + 317, + 126 + ], + [ + 367, + 115 + ] + ] + ] + }, + { + "task_id": "0003168", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104422070011#157", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003168/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 339, + 121 + ], + [ + 340, + 104 + ], + [ + 327, + 141 + ], + [ + 352, + 100 + ], + [ + 325, + 160 + ], + [ + 351, + 118 + ] + ], + [ + [ + 333, + 119 + ], + [ + 327, + 87 + ], + [ + 365, + 112 + ], + [ + 302, + 126 + ], + [ + 371, + 145 + ], + [ + 308, + 157 + ] + ], + [ + [ + 338, + 130 + ], + [ + 353, + 115 + ], + [ + 314, + 137 + ], + [ + 361, + 123 + ], + [ + 302, + 151 + ], + [ + 348, + 137 + ] + ], + [ + [ + 368, + 142 + ], + [ + 373, + 114 + ], + [ + 379, + 130 + ], + [ + 355, + 155 + ], + [ + 373, + 156 + ], + [ + 350, + 182 + ] + ] + ] + }, + { + "task_id": "0003169", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104422070011#203", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003169/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 320, + 136 + ], + [ + 331, + 134 + ], + [ + 343, + 123 + ], + [ + 297, + 150 + ], + [ + 333, + 126 + ], + [ + 289, + 152 + ] + ], + [ + [ + 355, + 93 + ], + [ + 357, + 103 + ], + [ + 362, + 65 + ], + [ + 347, + 121 + ], + [ + 360, + 52 + ], + [ + 344, + 111 + ] + ], + [ + [ + 350, + 82 + ], + [ + 371, + 84 + ], + [ + 369, + 65 + ], + [ + 331, + 100 + ], + [ + 349, + 64 + ], + [ + 313, + 98 + ] + ], + [ + [ + 327, + 104 + ], + [ + 339, + 94 + ], + [ + 303, + 114 + ], + [ + 350, + 95 + ], + [ + 293, + 122 + ], + [ + 339, + 105 + ] + ] + ] + }, + { + "task_id": "0003170", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104422070011#249", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003170/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 370, + 140 + ], + [ + 366, + 163 + ], + [ + 353, + 127 + ], + [ + 387, + 155 + ], + [ + 356, + 103 + ], + [ + 392, + 130 + ] + ], + [ + [ + 364, + 143 + ], + [ + 379, + 128 + ], + [ + 342, + 153 + ], + [ + 386, + 133 + ], + [ + 330, + 167 + ], + [ + 372, + 147 + ] + ], + [ + [ + 346, + 134 + ], + [ + 336, + 143 + ], + [ + 350, + 102 + ], + [ + 342, + 164 + ], + [ + 358, + 96 + ], + [ + 351, + 155 + ] + ], + [ + [ + 392, + 113 + ], + [ + 409, + 133 + ], + [ + 395, + 131 + ], + [ + 388, + 93 + ], + [ + 380, + 113 + ], + [ + 372, + 74 + ] + ] + ] + }, + { + "task_id": "0003171", + "id": "RH20T_cfg6_task_0039_user_0014_scene_0002_cfg_0006_104122061602#125", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003171/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 448, + 165 + ], + [ + 453, + 142 + ], + [ + 470, + 181 + ], + [ + 428, + 151 + ], + [ + 465, + 202 + ], + [ + 425, + 171 + ] + ], + [ + [ + 469, + 175 + ], + [ + 462, + 158 + ], + [ + 461, + 155 + ], + [ + 479, + 196 + ], + [ + 468, + 170 + ], + [ + 485, + 211 + ] + ], + [ + [ + 428, + 177 + ], + [ + 439, + 200 + ], + [ + 448, + 178 + ], + [ + 410, + 175 + ], + [ + 436, + 152 + ], + [ + 398, + 152 + ] + ], + [ + [ + 445, + 138 + ], + [ + 418, + 148 + ], + [ + 445, + 124 + ], + [ + 445, + 154 + ], + [ + 472, + 114 + ], + [ + 475, + 143 + ] + ] + ] + }, + { + "task_id": "0003173", + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_819112070044#73", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003173/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 474, + 168 + ], + [ + 497, + 164 + ], + [ + 474, + 180 + ], + [ + 474, + 155 + ], + [ + 452, + 184 + ], + [ + 450, + 160 + ] + ], + [ + [ + 480, + 179 + ], + [ + 503, + 185 + ], + [ + 486, + 158 + ], + [ + 475, + 200 + ], + [ + 463, + 153 + ], + [ + 453, + 194 + ] + ], + [ + [ + 478, + 174 + ], + [ + 502, + 180 + ], + [ + 484, + 152 + ], + [ + 472, + 195 + ], + [ + 460, + 146 + ], + [ + 450, + 189 + ] + ], + [ + [ + 467, + 168 + ], + [ + 470, + 148 + ], + [ + 472, + 158 + ], + [ + 461, + 179 + ], + [ + 469, + 177 + ], + [ + 458, + 199 + ] + ] + ] + }, + { + "task_id": "0003175", + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_819112070044#156", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003175/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 469, + 204 + ], + [ + 447, + 196 + ], + [ + 478, + 183 + ], + [ + 461, + 226 + ], + [ + 501, + 191 + ], + [ + 483, + 235 + ] + ], + [ + [ + 443, + 193 + ], + [ + 422, + 186 + ], + [ + 450, + 198 + ], + [ + 436, + 188 + ], + [ + 472, + 205 + ], + [ + 460, + 195 + ] + ], + [ + [ + 453, + 199 + ], + [ + 462, + 221 + ], + [ + 472, + 194 + ], + [ + 433, + 204 + ], + [ + 463, + 173 + ], + [ + 424, + 182 + ] + ], + [ + [ + 452, + 186 + ], + [ + 455, + 165 + ], + [ + 460, + 177 + ], + [ + 444, + 196 + ], + [ + 457, + 196 + ], + [ + 441, + 217 + ] + ] + ] + }, + { + "task_id": "0003176", + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_819112070044#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003176/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 438, + 156 + ], + [ + 438, + 137 + ], + [ + 442, + 145 + ], + [ + 434, + 167 + ], + [ + 443, + 163 + ], + [ + 435, + 186 + ] + ], + [ + [ + 431, + 176 + ], + [ + 450, + 164 + ], + [ + 443, + 194 + ], + [ + 419, + 157 + ], + [ + 425, + 206 + ], + [ + 401, + 169 + ] + ], + [ + [ + 435, + 149 + ], + [ + 421, + 145 + ], + [ + 420, + 150 + ], + [ + 452, + 148 + ], + [ + 434, + 154 + ], + [ + 466, + 152 + ] + ], + [ + [ + 430, + 148 + ], + [ + 433, + 136 + ], + [ + 448, + 158 + ], + [ + 412, + 138 + ], + [ + 445, + 169 + ], + [ + 410, + 150 + ] + ] + ] + }, + { + "task_id": "0003177", + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522063145#152", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003177/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 499, + 182 + ], + [ + 490, + 198 + ], + [ + 517, + 197 + ], + [ + 482, + 168 + ], + [ + 528, + 179 + ], + [ + 491, + 150 + ] + ], + [ + [ + 497, + 157 + ], + [ + 504, + 141 + ], + [ + 510, + 147 + ], + [ + 485, + 168 + ], + [ + 503, + 162 + ], + [ + 479, + 184 + ] + ], + [ + [ + 492, + 133 + ], + [ + 510, + 146 + ], + [ + 483, + 151 + ], + [ + 502, + 115 + ], + [ + 467, + 139 + ], + [ + 484, + 103 + ] + ], + [ + [ + 495, + 164 + ], + [ + 488, + 151 + ], + [ + 505, + 148 + ], + [ + 485, + 181 + ], + [ + 512, + 161 + ], + [ + 492, + 194 + ] + ] + ] + }, + { + "task_id": "0003178", + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522063145#256", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003178/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 475, + 149 + ], + [ + 483, + 168 + ], + [ + 493, + 145 + ], + [ + 455, + 153 + ], + [ + 485, + 126 + ], + [ + 448, + 133 + ] + ], + [ + [ + 491, + 140 + ], + [ + 498, + 122 + ], + [ + 503, + 130 + ], + [ + 478, + 151 + ], + [ + 496, + 147 + ], + [ + 472, + 167 + ] + ], + [ + [ + 492, + 116 + ], + [ + 514, + 108 + ], + [ + 482, + 99 + ], + [ + 502, + 134 + ], + [ + 461, + 107 + ], + [ + 479, + 143 + ] + ], + [ + [ + 473, + 140 + ], + [ + 474, + 152 + ], + [ + 496, + 146 + ], + [ + 450, + 135 + ], + [ + 496, + 134 + ], + [ + 448, + 123 + ] + ] + ] + }, + { + "task_id": "0003179", + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_038522063145#387", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003179/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 478, + 140 + ], + [ + 483, + 122 + ], + [ + 490, + 131 + ], + [ + 465, + 150 + ], + [ + 484, + 148 + ], + [ + 460, + 167 + ] + ], + [ + [ + 483, + 168 + ], + [ + 486, + 149 + ], + [ + 495, + 158 + ], + [ + 472, + 177 + ], + [ + 492, + 179 + ], + [ + 469, + 196 + ] + ], + [ + [ + 495, + 134 + ], + [ + 477, + 119 + ], + [ + 511, + 116 + ], + [ + 479, + 152 + ], + [ + 530, + 132 + ], + [ + 497, + 167 + ] + ], + [ + [ + 480, + 148 + ], + [ + 461, + 139 + ], + [ + 471, + 146 + ], + [ + 488, + 150 + ], + [ + 491, + 156 + ], + [ + 507, + 160 + ] + ] + ] + }, + { + "task_id": "0003183", + "id": "RH20T_cfg7_task_0020_user_0014_scene_0010_cfg_0007_104122060811#124", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003183/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 350, + 143 + ], + [ + 343, + 131 + ], + [ + 353, + 137 + ], + [ + 347, + 149 + ], + [ + 359, + 148 + ], + [ + 355, + 160 + ] + ], + [ + [ + 354, + 138 + ], + [ + 344, + 128 + ], + [ + 363, + 128 + ], + [ + 345, + 148 + ], + [ + 374, + 138 + ], + [ + 355, + 158 + ] + ], + [ + [ + 349, + 148 + ], + [ + 363, + 144 + ], + [ + 350, + 154 + ], + [ + 348, + 141 + ], + [ + 336, + 157 + ], + [ + 334, + 145 + ] + ], + [ + [ + 340, + 141 + ], + [ + 341, + 129 + ], + [ + 338, + 134 + ], + [ + 342, + 148 + ], + [ + 336, + 147 + ], + [ + 341, + 161 + ] + ] + ] + }, + { + "task_id": "0003184", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_104422070011#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003184/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 290, + 304 + ], + [ + 270, + 280 + ], + [ + 267, + 326 + ], + [ + 313, + 283 + ], + [ + 287, + 351 + ], + [ + 335, + 308 + ] + ], + [ + [ + 286, + 261 + ], + [ + 318, + 264 + ], + [ + 284, + 277 + ], + [ + 287, + 248 + ], + [ + 249, + 274 + ], + [ + 256, + 246 + ] + ], + [ + [ + 227, + 287 + ], + [ + 212, + 284 + ], + [ + 247, + 260 + ], + [ + 205, + 316 + ], + [ + 264, + 261 + ], + [ + 220, + 321 + ] + ], + [ + [ + 264, + 307 + ], + [ + 270, + 298 + ], + [ + 234, + 322 + ], + [ + 292, + 293 + ], + [ + 230, + 330 + ], + [ + 285, + 302 + ] + ] + ] + }, + { + "task_id": "0003185", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_104422070011#350", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003185/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 263, + 325 + ], + [ + 228, + 327 + ], + [ + 262, + 301 + ], + [ + 264, + 352 + ], + [ + 294, + 299 + ], + [ + 300, + 350 + ] + ], + [ + [ + 291, + 275 + ], + [ + 311, + 298 + ], + [ + 297, + 255 + ], + [ + 286, + 294 + ], + [ + 277, + 233 + ], + [ + 268, + 273 + ] + ], + [ + [ + 259, + 287 + ], + [ + 280, + 276 + ], + [ + 254, + 313 + ], + [ + 263, + 264 + ], + [ + 233, + 323 + ], + [ + 244, + 275 + ] + ], + [ + [ + 278, + 303 + ], + [ + 287, + 293 + ], + [ + 248, + 313 + ], + [ + 305, + 295 + ], + [ + 242, + 322 + ], + [ + 296, + 304 + ] + ] + ] + }, + { + "task_id": "0003186", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_104422070011#707", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003186/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 313, + 280 + ], + [ + 321, + 302 + ], + [ + 329, + 293 + ], + [ + 295, + 266 + ], + [ + 321, + 274 + ], + [ + 288, + 247 + ] + ], + [ + [ + 300, + 289 + ], + [ + 308, + 279 + ], + [ + 277, + 308 + ], + [ + 321, + 272 + ], + [ + 270, + 317 + ], + [ + 312, + 282 + ] + ], + [ + [ + 280, + 301 + ], + [ + 290, + 304 + ], + [ + 272, + 272 + ], + [ + 289, + 330 + ], + [ + 263, + 271 + ], + [ + 279, + 326 + ] + ], + [ + [ + 339, + 284 + ], + [ + 362, + 305 + ], + [ + 350, + 270 + ], + [ + 329, + 297 + ], + [ + 326, + 249 + ], + [ + 307, + 277 + ] + ] + ] + }, + { + "task_id": "0003187", + "id": "RH20T_cfg4_task_0065_user_0007_scene_0004_cfg_0004_104422070011#869", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003187/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 331, + 273 + ], + [ + 341, + 263 + ], + [ + 304, + 279 + ], + [ + 356, + 267 + ], + [ + 296, + 288 + ], + [ + 346, + 275 + ] + ], + [ + [ + 329, + 301 + ], + [ + 329, + 283 + ], + [ + 354, + 290 + ], + [ + 303, + 313 + ], + [ + 353, + 306 + ], + [ + 304, + 330 + ] + ], + [ + [ + 329, + 287 + ], + [ + 327, + 317 + ], + [ + 345, + 297 + ], + [ + 311, + 277 + ], + [ + 347, + 270 + ], + [ + 314, + 248 + ] + ], + [ + [ + 331, + 277 + ], + [ + 339, + 273 + ], + [ + 313, + 255 + ], + [ + 350, + 300 + ], + [ + 307, + 260 + ], + [ + 342, + 302 + ] + ] + ] + }, + { + "task_id": "0003188", + "id": "RH20T_cfg7_task_0033_user_0020_scene_0008_cfg_0007_104122061018#189", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003188/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 265, + 119 + ], + [ + 249, + 103 + ], + [ + 262, + 100 + ], + [ + 268, + 138 + ], + [ + 276, + 115 + ], + [ + 283, + 152 + ] + ], + [ + [ + 244, + 136 + ], + [ + 255, + 115 + ], + [ + 229, + 118 + ], + [ + 257, + 153 + ], + [ + 216, + 140 + ], + [ + 245, + 174 + ] + ], + [ + [ + 242, + 103 + ], + [ + 232, + 82 + ], + [ + 242, + 119 + ], + [ + 243, + 85 + ], + [ + 252, + 140 + ], + [ + 254, + 108 + ] + ], + [ + [ + 249, + 83 + ], + [ + 224, + 95 + ], + [ + 262, + 108 + ], + [ + 236, + 57 + ], + [ + 285, + 96 + ], + [ + 261, + 45 + ] + ] + ] + }, + { + "task_id": "0003189", + "id": "RH20T_cfg1_task_0059_user_0001_scene_0001_cfg_0001_819112070044#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003189/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 444, + 205 + ], + [ + 447, + 190 + ], + [ + 433, + 189 + ], + [ + 455, + 221 + ], + [ + 430, + 203 + ], + [ + 451, + 234 + ] + ], + [ + [ + 452, + 211 + ], + [ + 436, + 208 + ], + [ + 463, + 194 + ], + [ + 440, + 228 + ], + [ + 478, + 198 + ], + [ + 456, + 230 + ] + ], + [ + [ + 464, + 205 + ], + [ + 466, + 209 + ], + [ + 442, + 209 + ], + [ + 486, + 202 + ], + [ + 439, + 205 + ], + [ + 484, + 197 + ] + ], + [ + [ + 429, + 212 + ], + [ + 442, + 216 + ], + [ + 445, + 223 + ], + [ + 415, + 203 + ], + [ + 431, + 220 + ], + [ + 401, + 199 + ] + ] + ] + }, + { + "task_id": "0003190", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_750612070853#145", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003190/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 381, + 277 + ], + [ + 412, + 243 + ], + [ + 319, + 247 + ], + [ + 443, + 307 + ], + [ + 301, + 277 + ], + [ + 412, + 331 + ] + ], + [ + [ + 354, + 271 + ], + [ + 295, + 299 + ], + [ + 335, + 210 + ], + [ + 372, + 330 + ], + [ + 390, + 187 + ], + [ + 423, + 303 + ] + ], + [ + [ + 416, + 242 + ], + [ + 421, + 261 + ], + [ + 485, + 237 + ], + [ + 350, + 246 + ], + [ + 472, + 223 + ], + [ + 354, + 232 + ] + ], + [ + [ + 352, + 330 + ], + [ + 403, + 359 + ], + [ + 319, + 360 + ], + [ + 383, + 273 + ], + [ + 263, + 360 + ], + [ + 331, + 241 + ] + ] + ] + }, + { + "task_id": "0003192", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_750612070853#945", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003192/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 151, + 297 + ], + [ + 232, + 243 + ], + [ + 152, + 273 + ], + [ + 150, + 333 + ], + [ + 78, + 321 + ], + [ + 38, + 360 + ] + ], + [ + [ + 148, + 291 + ], + [ + 52, + 351 + ], + [ + 187, + 293 + ], + [ + 119, + 290 + ], + [ + 267, + 239 + ], + [ + 187, + 250 + ] + ], + [ + [ + 129, + 223 + ], + [ + 192, + 247 + ], + [ + 131, + 186 + ], + [ + 127, + 271 + ], + [ + 66, + 158 + ], + [ + 41, + 244 + ] + ], + [ + [ + 156, + 286 + ], + [ + 115, + 250 + ], + [ + 112, + 252 + ], + [ + 207, + 326 + ], + [ + 146, + 282 + ], + [ + 234, + 351 + ] + ] + ] + }, + { + "task_id": "0003194", + "id": "RH20T_cfg3_task_0070_user_0010_scene_0004_cfg_0003_104122062823#175", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003194/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 244, + 49 + ], + [ + 265, + 69 + ], + [ + 235, + 49 + ], + [ + 253, + 50 + ], + [ + 214, + 28 + ], + [ + 230, + 27 + ] + ], + [ + [ + 245, + 84 + ], + [ + 253, + 55 + ], + [ + 271, + 95 + ], + [ + 221, + 73 + ], + [ + 263, + 123 + ], + [ + 214, + 101 + ] + ], + [ + [ + 210, + 69 + ], + [ + 195, + 74 + ], + [ + 221, + 44 + ], + [ + 198, + 96 + ], + [ + 239, + 38 + ], + [ + 215, + 93 + ] + ], + [ + [ + 259, + 112 + ], + [ + 268, + 127 + ], + [ + 271, + 130 + ], + [ + 247, + 94 + ], + [ + 261, + 113 + ], + [ + 236, + 76 + ] + ] + ] + }, + { + "task_id": "0003196", + "id": "RH20T_cfg5_task_0044_user_0010_scene_0002_cfg_0005_105422061350#226", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003196/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 212, + 43 + ], + [ + 215, + 29 + ], + [ + 184, + 27 + ], + [ + 237, + 58 + ], + [ + 179, + 41 + ], + [ + 235, + 73 + ] + ], + [ + [ + 220, + 62 + ], + [ + 213, + 46 + ], + [ + 233, + 40 + ], + [ + 207, + 83 + ], + [ + 241, + 57 + ], + [ + 214, + 101 + ] + ], + [ + [ + 215, + 55 + ], + [ + 214, + 33 + ], + [ + 230, + 38 + ], + [ + 198, + 73 + ], + [ + 230, + 58 + ], + [ + 200, + 92 + ] + ], + [ + [ + 203, + 69 + ], + [ + 210, + 39 + ], + [ + 183, + 68 + ], + [ + 220, + 70 + ], + [ + 176, + 98 + ], + [ + 213, + 97 + ] + ] + ] + }, + { + "task_id": "0003198", + "id": "RH20T_cfg1_task_0089_user_0014_scene_0002_cfg_0001_035622060973#130", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push an object with obstacles'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003198/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 240, + 259 + ], + [ + 237, + 252 + ], + [ + 227, + 242 + ], + [ + 253, + 277 + ], + [ + 230, + 250 + ], + [ + 255, + 283 + ] + ], + [ + [ + 260, + 246 + ], + [ + 272, + 227 + ], + [ + 278, + 251 + ], + [ + 241, + 241 + ], + [ + 267, + 269 + ], + [ + 230, + 260 + ] + ], + [ + [ + 239, + 239 + ], + [ + 219, + 251 + ], + [ + 250, + 258 + ], + [ + 228, + 220 + ], + [ + 269, + 247 + ], + [ + 248, + 209 + ] + ], + [ + [ + 260, + 258 + ], + [ + 237, + 259 + ], + [ + 256, + 243 + ], + [ + 263, + 272 + ], + [ + 279, + 241 + ], + [ + 284, + 271 + ] + ] + ] + }, + { + "task_id": "0003199", + "id": "RH20T_cfg5_task_0083_user_0007_scene_0005_cfg_0005_105422061350#310", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Swat with a flyswatter'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003199/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 87, + 220 + ], + [ + 69, + 241 + ], + [ + 62, + 226 + ], + [ + 112, + 214 + ], + [ + 79, + 208 + ], + [ + 127, + 195 + ] + ], + [ + [ + 74, + 205 + ], + [ + 98, + 189 + ], + [ + 59, + 219 + ], + [ + 90, + 191 + ], + [ + 32, + 237 + ], + [ + 62, + 209 + ] + ], + [ + [ + 103, + 224 + ], + [ + 125, + 199 + ], + [ + 76, + 207 + ], + [ + 131, + 241 + ], + [ + 53, + 234 + ], + [ + 109, + 268 + ] + ], + [ + [ + 108, + 202 + ], + [ + 99, + 187 + ], + [ + 136, + 215 + ], + [ + 80, + 189 + ], + [ + 141, + 229 + ], + [ + 88, + 204 + ] + ] + ] + }, + { + "task_id": "0003202", + "id": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_037522061512#383", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003202/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 183, + 162 + ], + [ + 199, + 172 + ], + [ + 179, + 145 + ], + [ + 187, + 178 + ], + [ + 163, + 136 + ], + [ + 173, + 169 + ] + ], + [ + [ + 173, + 166 + ], + [ + 188, + 150 + ], + [ + 181, + 181 + ], + [ + 165, + 151 + ], + [ + 166, + 197 + ], + [ + 148, + 168 + ] + ], + [ + [ + 171, + 154 + ], + [ + 190, + 166 + ], + [ + 185, + 152 + ], + [ + 156, + 156 + ], + [ + 168, + 140 + ], + [ + 137, + 143 + ] + ], + [ + [ + 175, + 151 + ], + [ + 194, + 161 + ], + [ + 188, + 137 + ], + [ + 163, + 165 + ], + [ + 168, + 126 + ], + [ + 143, + 156 + ] + ] + ] + }, + { + "task_id": "0003204", + "id": "RH20T_cfg4_task_0012_user_0010_scene_0005_cfg_0004_104122062823#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push the soccer ball into the goal'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003204/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 185, + 19 + ], + [ + 173, + 39 + ], + [ + 192, + 36 + ], + [ + 179, + 3 + ], + [ + 205, + 15 + ], + [ + 191, + 0 + ] + ], + [ + [ + 184, + 13 + ], + [ + 183, + 0 + ], + [ + 205, + 25 + ], + [ + 163, + 1 + ], + [ + 204, + 46 + ], + [ + 165, + 23 + ] + ], + [ + [ + 196, + 29 + ], + [ + 197, + 10 + ], + [ + 184, + 45 + ], + [ + 207, + 13 + ], + [ + 183, + 62 + ], + [ + 205, + 31 + ] + ], + [ + [ + 202, + 40 + ], + [ + 185, + 54 + ], + [ + 192, + 17 + ], + [ + 211, + 61 + ], + [ + 210, + 3 + ], + [ + 227, + 47 + ] + ] + ] + }, + { + "task_id": "0003207", + "id": "RH20T_cfg4_task_0012_user_0010_scene_0005_cfg_0004_104122062823#220", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push the soccer ball into the goal'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003207/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 233, + 123 + ], + [ + 232, + 90 + ], + [ + 267, + 131 + ], + [ + 200, + 115 + ], + [ + 266, + 160 + ], + [ + 202, + 144 + ] + ], + [ + [ + 247, + 108 + ], + [ + 215, + 101 + ], + [ + 229, + 93 + ], + [ + 262, + 121 + ], + [ + 266, + 99 + ], + [ + 294, + 127 + ] + ], + [ + [ + 260, + 112 + ], + [ + 292, + 121 + ], + [ + 239, + 131 + ], + [ + 278, + 96 + ], + [ + 208, + 121 + ], + [ + 249, + 88 + ] + ], + [ + [ + 251, + 155 + ], + [ + 250, + 166 + ], + [ + 221, + 136 + ], + [ + 280, + 173 + ], + [ + 219, + 122 + ], + [ + 283, + 162 + ] + ] + ] + }, + { + "task_id": "0003213", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0006_cfg_0004_038522062288#367", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003213/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 298, + 270 + ], + [ + 306, + 300 + ], + [ + 310, + 263 + ], + [ + 284, + 278 + ], + [ + 302, + 234 + ], + [ + 275, + 246 + ] + ], + [ + [ + 276, + 246 + ], + [ + 266, + 270 + ], + [ + 294, + 269 + ], + [ + 257, + 223 + ], + [ + 302, + 247 + ], + [ + 267, + 202 + ] + ], + [ + [ + 289, + 245 + ], + [ + 285, + 216 + ], + [ + 289, + 257 + ], + [ + 289, + 235 + ], + [ + 293, + 287 + ], + [ + 293, + 261 + ] + ], + [ + [ + 303, + 266 + ], + [ + 303, + 249 + ], + [ + 271, + 264 + ], + [ + 335, + 268 + ], + [ + 273, + 279 + ], + [ + 334, + 284 + ] + ] + ] + }, + { + "task_id": "0003215", + "id": "RH20T_cfg3_task_0035_user_0010_scene_0001_cfg_0003_104122063550#101", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up one small block'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003215/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 320 + ], + [ + 317, + 317 + ], + [ + 319, + 352 + ], + [ + 310, + 290 + ], + [ + 316, + 353 + ], + [ + 308, + 294 + ] + ], + [ + [ + 324, + 292 + ], + [ + 327, + 316 + ], + [ + 345, + 279 + ], + [ + 304, + 304 + ], + [ + 341, + 256 + ], + [ + 302, + 281 + ] + ], + [ + [ + 316, + 300 + ], + [ + 321, + 272 + ], + [ + 321, + 305 + ], + [ + 311, + 294 + ], + [ + 316, + 333 + ], + [ + 306, + 324 + ] + ], + [ + [ + 296, + 313 + ], + [ + 318, + 323 + ], + [ + 294, + 344 + ], + [ + 299, + 284 + ], + [ + 270, + 333 + ], + [ + 277, + 272 + ] + ] + ] + }, + { + "task_id": "0003218", + "id": "RH20T_cfg6_task_0078_user_0014_scene_0008_cfg_0006_104122061018#134", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close the microwave door'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003218/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 250, + 0 + ], + [ + 216, + 0 + ], + [ + 258, + 0 + ], + [ + 240, + 0 + ], + [ + 290, + 0 + ], + [ + 277, + 0 + ] + ], + [ + [ + 288, + 0 + ], + [ + 313, + 0 + ], + [ + 310, + 0 + ], + [ + 267, + 0 + ], + [ + 282, + 0 + ], + [ + 240, + 0 + ] + ], + [ + [ + 287, + 0 + ], + [ + 279, + 0 + ], + [ + 267, + 0 + ], + [ + 308, + 0 + ], + [ + 274, + 0 + ], + [ + 314, + 0 + ] + ], + [ + [ + 289, + 0 + ], + [ + 287, + 0 + ], + [ + 288, + 0 + ], + [ + 290, + 0 + ], + [ + 290, + 0 + ], + [ + 291, + 0 + ] + ] + ] + }, + { + "task_id": "0003219", + "id": "RH20T_cfg6_task_0078_user_0014_scene_0008_cfg_0006_104122061018#155", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close the microwave door'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003219/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 0 + ], + [ + 310, + 0 + ], + [ + 336, + 0 + ], + [ + 276, + 0 + ], + [ + 333, + 0 + ], + [ + 270, + 0 + ] + ], + [ + [ + 286, + 0 + ], + [ + 266, + 0 + ], + [ + 313, + 0 + ], + [ + 257, + 0 + ], + [ + 334, + 0 + ], + [ + 278, + 0 + ] + ], + [ + [ + 296, + 0 + ], + [ + 287, + 0 + ], + [ + 278, + 0 + ], + [ + 316, + 0 + ], + [ + 287, + 0 + ], + [ + 324, + 0 + ] + ], + [ + [ + 303, + 0 + ], + [ + 303, + 0 + ], + [ + 293, + 0 + ], + [ + 315, + 0 + ], + [ + 293, + 0 + ], + [ + 315, + 0 + ] + ] + ] + }, + { + "task_id": "0003222", + "id": "RH20T_cfg4_task_0081_user_0007_scene_0004_cfg_0004_104122062295#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003222/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 184, + 187 + ], + [ + 180, + 148 + ], + [ + 176, + 201 + ], + [ + 194, + 170 + ], + [ + 181, + 240 + ], + [ + 200, + 213 + ] + ], + [ + [ + 151, + 160 + ], + [ + 161, + 126 + ], + [ + 118, + 136 + ], + [ + 185, + 185 + ], + [ + 111, + 167 + ], + [ + 174, + 214 + ] + ], + [ + [ + 144, + 178 + ], + [ + 166, + 157 + ], + [ + 182, + 182 + ], + [ + 109, + 174 + ], + [ + 160, + 207 + ], + [ + 81, + 197 + ] + ], + [ + [ + 162, + 194 + ], + [ + 179, + 183 + ], + [ + 145, + 236 + ], + [ + 178, + 156 + ], + [ + 131, + 243 + ], + [ + 164, + 167 + ] + ] + ] + }, + { + "task_id": "0003225", + "id": "RH20T_cfg4_task_0081_user_0007_scene_0004_cfg_0004_104122062295#311", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003225/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 247, + 206 + ], + [ + 263, + 171 + ], + [ + 274, + 218 + ], + [ + 217, + 193 + ], + [ + 259, + 252 + ], + [ + 199, + 231 + ] + ], + [ + [ + 275, + 214 + ], + [ + 255, + 217 + ], + [ + 299, + 192 + ], + [ + 249, + 239 + ], + [ + 322, + 188 + ], + [ + 271, + 238 + ] + ], + [ + [ + 244, + 188 + ], + [ + 231, + 153 + ], + [ + 264, + 174 + ], + [ + 222, + 203 + ], + [ + 275, + 206 + ], + [ + 236, + 239 + ] + ], + [ + [ + 245, + 187 + ], + [ + 258, + 155 + ], + [ + 206, + 186 + ], + [ + 281, + 187 + ], + [ + 196, + 216 + ], + [ + 268, + 214 + ] + ] + ] + }, + { + "task_id": "0003226", + "id": "RH20T_cfg1_task_0018_user_0008_scene_0007_cfg_0001_038522062547#147", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003226/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 413, + 93 + ], + [ + 399, + 108 + ], + [ + 428, + 112 + ], + [ + 399, + 74 + ], + [ + 444, + 96 + ], + [ + 414, + 57 + ] + ], + [ + [ + 403, + 100 + ], + [ + 408, + 86 + ], + [ + 427, + 100 + ], + [ + 379, + 100 + ], + [ + 422, + 113 + ], + [ + 377, + 113 + ] + ], + [ + [ + 408, + 75 + ], + [ + 406, + 56 + ], + [ + 385, + 73 + ], + [ + 432, + 76 + ], + [ + 387, + 93 + ], + [ + 435, + 96 + ] + ], + [ + [ + 392, + 130 + ], + [ + 398, + 137 + ], + [ + 372, + 122 + ], + [ + 413, + 137 + ], + [ + 365, + 114 + ], + [ + 407, + 129 + ] + ] + ] + }, + { + "task_id": "0003227", + "id": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_038522063145#101", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003227/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 354, + 63 + ], + [ + 378, + 66 + ], + [ + 356, + 38 + ], + [ + 353, + 87 + ], + [ + 332, + 36 + ], + [ + 330, + 85 + ] + ], + [ + [ + 351, + 62 + ], + [ + 354, + 41 + ], + [ + 358, + 51 + ], + [ + 342, + 73 + ], + [ + 355, + 70 + ], + [ + 340, + 93 + ] + ], + [ + [ + 359, + 85 + ], + [ + 345, + 82 + ], + [ + 366, + 103 + ], + [ + 351, + 65 + ], + [ + 381, + 107 + ], + [ + 367, + 68 + ] + ], + [ + [ + 356, + 85 + ], + [ + 340, + 76 + ], + [ + 347, + 106 + ], + [ + 365, + 62 + ], + [ + 363, + 116 + ], + [ + 383, + 71 + ] + ] + ] + }, + { + "task_id": "0003228", + "id": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_038522063145#209", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003228/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 364, + 42 + ], + [ + 365, + 52 + ], + [ + 383, + 55 + ], + [ + 345, + 29 + ], + [ + 383, + 45 + ], + [ + 343, + 18 + ] + ], + [ + [ + 381, + 70 + ], + [ + 391, + 54 + ], + [ + 390, + 56 + ], + [ + 371, + 84 + ], + [ + 380, + 71 + ], + [ + 362, + 98 + ] + ], + [ + [ + 356, + 71 + ], + [ + 360, + 50 + ], + [ + 369, + 62 + ], + [ + 342, + 80 + ], + [ + 365, + 81 + ], + [ + 338, + 100 + ] + ], + [ + [ + 368, + 65 + ], + [ + 371, + 89 + ], + [ + 350, + 75 + ], + [ + 388, + 55 + ], + [ + 348, + 53 + ], + [ + 385, + 31 + ] + ] + ] + }, + { + "task_id": "0003230", + "id": "RH20T_cfg7_task_0013_user_0020_scene_0010_cfg_0007_104122061330#189", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003230/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 354, + 271 + ], + [ + 336, + 294 + ], + [ + 338, + 256 + ], + [ + 370, + 284 + ], + [ + 356, + 232 + ], + [ + 387, + 262 + ] + ], + [ + [ + 334, + 253 + ], + [ + 312, + 248 + ], + [ + 340, + 277 + ], + [ + 328, + 230 + ], + [ + 362, + 281 + ], + [ + 348, + 235 + ] + ], + [ + [ + 322, + 256 + ], + [ + 309, + 265 + ], + [ + 346, + 253 + ], + [ + 300, + 259 + ], + [ + 357, + 245 + ], + [ + 312, + 251 + ] + ], + [ + [ + 339, + 249 + ], + [ + 336, + 236 + ], + [ + 326, + 274 + ], + [ + 351, + 226 + ], + [ + 329, + 285 + ], + [ + 353, + 239 + ] + ] + ] + }, + { + "task_id": "0003232", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_038522063145#93", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003232/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 558, + 265 + ], + [ + 534, + 247 + ], + [ + 568, + 244 + ], + [ + 549, + 284 + ], + [ + 596, + 264 + ], + [ + 575, + 303 + ] + ], + [ + [ + 530, + 271 + ], + [ + 530, + 241 + ], + [ + 553, + 271 + ], + [ + 510, + 271 + ], + [ + 553, + 304 + ], + [ + 510, + 301 + ] + ], + [ + [ + 519, + 258 + ], + [ + 530, + 243 + ], + [ + 544, + 254 + ], + [ + 493, + 262 + ], + [ + 533, + 267 + ], + [ + 484, + 275 + ] + ], + [ + [ + 530, + 245 + ], + [ + 514, + 225 + ], + [ + 550, + 242 + ], + [ + 509, + 249 + ], + [ + 567, + 264 + ], + [ + 525, + 272 + ] + ] + ] + }, + { + "task_id": "0003234", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_038522063145#200", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003234/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 517, + 239 + ], + [ + 526, + 224 + ], + [ + 543, + 236 + ], + [ + 491, + 241 + ], + [ + 533, + 250 + ], + [ + 484, + 255 + ] + ], + [ + [ + 524, + 208 + ], + [ + 507, + 211 + ], + [ + 513, + 184 + ], + [ + 535, + 234 + ], + [ + 528, + 182 + ], + [ + 550, + 231 + ] + ], + [ + [ + 520, + 213 + ], + [ + 510, + 192 + ], + [ + 506, + 204 + ], + [ + 533, + 222 + ], + [ + 517, + 228 + ], + [ + 543, + 244 + ] + ], + [ + [ + 532, + 257 + ], + [ + 551, + 272 + ], + [ + 545, + 236 + ], + [ + 519, + 278 + ], + [ + 526, + 220 + ], + [ + 499, + 263 + ] + ] + ] + }, + { + "task_id": "0003235", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_038522063145#245", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003235/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 514, + 201 + ], + [ + 498, + 215 + ], + [ + 516, + 180 + ], + [ + 512, + 221 + ], + [ + 532, + 168 + ], + [ + 527, + 208 + ] + ], + [ + [ + 511, + 220 + ], + [ + 520, + 204 + ], + [ + 536, + 217 + ], + [ + 486, + 223 + ], + [ + 527, + 232 + ], + [ + 480, + 238 + ] + ], + [ + [ + 510, + 215 + ], + [ + 498, + 200 + ], + [ + 534, + 205 + ], + [ + 486, + 225 + ], + [ + 549, + 221 + ], + [ + 499, + 242 + ] + ], + [ + [ + 479, + 218 + ], + [ + 492, + 199 + ], + [ + 489, + 232 + ], + [ + 468, + 203 + ], + [ + 475, + 251 + ], + [ + 454, + 222 + ] + ] + ] + }, + { + "task_id": "0003236", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_038522063145#299", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003236/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 520, + 188 + ], + [ + 502, + 180 + ], + [ + 535, + 166 + ], + [ + 505, + 209 + ], + [ + 555, + 174 + ], + [ + 523, + 219 + ] + ], + [ + [ + 502, + 205 + ], + [ + 509, + 188 + ], + [ + 526, + 201 + ], + [ + 478, + 208 + ], + [ + 518, + 217 + ], + [ + 472, + 223 + ] + ], + [ + [ + 509, + 216 + ], + [ + 535, + 219 + ], + [ + 515, + 193 + ], + [ + 504, + 238 + ], + [ + 489, + 190 + ], + [ + 480, + 235 + ] + ], + [ + [ + 513, + 187 + ], + [ + 516, + 203 + ], + [ + 487, + 187 + ], + [ + 540, + 187 + ], + [ + 482, + 171 + ], + [ + 537, + 169 + ] + ] + ] + }, + { + "task_id": "0003237", + "id": "RH20T_cfg5_task_0010_user_0010_scene_0005_cfg_0005_036422060215#370", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003237/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 335, + 163 + ], + [ + 332, + 144 + ], + [ + 336, + 172 + ], + [ + 333, + 153 + ], + [ + 339, + 190 + ], + [ + 336, + 170 + ] + ], + [ + [ + 322, + 175 + ], + [ + 330, + 188 + ], + [ + 331, + 186 + ], + [ + 314, + 164 + ], + [ + 322, + 172 + ], + [ + 305, + 150 + ] + ], + [ + [ + 314, + 168 + ], + [ + 300, + 181 + ], + [ + 314, + 158 + ], + [ + 314, + 179 + ], + [ + 329, + 145 + ], + [ + 330, + 166 + ] + ], + [ + [ + 323, + 171 + ], + [ + 331, + 160 + ], + [ + 340, + 169 + ], + [ + 305, + 174 + ], + [ + 333, + 179 + ], + [ + 299, + 185 + ] + ] + ] + }, + { + "task_id": "0003238", + "id": "RH20T_cfg1_task_0016_user_0014_scene_0003_cfg_0001_819112071037#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003238/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 510, + 154 + ], + [ + 568, + 158 + ], + [ + 514, + 112 + ], + [ + 506, + 189 + ], + [ + 456, + 109 + ], + [ + 456, + 184 + ] + ], + [ + [ + 584, + 175 + ], + [ + 554, + 183 + ], + [ + 543, + 187 + ], + [ + 633, + 160 + ], + [ + 570, + 180 + ], + [ + 640, + 153 + ] + ], + [ + [ + 577, + 203 + ], + [ + 591, + 167 + ], + [ + 617, + 182 + ], + [ + 544, + 220 + ], + [ + 603, + 225 + ], + [ + 529, + 256 + ] + ], + [ + [ + 546, + 148 + ], + [ + 550, + 100 + ], + [ + 573, + 167 + ], + [ + 525, + 133 + ], + [ + 568, + 215 + ], + [ + 522, + 174 + ] + ] + ] + }, + { + "task_id": "0003240", + "id": "RH20T_cfg7_task_0013_user_0020_scene_0004_cfg_0007_104122061330#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003240/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 346, + 138 + ], + [ + 371, + 133 + ], + [ + 349, + 158 + ], + [ + 343, + 118 + ], + [ + 323, + 163 + ], + [ + 319, + 123 + ] + ], + [ + [ + 350, + 132 + ], + [ + 362, + 153 + ], + [ + 330, + 145 + ], + [ + 369, + 119 + ], + [ + 317, + 123 + ], + [ + 357, + 98 + ] + ], + [ + [ + 344, + 114 + ], + [ + 334, + 91 + ], + [ + 324, + 120 + ], + [ + 364, + 108 + ], + [ + 335, + 143 + ], + [ + 374, + 130 + ] + ], + [ + [ + 351, + 137 + ], + [ + 347, + 124 + ], + [ + 340, + 159 + ], + [ + 362, + 116 + ], + [ + 345, + 169 + ], + [ + 366, + 128 + ] + ] + ] + }, + { + "task_id": "0003246", + "id": "RH20T_cfg1_task_0001_user_0014_scene_0006_cfg_0001_039422060546#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003246/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 59 + ], + [ + 355, + 63 + ], + [ + 331, + 35 + ], + [ + 320, + 85 + ], + [ + 303, + 32 + ], + [ + 291, + 81 + ] + ], + [ + [ + 311, + 81 + ], + [ + 291, + 73 + ], + [ + 326, + 85 + ], + [ + 294, + 76 + ], + [ + 347, + 93 + ], + [ + 315, + 85 + ] + ], + [ + [ + 347, + 69 + ], + [ + 328, + 45 + ], + [ + 364, + 52 + ], + [ + 329, + 87 + ], + [ + 382, + 75 + ], + [ + 348, + 111 + ] + ], + [ + [ + 321, + 80 + ], + [ + 319, + 56 + ], + [ + 316, + 99 + ], + [ + 326, + 61 + ], + [ + 318, + 121 + ], + [ + 328, + 83 + ] + ] + ] + }, + { + "task_id": "0003247", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0005_cfg_0006_104122061330#164", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003247/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 370, + 46 + ], + [ + 350, + 65 + ], + [ + 349, + 30 + ], + [ + 391, + 63 + ], + [ + 369, + 11 + ], + [ + 412, + 43 + ] + ], + [ + [ + 355, + 96 + ], + [ + 364, + 107 + ], + [ + 377, + 104 + ], + [ + 331, + 88 + ], + [ + 368, + 94 + ], + [ + 323, + 78 + ] + ], + [ + [ + 356, + 67 + ], + [ + 347, + 57 + ], + [ + 344, + 93 + ], + [ + 368, + 41 + ], + [ + 353, + 101 + ], + [ + 375, + 51 + ] + ], + [ + [ + 347, + 64 + ], + [ + 357, + 40 + ], + [ + 366, + 81 + ], + [ + 325, + 46 + ], + [ + 357, + 105 + ], + [ + 314, + 71 + ] + ] + ] + }, + { + "task_id": "0003248", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0005_cfg_0006_104122061330#452", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003248/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 298, + 249 + ], + [ + 289, + 241 + ], + [ + 285, + 276 + ], + [ + 311, + 223 + ], + [ + 295, + 282 + ], + [ + 319, + 231 + ] + ], + [ + [ + 327, + 236 + ], + [ + 306, + 218 + ], + [ + 335, + 237 + ], + [ + 319, + 236 + ], + [ + 357, + 257 + ], + [ + 338, + 253 + ] + ], + [ + [ + 293, + 279 + ], + [ + 309, + 304 + ], + [ + 285, + 297 + ], + [ + 302, + 260 + ], + [ + 270, + 274 + ], + [ + 286, + 236 + ] + ], + [ + [ + 305, + 220 + ], + [ + 300, + 212 + ], + [ + 327, + 238 + ], + [ + 284, + 203 + ], + [ + 330, + 244 + ], + [ + 289, + 211 + ] + ] + ] + }, + { + "task_id": "0003250", + "id": "RH20T_cfg7_task_0015_user_0014_scene_0008_cfg_0007_037522061512#149", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003250/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 267, + 97 + ], + [ + 249, + 94 + ], + [ + 273, + 78 + ], + [ + 260, + 116 + ], + [ + 292, + 81 + ], + [ + 278, + 120 + ] + ], + [ + [ + 252, + 109 + ], + [ + 246, + 89 + ], + [ + 264, + 101 + ], + [ + 238, + 118 + ], + [ + 270, + 120 + ], + [ + 245, + 138 + ] + ], + [ + [ + 254, + 124 + ], + [ + 238, + 138 + ], + [ + 259, + 130 + ], + [ + 250, + 118 + ], + [ + 276, + 115 + ], + [ + 265, + 104 + ] + ], + [ + [ + 243, + 93 + ], + [ + 227, + 85 + ], + [ + 258, + 82 + ], + [ + 229, + 104 + ], + [ + 272, + 90 + ], + [ + 243, + 111 + ] + ] + ] + }, + { + "task_id": "0003251", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_104122062823#102", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003251/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 429, + 81 + ], + [ + 430, + 52 + ], + [ + 456, + 70 + ], + [ + 401, + 91 + ], + [ + 453, + 96 + ], + [ + 400, + 117 + ] + ], + [ + [ + 416, + 98 + ], + [ + 396, + 77 + ], + [ + 404, + 99 + ], + [ + 426, + 97 + ], + [ + 426, + 121 + ], + [ + 446, + 117 + ] + ], + [ + [ + 470, + 75 + ], + [ + 493, + 91 + ], + [ + 497, + 49 + ], + [ + 445, + 99 + ], + [ + 475, + 35 + ], + [ + 427, + 83 + ] + ], + [ + [ + 409, + 89 + ], + [ + 440, + 86 + ], + [ + 418, + 114 + ], + [ + 399, + 63 + ], + [ + 391, + 115 + ], + [ + 370, + 67 + ] + ] + ] + }, + { + "task_id": "0003252", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_104122062823#175", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003252/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 462, + 45 + ], + [ + 440, + 31 + ], + [ + 438, + 67 + ], + [ + 485, + 23 + ], + [ + 461, + 82 + ], + [ + 509, + 37 + ] + ], + [ + [ + 433, + 97 + ], + [ + 404, + 101 + ], + [ + 439, + 77 + ], + [ + 428, + 114 + ], + [ + 469, + 72 + ], + [ + 455, + 110 + ] + ], + [ + [ + 451, + 73 + ], + [ + 461, + 89 + ], + [ + 427, + 91 + ], + [ + 476, + 53 + ], + [ + 419, + 77 + ], + [ + 465, + 39 + ] + ], + [ + [ + 437, + 76 + ], + [ + 442, + 52 + ], + [ + 467, + 71 + ], + [ + 408, + 81 + ], + [ + 461, + 93 + ], + [ + 405, + 103 + ] + ] + ] + }, + { + "task_id": "0003253", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_104122062823#327", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003253/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 330, + 80 + ], + [ + 337, + 76 + ], + [ + 315, + 51 + ], + [ + 345, + 107 + ], + [ + 310, + 56 + ], + [ + 338, + 109 + ] + ], + [ + [ + 364, + 57 + ], + [ + 361, + 83 + ], + [ + 366, + 36 + ], + [ + 363, + 76 + ], + [ + 370, + 11 + ], + [ + 366, + 53 + ] + ], + [ + [ + 351, + 63 + ], + [ + 349, + 39 + ], + [ + 380, + 70 + ], + [ + 323, + 57 + ], + [ + 379, + 92 + ], + [ + 325, + 78 + ] + ], + [ + [ + 382, + 77 + ], + [ + 353, + 78 + ], + [ + 383, + 102 + ], + [ + 381, + 53 + ], + [ + 413, + 101 + ], + [ + 410, + 51 + ] + ] + ] + }, + { + "task_id": "0003255", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0001_cfg_0005_104122061850#408", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003255/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 331, + 185 + ], + [ + 336, + 169 + ], + [ + 326, + 164 + ], + [ + 336, + 204 + ], + [ + 321, + 180 + ], + [ + 332, + 221 + ] + ], + [ + [ + 326, + 166 + ], + [ + 318, + 151 + ], + [ + 303, + 179 + ], + [ + 349, + 153 + ], + [ + 311, + 192 + ], + [ + 355, + 168 + ] + ], + [ + [ + 313, + 144 + ], + [ + 312, + 127 + ], + [ + 329, + 128 + ], + [ + 297, + 161 + ], + [ + 330, + 143 + ], + [ + 299, + 176 + ] + ], + [ + [ + 352, + 181 + ], + [ + 336, + 181 + ], + [ + 330, + 185 + ], + [ + 372, + 177 + ], + [ + 347, + 185 + ], + [ + 390, + 177 + ] + ] + ] + }, + { + "task_id": "0003256", + "id": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_750612070853#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003256/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 329, + 267 + ], + [ + 326, + 317 + ], + [ + 302, + 246 + ], + [ + 352, + 285 + ], + [ + 306, + 200 + ], + [ + 353, + 243 + ] + ], + [ + [ + 337, + 266 + ], + [ + 326, + 223 + ], + [ + 305, + 251 + ], + [ + 376, + 284 + ], + [ + 316, + 288 + ], + [ + 385, + 325 + ] + ], + [ + [ + 345, + 262 + ], + [ + 340, + 213 + ], + [ + 288, + 263 + ], + [ + 404, + 262 + ], + [ + 295, + 306 + ], + [ + 405, + 307 + ] + ], + [ + [ + 329, + 218 + ], + [ + 314, + 247 + ], + [ + 352, + 179 + ], + [ + 308, + 253 + ], + [ + 362, + 157 + ], + [ + 321, + 227 + ] + ] + ] + }, + { + "task_id": "0003258", + "id": "RH20T_cfg1_task_0033_user_0015_scene_0008_cfg_0001_750612070853#225", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003258/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 444, + 208 + ], + [ + 441, + 165 + ], + [ + 393, + 214 + ], + [ + 497, + 203 + ], + [ + 398, + 252 + ], + [ + 497, + 243 + ] + ], + [ + [ + 420, + 246 + ], + [ + 445, + 209 + ], + [ + 425, + 273 + ], + [ + 416, + 223 + ], + [ + 400, + 310 + ], + [ + 395, + 256 + ] + ], + [ + [ + 374, + 224 + ], + [ + 333, + 235 + ], + [ + 391, + 264 + ], + [ + 358, + 187 + ], + [ + 430, + 252 + ], + [ + 395, + 178 + ] + ], + [ + [ + 394, + 229 + ], + [ + 390, + 188 + ], + [ + 359, + 213 + ], + [ + 435, + 247 + ], + [ + 364, + 248 + ], + [ + 437, + 285 + ] + ] + ] + }, + { + "task_id": "0003263", + "id": "RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_104122061602#177", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003263/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 426, + 115 + ], + [ + 446, + 107 + ], + [ + 421, + 125 + ], + [ + 431, + 105 + ], + [ + 401, + 132 + ], + [ + 410, + 113 + ] + ], + [ + [ + 428, + 106 + ], + [ + 416, + 86 + ], + [ + 416, + 115 + ], + [ + 440, + 96 + ], + [ + 428, + 135 + ], + [ + 451, + 115 + ] + ], + [ + [ + 408, + 91 + ], + [ + 386, + 87 + ], + [ + 416, + 90 + ], + [ + 402, + 92 + ], + [ + 439, + 94 + ], + [ + 423, + 96 + ] + ], + [ + [ + 416, + 99 + ], + [ + 425, + 120 + ], + [ + 406, + 109 + ], + [ + 426, + 89 + ], + [ + 398, + 90 + ], + [ + 416, + 68 + ] + ] + ] + }, + { + "task_id": "0003265", + "id": "RH20T_cfg1_task_0005_user_0015_scene_0007_cfg_0001_750612070853#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003265/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 366, + 145 + ], + [ + 338, + 173 + ], + [ + 328, + 124 + ], + [ + 406, + 168 + ], + [ + 354, + 101 + ], + [ + 430, + 140 + ] + ], + [ + [ + 392, + 162 + ], + [ + 405, + 128 + ], + [ + 355, + 141 + ], + [ + 432, + 184 + ], + [ + 347, + 171 + ], + [ + 418, + 213 + ] + ], + [ + [ + 406, + 117 + ], + [ + 393, + 79 + ], + [ + 382, + 148 + ], + [ + 429, + 88 + ], + [ + 396, + 180 + ], + [ + 439, + 121 + ] + ], + [ + [ + 360, + 127 + ], + [ + 380, + 104 + ], + [ + 358, + 162 + ], + [ + 362, + 86 + ], + [ + 336, + 189 + ], + [ + 338, + 112 + ] + ] + ] + }, + { + "task_id": "0003266", + "id": "RH20T_cfg5_task_0033_user_0010_scene_0010_cfg_0005_037522062165#232", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003266/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 231, + 214 + ], + [ + 235, + 197 + ], + [ + 243, + 197 + ], + [ + 220, + 229 + ], + [ + 240, + 215 + ], + [ + 215, + 248 + ] + ], + [ + [ + 247, + 200 + ], + [ + 245, + 187 + ], + [ + 229, + 184 + ], + [ + 265, + 218 + ], + [ + 232, + 196 + ], + [ + 266, + 229 + ] + ], + [ + [ + 237, + 208 + ], + [ + 233, + 234 + ], + [ + 258, + 213 + ], + [ + 215, + 202 + ], + [ + 261, + 189 + ], + [ + 220, + 175 + ] + ], + [ + [ + 254, + 210 + ], + [ + 231, + 227 + ], + [ + 247, + 209 + ], + [ + 262, + 211 + ], + [ + 267, + 193 + ], + [ + 285, + 194 + ] + ] + ] + }, + { + "task_id": "0003269", + "id": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_037522061512#288", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the pen into the pen holder'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003269/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 321, + 190 + ], + [ + 333, + 182 + ], + [ + 343, + 195 + ], + [ + 301, + 185 + ], + [ + 332, + 203 + ], + [ + 288, + 192 + ] + ], + [ + [ + 291, + 167 + ], + [ + 284, + 184 + ], + [ + 279, + 150 + ], + [ + 304, + 185 + ], + [ + 286, + 131 + ], + [ + 312, + 167 + ] + ], + [ + [ + 282, + 176 + ], + [ + 262, + 166 + ], + [ + 288, + 179 + ], + [ + 276, + 171 + ], + [ + 308, + 188 + ], + [ + 298, + 181 + ] + ], + [ + [ + 291, + 191 + ], + [ + 299, + 168 + ], + [ + 300, + 196 + ], + [ + 281, + 187 + ], + [ + 292, + 218 + ], + [ + 272, + 211 + ] + ] + ] + }, + { + "task_id": "0003272", + "id": "RH20T_cfg1_task_0038_user_0016_scene_0009_cfg_0001_035622060973#120", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003272/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 211, + 160 + ], + [ + 232, + 164 + ], + [ + 206, + 178 + ], + [ + 215, + 143 + ], + [ + 184, + 174 + ], + [ + 194, + 140 + ] + ], + [ + [ + 228, + 149 + ], + [ + 229, + 135 + ], + [ + 216, + 134 + ], + [ + 241, + 164 + ], + [ + 216, + 147 + ], + [ + 241, + 176 + ] + ], + [ + [ + 220, + 150 + ], + [ + 240, + 144 + ], + [ + 225, + 168 + ], + [ + 214, + 131 + ], + [ + 205, + 173 + ], + [ + 192, + 136 + ] + ], + [ + [ + 213, + 143 + ], + [ + 235, + 138 + ], + [ + 212, + 124 + ], + [ + 213, + 160 + ], + [ + 188, + 128 + ], + [ + 191, + 165 + ] + ] + ] + }, + { + "task_id": "0003273", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_038522063145#91", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003273/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 555, + 177 + ], + [ + 542, + 199 + ], + [ + 579, + 183 + ], + [ + 531, + 170 + ], + [ + 594, + 160 + ], + [ + 546, + 146 + ] + ], + [ + [ + 540, + 180 + ], + [ + 552, + 160 + ], + [ + 569, + 190 + ], + [ + 513, + 171 + ], + [ + 557, + 208 + ], + [ + 503, + 189 + ] + ], + [ + [ + 530, + 164 + ], + [ + 542, + 182 + ], + [ + 504, + 178 + ], + [ + 557, + 151 + ], + [ + 494, + 162 + ], + [ + 544, + 135 + ] + ], + [ + [ + 518, + 188 + ], + [ + 546, + 193 + ], + [ + 511, + 214 + ], + [ + 524, + 160 + ], + [ + 483, + 209 + ], + [ + 495, + 155 + ] + ] + ] + }, + { + "task_id": "0003274", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_038522063145#267", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003274/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 451, + 96 + ], + [ + 432, + 95 + ], + [ + 461, + 76 + ], + [ + 442, + 118 + ], + [ + 483, + 76 + ], + [ + 463, + 120 + ] + ], + [ + [ + 418, + 122 + ], + [ + 413, + 107 + ], + [ + 404, + 110 + ], + [ + 431, + 133 + ], + [ + 408, + 126 + ], + [ + 437, + 148 + ] + ], + [ + [ + 437, + 116 + ], + [ + 444, + 98 + ], + [ + 455, + 130 + ], + [ + 420, + 103 + ], + [ + 448, + 147 + ], + [ + 415, + 120 + ] + ], + [ + [ + 432, + 101 + ], + [ + 416, + 111 + ], + [ + 447, + 93 + ], + [ + 419, + 109 + ], + [ + 464, + 84 + ], + [ + 434, + 100 + ] + ] + ] + }, + { + "task_id": "0003276", + "id": "RH20T_cfg5_task_0013_user_0007_scene_0001_cfg_0005_104422070044#149", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003276/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 129 + ], + [ + 381, + 141 + ], + [ + 375, + 134 + ], + [ + 367, + 123 + ], + [ + 366, + 122 + ], + [ + 358, + 110 + ] + ], + [ + [ + 372, + 138 + ], + [ + 371, + 125 + ], + [ + 386, + 143 + ], + [ + 359, + 132 + ], + [ + 388, + 156 + ], + [ + 361, + 145 + ] + ], + [ + [ + 370, + 134 + ], + [ + 381, + 142 + ], + [ + 368, + 148 + ], + [ + 372, + 120 + ], + [ + 356, + 140 + ], + [ + 361, + 111 + ] + ], + [ + [ + 366, + 146 + ], + [ + 375, + 137 + ], + [ + 356, + 133 + ], + [ + 376, + 159 + ], + [ + 346, + 143 + ], + [ + 367, + 169 + ] + ] + ] + }, + { + "task_id": "0003277", + "id": "RH20T_cfg1_task_0065_user_0014_scene_0004_cfg_0001_039422060546#72", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003277/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 311 + ], + [ + 274, + 283 + ], + [ + 248, + 332 + ], + [ + 297, + 293 + ], + [ + 249, + 360 + ], + [ + 295, + 319 + ] + ], + [ + [ + 286, + 281 + ], + [ + 283, + 265 + ], + [ + 275, + 315 + ], + [ + 297, + 249 + ], + [ + 279, + 328 + ], + [ + 299, + 265 + ] + ], + [ + [ + 309, + 299 + ], + [ + 282, + 285 + ], + [ + 322, + 266 + ], + [ + 295, + 332 + ], + [ + 351, + 279 + ], + [ + 324, + 349 + ] + ], + [ + [ + 268, + 275 + ], + [ + 291, + 247 + ], + [ + 296, + 288 + ], + [ + 238, + 260 + ], + [ + 273, + 314 + ], + [ + 214, + 288 + ] + ] + ] + }, + { + "task_id": "0003286", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0002_cfg_0006_037522061512#355", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003286/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 231, + 146 + ], + [ + 240, + 120 + ], + [ + 246, + 149 + ], + [ + 215, + 143 + ], + [ + 238, + 173 + ], + [ + 206, + 170 + ] + ], + [ + [ + 233, + 158 + ], + [ + 248, + 142 + ], + [ + 235, + 179 + ], + [ + 231, + 139 + ], + [ + 220, + 194 + ], + [ + 217, + 155 + ] + ], + [ + [ + 201, + 158 + ], + [ + 179, + 155 + ], + [ + 186, + 151 + ], + [ + 214, + 165 + ], + [ + 210, + 155 + ], + [ + 236, + 169 + ] + ], + [ + [ + 221, + 150 + ], + [ + 226, + 133 + ], + [ + 217, + 169 + ], + [ + 224, + 130 + ], + [ + 212, + 188 + ], + [ + 219, + 150 + ] + ] + ] + }, + { + "task_id": "0003287", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0002_cfg_0006_037522061512#466", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003287/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 240, + 89 + ], + [ + 251, + 66 + ], + [ + 231, + 77 + ], + [ + 249, + 102 + ], + [ + 222, + 98 + ], + [ + 238, + 124 + ] + ], + [ + [ + 210, + 87 + ], + [ + 217, + 71 + ], + [ + 196, + 101 + ], + [ + 225, + 73 + ], + [ + 188, + 118 + ], + [ + 217, + 91 + ] + ], + [ + [ + 239, + 83 + ], + [ + 221, + 78 + ], + [ + 219, + 86 + ], + [ + 261, + 79 + ], + [ + 236, + 90 + ], + [ + 277, + 84 + ] + ], + [ + [ + 224, + 97 + ], + [ + 216, + 77 + ], + [ + 239, + 81 + ], + [ + 209, + 113 + ], + [ + 245, + 100 + ], + [ + 217, + 132 + ] + ] + ] + }, + { + "task_id": "0003288", + "id": "RH20T_cfg1_task_0119_user_0014_scene_0004_cfg_0001_035622060973#107", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn off the water tap'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003288/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 203, + 179 + ], + [ + 192, + 180 + ], + [ + 188, + 190 + ], + [ + 217, + 167 + ], + [ + 201, + 189 + ], + [ + 230, + 166 + ] + ], + [ + [ + 204, + 191 + ], + [ + 199, + 180 + ], + [ + 187, + 181 + ], + [ + 222, + 202 + ], + [ + 193, + 192 + ], + [ + 226, + 212 + ] + ], + [ + [ + 213, + 180 + ], + [ + 225, + 162 + ], + [ + 229, + 191 + ], + [ + 199, + 169 + ], + [ + 217, + 209 + ], + [ + 187, + 187 + ] + ], + [ + [ + 217, + 210 + ], + [ + 200, + 225 + ], + [ + 202, + 195 + ], + [ + 232, + 226 + ], + [ + 219, + 181 + ], + [ + 248, + 212 + ] + ] + ] + }, + { + "task_id": "0003293", + "id": "RH20T_cfg1_task_0043_user_0016_scene_0002_cfg_0001_038522062547#130", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003293/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 359, + 173 + ], + [ + 331, + 170 + ], + [ + 361, + 145 + ], + [ + 356, + 201 + ], + [ + 389, + 148 + ], + [ + 384, + 203 + ] + ], + [ + [ + 333, + 166 + ], + [ + 359, + 179 + ], + [ + 334, + 149 + ], + [ + 332, + 181 + ], + [ + 308, + 135 + ], + [ + 308, + 168 + ] + ], + [ + [ + 350, + 167 + ], + [ + 350, + 155 + ], + [ + 359, + 193 + ], + [ + 341, + 143 + ], + [ + 358, + 203 + ], + [ + 342, + 155 + ] + ], + [ + [ + 338, + 164 + ], + [ + 362, + 158 + ], + [ + 337, + 184 + ], + [ + 340, + 142 + ], + [ + 313, + 190 + ], + [ + 314, + 148 + ] + ] + ] + }, + { + "task_id": "0003296", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_038522062547#439", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003296/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 32 + ], + [ + 427, + 39 + ], + [ + 395, + 44 + ], + [ + 419, + 20 + ], + [ + 374, + 38 + ], + [ + 397, + 13 + ] + ], + [ + [ + 403, + 50 + ], + [ + 407, + 35 + ], + [ + 410, + 70 + ], + [ + 396, + 31 + ], + [ + 407, + 83 + ], + [ + 393, + 45 + ] + ], + [ + [ + 414, + 40 + ], + [ + 435, + 39 + ], + [ + 406, + 25 + ], + [ + 422, + 56 + ], + [ + 385, + 26 + ], + [ + 400, + 57 + ] + ], + [ + [ + 381, + 28 + ], + [ + 397, + 38 + ], + [ + 363, + 44 + ], + [ + 399, + 12 + ], + [ + 349, + 34 + ], + [ + 384, + 2 + ] + ] + ] + }, + { + "task_id": "0003297", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_038522062547#656", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003297/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 447, + 206 + ], + [ + 452, + 197 + ], + [ + 472, + 222 + ], + [ + 423, + 192 + ], + [ + 467, + 230 + ], + [ + 420, + 201 + ] + ], + [ + [ + 452, + 237 + ], + [ + 473, + 245 + ], + [ + 471, + 221 + ], + [ + 435, + 251 + ], + [ + 448, + 212 + ], + [ + 413, + 244 + ] + ], + [ + [ + 480, + 201 + ], + [ + 458, + 207 + ], + [ + 461, + 183 + ], + [ + 501, + 221 + ], + [ + 481, + 178 + ], + [ + 522, + 215 + ] + ], + [ + [ + 446, + 211 + ], + [ + 452, + 233 + ], + [ + 467, + 196 + ], + [ + 426, + 224 + ], + [ + 461, + 176 + ], + [ + 422, + 204 + ] + ] + ] + }, + { + "task_id": "0003299", + "id": "RH20T_cfg4_task_0024_user_0014_scene_0003_cfg_0004_104122063550#70", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn off the power strip by pressing the button'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003299/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 248 + ], + [ + 267, + 272 + ], + [ + 260, + 236 + ], + [ + 310, + 261 + ], + [ + 276, + 215 + ], + [ + 325, + 238 + ] + ], + [ + [ + 297, + 258 + ], + [ + 293, + 241 + ], + [ + 322, + 270 + ], + [ + 273, + 247 + ], + [ + 325, + 286 + ], + [ + 278, + 264 + ] + ], + [ + [ + 290, + 238 + ], + [ + 299, + 223 + ], + [ + 283, + 257 + ], + [ + 297, + 218 + ], + [ + 273, + 274 + ], + [ + 286, + 234 + ] + ], + [ + [ + 281, + 252 + ], + [ + 299, + 232 + ], + [ + 263, + 229 + ], + [ + 299, + 273 + ], + [ + 243, + 250 + ], + [ + 280, + 295 + ] + ] + ] + }, + { + "task_id": "0003300", + "id": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_750612070851#88", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003300/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 484, + 162 + ], + [ + 498, + 136 + ], + [ + 510, + 191 + ], + [ + 459, + 135 + ], + [ + 495, + 213 + ], + [ + 449, + 159 + ] + ], + [ + [ + 529, + 143 + ], + [ + 531, + 181 + ], + [ + 516, + 157 + ], + [ + 541, + 130 + ], + [ + 514, + 114 + ], + [ + 540, + 91 + ] + ], + [ + [ + 468, + 155 + ], + [ + 506, + 146 + ], + [ + 471, + 167 + ], + [ + 464, + 140 + ], + [ + 436, + 175 + ], + [ + 425, + 149 + ] + ], + [ + [ + 475, + 117 + ], + [ + 454, + 130 + ], + [ + 447, + 90 + ], + [ + 502, + 145 + ], + [ + 469, + 72 + ], + [ + 529, + 132 + ] + ] + ] + }, + { + "task_id": "0003301", + "id": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_750612070851#203", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003301/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 451, + 148 + ], + [ + 462, + 121 + ], + [ + 466, + 179 + ], + [ + 438, + 120 + ], + [ + 455, + 202 + ], + [ + 429, + 144 + ] + ], + [ + [ + 443, + 184 + ], + [ + 452, + 153 + ], + [ + 478, + 182 + ], + [ + 407, + 187 + ], + [ + 468, + 211 + ], + [ + 401, + 216 + ] + ], + [ + [ + 486, + 134 + ], + [ + 518, + 111 + ], + [ + 486, + 104 + ], + [ + 486, + 160 + ], + [ + 451, + 130 + ], + [ + 456, + 182 + ] + ], + [ + [ + 442, + 135 + ], + [ + 412, + 156 + ], + [ + 438, + 154 + ], + [ + 446, + 112 + ], + [ + 465, + 135 + ], + [ + 477, + 91 + ] + ] + ] + }, + { + "task_id": "0003305", + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_104422070044#223", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003305/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 292, + 186 + ], + [ + 289, + 173 + ], + [ + 301, + 196 + ], + [ + 284, + 177 + ], + [ + 303, + 209 + ], + [ + 286, + 190 + ] + ], + [ + [ + 307, + 198 + ], + [ + 303, + 202 + ], + [ + 307, + 215 + ], + [ + 307, + 181 + ], + [ + 310, + 211 + ], + [ + 311, + 178 + ] + ], + [ + [ + 281, + 183 + ], + [ + 298, + 177 + ], + [ + 279, + 176 + ], + [ + 284, + 190 + ], + [ + 262, + 182 + ], + [ + 268, + 196 + ] + ], + [ + [ + 287, + 176 + ], + [ + 273, + 175 + ], + [ + 277, + 182 + ], + [ + 299, + 169 + ], + [ + 290, + 182 + ], + [ + 313, + 169 + ] + ] + ] + }, + { + "task_id": "0003307", + "id": "RH20T_cfg5_task_0096_user_0007_scene_0003_cfg_0005_105422061350#292", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003307/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 261, + 77 + ], + [ + 262, + 54 + ], + [ + 293, + 80 + ], + [ + 229, + 73 + ], + [ + 291, + 100 + ], + [ + 231, + 94 + ] + ], + [ + [ + 233, + 40 + ], + [ + 213, + 25 + ], + [ + 218, + 20 + ], + [ + 249, + 60 + ], + [ + 236, + 34 + ], + [ + 266, + 73 + ] + ], + [ + [ + 272, + 43 + ], + [ + 290, + 64 + ], + [ + 262, + 67 + ], + [ + 282, + 15 + ], + [ + 246, + 47 + ], + [ + 263, + 0 + ] + ], + [ + [ + 268, + 55 + ], + [ + 271, + 30 + ], + [ + 260, + 33 + ], + [ + 275, + 80 + ], + [ + 258, + 56 + ], + [ + 271, + 101 + ] + ] + ] + }, + { + "task_id": "0003309", + "id": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_036422060909#175", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003309/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 222, + 153 + ], + [ + 204, + 171 + ], + [ + 223, + 155 + ], + [ + 220, + 152 + ], + [ + 241, + 137 + ], + [ + 237, + 136 + ] + ], + [ + [ + 178, + 130 + ], + [ + 159, + 148 + ], + [ + 197, + 149 + ], + [ + 159, + 110 + ], + [ + 215, + 132 + ], + [ + 178, + 93 + ] + ], + [ + [ + 202, + 140 + ], + [ + 198, + 122 + ], + [ + 224, + 131 + ], + [ + 178, + 150 + ], + [ + 226, + 148 + ], + [ + 182, + 167 + ] + ], + [ + [ + 179, + 123 + ], + [ + 195, + 107 + ], + [ + 161, + 103 + ], + [ + 195, + 142 + ], + [ + 143, + 121 + ], + [ + 179, + 160 + ] + ] + ] + }, + { + "task_id": "0003314", + "id": "RH20T_cfg4_task_0103_user_0007_scene_0003_cfg_0004_104422070011#120", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003314/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 165 + ], + [ + 249, + 155 + ], + [ + 283, + 171 + ], + [ + 265, + 160 + ], + [ + 308, + 180 + ], + [ + 288, + 169 + ] + ], + [ + [ + 249, + 148 + ], + [ + 256, + 135 + ], + [ + 224, + 157 + ], + [ + 272, + 140 + ], + [ + 219, + 169 + ], + [ + 265, + 152 + ] + ], + [ + [ + 219, + 155 + ], + [ + 229, + 151 + ], + [ + 193, + 148 + ], + [ + 246, + 162 + ], + [ + 180, + 152 + ], + [ + 236, + 166 + ] + ], + [ + [ + 239, + 152 + ], + [ + 237, + 164 + ], + [ + 210, + 157 + ], + [ + 266, + 147 + ], + [ + 210, + 143 + ], + [ + 270, + 133 + ] + ] + ] + }, + { + "task_id": "0003315", + "id": "RH20T_cfg1_task_0110_user_0015_scene_0003_cfg_0001_750612070853#87", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the pencil out from the pencil sharpener'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003315/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 399, + 265 + ], + [ + 402, + 312 + ], + [ + 414, + 285 + ], + [ + 387, + 249 + ], + [ + 410, + 230 + ], + [ + 384, + 205 + ] + ], + [ + [ + 427, + 231 + ], + [ + 427, + 190 + ], + [ + 389, + 208 + ], + [ + 472, + 259 + ], + [ + 392, + 243 + ], + [ + 469, + 296 + ] + ], + [ + [ + 365, + 207 + ], + [ + 403, + 185 + ], + [ + 368, + 247 + ], + [ + 362, + 162 + ], + [ + 330, + 272 + ], + [ + 317, + 186 + ] + ], + [ + [ + 420, + 220 + ], + [ + 435, + 195 + ], + [ + 466, + 209 + ], + [ + 381, + 229 + ], + [ + 451, + 239 + ], + [ + 361, + 256 + ] + ] + ] + }, + { + "task_id": "0003318", + "id": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122063633#115", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003318/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 178, + 167 + ], + [ + 163, + 159 + ], + [ + 185, + 185 + ], + [ + 170, + 149 + ], + [ + 199, + 191 + ], + [ + 184, + 157 + ] + ], + [ + [ + 159, + 184 + ], + [ + 176, + 184 + ], + [ + 143, + 170 + ], + [ + 174, + 197 + ], + [ + 127, + 171 + ], + [ + 158, + 196 + ] + ], + [ + [ + 177, + 169 + ], + [ + 173, + 154 + ], + [ + 180, + 151 + ], + [ + 174, + 187 + ], + [ + 183, + 165 + ], + [ + 178, + 201 + ] + ], + [ + [ + 182, + 171 + ], + [ + 182, + 192 + ], + [ + 161, + 167 + ], + [ + 203, + 174 + ], + [ + 161, + 145 + ], + [ + 203, + 152 + ] + ] + ] + }, + { + "task_id": "0003320", + "id": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122063633#398", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003320/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 199, + 170 + ], + [ + 195, + 154 + ], + [ + 202, + 153 + ], + [ + 196, + 187 + ], + [ + 205, + 169 + ], + [ + 200, + 202 + ] + ], + [ + [ + 214, + 162 + ], + [ + 191, + 162 + ], + [ + 211, + 141 + ], + [ + 217, + 181 + ], + [ + 235, + 141 + ], + [ + 239, + 181 + ] + ], + [ + [ + 190, + 195 + ], + [ + 213, + 195 + ], + [ + 194, + 181 + ], + [ + 185, + 209 + ], + [ + 171, + 181 + ], + [ + 160, + 208 + ] + ], + [ + [ + 199, + 154 + ], + [ + 195, + 173 + ], + [ + 198, + 167 + ], + [ + 201, + 143 + ], + [ + 202, + 147 + ], + [ + 205, + 124 + ] + ] + ] + }, + { + "task_id": "0003321", + "id": "RH20T_cfg7_task_0034_user_0014_scene_0007_cfg_0007_104122063633#543", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003321/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 241, + 145 + ], + [ + 222, + 154 + ], + [ + 248, + 163 + ], + [ + 233, + 126 + ], + [ + 267, + 154 + ], + [ + 251, + 118 + ] + ], + [ + [ + 233, + 149 + ], + [ + 229, + 134 + ], + [ + 233, + 134 + ], + [ + 232, + 165 + ], + [ + 237, + 148 + ], + [ + 236, + 179 + ] + ], + [ + [ + 246, + 162 + ], + [ + 229, + 147 + ], + [ + 260, + 146 + ], + [ + 233, + 177 + ], + [ + 277, + 161 + ], + [ + 250, + 191 + ] + ], + [ + [ + 223, + 177 + ], + [ + 209, + 168 + ], + [ + 216, + 196 + ], + [ + 230, + 157 + ], + [ + 229, + 206 + ], + [ + 244, + 167 + ] + ] + ] + }, + { + "task_id": "0003325", + "id": "RH20T_cfg6_task_0020_user_0014_scene_0001_cfg_0006_104122060811#238", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003325/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 291, + 196 + ], + [ + 288, + 204 + ], + [ + 297, + 210 + ], + [ + 285, + 182 + ], + [ + 300, + 203 + ], + [ + 289, + 175 + ] + ], + [ + [ + 288, + 195 + ], + [ + 294, + 185 + ], + [ + 303, + 200 + ], + [ + 274, + 190 + ], + [ + 297, + 209 + ], + [ + 268, + 199 + ] + ], + [ + [ + 278, + 195 + ], + [ + 287, + 189 + ], + [ + 268, + 201 + ], + [ + 288, + 189 + ], + [ + 258, + 207 + ], + [ + 278, + 196 + ] + ], + [ + [ + 291, + 199 + ], + [ + 297, + 210 + ], + [ + 278, + 208 + ], + [ + 304, + 191 + ], + [ + 272, + 197 + ], + [ + 298, + 181 + ] + ] + ] + }, + { + "task_id": "0003326", + "id": "RH20T_cfg1_task_0043_user_0016_scene_0005_cfg_0001_039422060546#136", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003326/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 392, + 40 + ], + [ + 389, + 27 + ], + [ + 371, + 62 + ], + [ + 416, + 17 + ], + [ + 373, + 77 + ], + [ + 421, + 31 + ] + ], + [ + [ + 369, + 44 + ], + [ + 354, + 52 + ], + [ + 376, + 74 + ], + [ + 363, + 12 + ], + [ + 389, + 65 + ], + [ + 377, + 5 + ] + ], + [ + [ + 385, + 50 + ], + [ + 385, + 24 + ], + [ + 379, + 69 + ], + [ + 390, + 32 + ], + [ + 379, + 92 + ], + [ + 390, + 56 + ] + ], + [ + [ + 378, + 35 + ], + [ + 395, + 11 + ], + [ + 400, + 42 + ], + [ + 358, + 29 + ], + [ + 383, + 68 + ], + [ + 342, + 53 + ] + ] + ] + }, + { + "task_id": "0003327", + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_035622060973#129", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003327/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 170, + 157 + ], + [ + 156, + 171 + ], + [ + 179, + 169 + ], + [ + 159, + 144 + ], + [ + 192, + 156 + ], + [ + 173, + 130 + ] + ], + [ + [ + 184, + 156 + ], + [ + 182, + 145 + ], + [ + 165, + 163 + ], + [ + 204, + 149 + ], + [ + 168, + 173 + ], + [ + 206, + 159 + ] + ], + [ + [ + 177, + 158 + ], + [ + 181, + 177 + ], + [ + 181, + 166 + ], + [ + 173, + 149 + ], + [ + 178, + 148 + ], + [ + 170, + 130 + ] + ], + [ + [ + 190, + 169 + ], + [ + 183, + 181 + ], + [ + 171, + 161 + ], + [ + 208, + 178 + ], + [ + 177, + 148 + ], + [ + 216, + 166 + ] + ] + ] + }, + { + "task_id": "0003328", + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_035622060973#243", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003328/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 198, + 167 + ], + [ + 187, + 168 + ], + [ + 207, + 151 + ], + [ + 189, + 183 + ], + [ + 219, + 149 + ], + [ + 201, + 183 + ] + ], + [ + [ + 190, + 151 + ], + [ + 204, + 152 + ], + [ + 182, + 168 + ], + [ + 198, + 132 + ], + [ + 167, + 168 + ], + [ + 183, + 130 + ] + ], + [ + [ + 187, + 167 + ], + [ + 184, + 157 + ], + [ + 168, + 175 + ], + [ + 206, + 160 + ], + [ + 171, + 184 + ], + [ + 208, + 169 + ] + ], + [ + [ + 195, + 174 + ], + [ + 208, + 160 + ], + [ + 209, + 187 + ], + [ + 181, + 161 + ], + [ + 196, + 202 + ], + [ + 169, + 175 + ] + ] + ] + }, + { + "task_id": "0003331", + "id": "RH20T_cfg1_task_0034_user_0014_scene_0004_cfg_0001_035622060973#592", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003331/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 207, + 181 + ], + [ + 214, + 200 + ], + [ + 220, + 174 + ], + [ + 195, + 187 + ], + [ + 213, + 155 + ], + [ + 188, + 169 + ] + ], + [ + [ + 198, + 177 + ], + [ + 209, + 192 + ], + [ + 182, + 178 + ], + [ + 213, + 175 + ], + [ + 171, + 163 + ], + [ + 202, + 160 + ] + ], + [ + [ + 202, + 180 + ], + [ + 201, + 170 + ], + [ + 182, + 188 + ], + [ + 221, + 172 + ], + [ + 183, + 197 + ], + [ + 221, + 181 + ] + ], + [ + [ + 220, + 188 + ], + [ + 230, + 205 + ], + [ + 238, + 179 + ], + [ + 201, + 198 + ], + [ + 227, + 163 + ], + [ + 191, + 181 + ] + ] + ] + }, + { + "task_id": "0003334", + "id": "RH20T_cfg2_task_0065_user_0007_scene_0003_cfg_0002_104122063678#214", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003334/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 516, + 270 + ], + [ + 508, + 248 + ], + [ + 521, + 258 + ], + [ + 512, + 281 + ], + [ + 530, + 283 + ], + [ + 520, + 304 + ] + ], + [ + [ + 547, + 270 + ], + [ + 574, + 266 + ], + [ + 542, + 249 + ], + [ + 551, + 291 + ], + [ + 516, + 254 + ], + [ + 525, + 295 + ] + ], + [ + [ + 542, + 273 + ], + [ + 552, + 266 + ], + [ + 568, + 274 + ], + [ + 518, + 273 + ], + [ + 557, + 283 + ], + [ + 507, + 281 + ] + ], + [ + [ + 523, + 261 + ], + [ + 530, + 247 + ], + [ + 499, + 266 + ], + [ + 547, + 256 + ], + [ + 494, + 278 + ], + [ + 540, + 269 + ] + ] + ] + }, + { + "task_id": "0003336", + "id": "RH20T_cfg2_task_0065_user_0007_scene_0003_cfg_0002_104122063678#932", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003336/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 504, + 199 + ], + [ + 510, + 211 + ], + [ + 486, + 186 + ], + [ + 522, + 211 + ], + [ + 483, + 176 + ], + [ + 517, + 200 + ] + ], + [ + [ + 488, + 205 + ], + [ + 511, + 206 + ], + [ + 491, + 211 + ], + [ + 485, + 200 + ], + [ + 467, + 210 + ], + [ + 463, + 199 + ] + ], + [ + [ + 513, + 211 + ], + [ + 520, + 197 + ], + [ + 491, + 213 + ], + [ + 535, + 209 + ], + [ + 485, + 226 + ], + [ + 528, + 222 + ] + ], + [ + [ + 500, + 212 + ], + [ + 508, + 197 + ], + [ + 478, + 212 + ], + [ + 522, + 213 + ], + [ + 472, + 226 + ], + [ + 515, + 227 + ] + ] + ] + }, + { + "task_id": "0003337", + "id": "RH20T_cfg2_task_0016_user_0003_scene_0009_cfg_0002_037522062165#170", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003337/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 225, + 227 + ], + [ + 234, + 235 + ], + [ + 224, + 200 + ], + [ + 227, + 255 + ], + [ + 214, + 189 + ], + [ + 216, + 248 + ] + ], + [ + [ + 266, + 219 + ], + [ + 240, + 208 + ], + [ + 256, + 236 + ], + [ + 274, + 205 + ], + [ + 284, + 248 + ], + [ + 299, + 216 + ] + ], + [ + [ + 260, + 241 + ], + [ + 237, + 254 + ], + [ + 251, + 214 + ], + [ + 269, + 268 + ], + [ + 273, + 203 + ], + [ + 289, + 255 + ] + ], + [ + [ + 245, + 245 + ], + [ + 247, + 230 + ], + [ + 258, + 223 + ], + [ + 232, + 268 + ], + [ + 256, + 237 + ], + [ + 232, + 281 + ] + ] + ] + }, + { + "task_id": "0003339", + "id": "RH20T_cfg4_task_0045_user_0014_scene_0007_cfg_0004_038522062288#100", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the brush on the pen rack'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003339/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 268, + 234 + ], + [ + 266, + 205 + ], + [ + 267, + 236 + ], + [ + 270, + 233 + ], + [ + 269, + 268 + ], + [ + 272, + 261 + ] + ], + [ + [ + 212, + 206 + ], + [ + 214, + 230 + ], + [ + 233, + 194 + ], + [ + 189, + 219 + ], + [ + 232, + 168 + ], + [ + 187, + 191 + ] + ], + [ + [ + 227, + 263 + ], + [ + 256, + 251 + ], + [ + 237, + 283 + ], + [ + 218, + 245 + ], + [ + 206, + 296 + ], + [ + 190, + 257 + ] + ], + [ + [ + 233, + 228 + ], + [ + 231, + 210 + ], + [ + 204, + 238 + ], + [ + 262, + 218 + ], + [ + 208, + 254 + ], + [ + 263, + 234 + ] + ] + ] + }, + { + "task_id": "0003346", + "id": "RH20T_cfg4_task_0033_user_0016_scene_0008_cfg_0004_036422060909#66", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003346/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 92 + ], + [ + 397, + 112 + ], + [ + 387, + 104 + ], + [ + 398, + 82 + ], + [ + 382, + 82 + ], + [ + 393, + 61 + ] + ], + [ + [ + 415, + 127 + ], + [ + 413, + 119 + ], + [ + 438, + 117 + ], + [ + 393, + 137 + ], + [ + 441, + 127 + ], + [ + 394, + 147 + ] + ], + [ + [ + 423, + 94 + ], + [ + 445, + 88 + ], + [ + 410, + 88 + ], + [ + 438, + 101 + ], + [ + 389, + 94 + ], + [ + 414, + 108 + ] + ], + [ + [ + 410, + 95 + ], + [ + 417, + 74 + ], + [ + 413, + 84 + ], + [ + 407, + 107 + ], + [ + 407, + 103 + ], + [ + 400, + 127 + ] + ] + ] + }, + { + "task_id": "0003348", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0002_cfg_0001_038522063145#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003348/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 359, + 111 + ], + [ + 355, + 127 + ], + [ + 370, + 96 + ], + [ + 349, + 124 + ], + [ + 373, + 80 + ], + [ + 352, + 108 + ] + ], + [ + [ + 337, + 98 + ], + [ + 338, + 77 + ], + [ + 360, + 102 + ], + [ + 315, + 94 + ], + [ + 359, + 121 + ], + [ + 315, + 113 + ] + ], + [ + [ + 328, + 100 + ], + [ + 325, + 124 + ], + [ + 349, + 104 + ], + [ + 307, + 96 + ], + [ + 352, + 81 + ], + [ + 310, + 72 + ] + ], + [ + [ + 328, + 96 + ], + [ + 308, + 107 + ], + [ + 326, + 98 + ], + [ + 330, + 93 + ], + [ + 346, + 87 + ], + [ + 351, + 81 + ] + ] + ] + }, + { + "task_id": "0003351", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_105422061350#239", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003351/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 292, + 9 + ], + [ + 292, + 0 + ], + [ + 304, + 0 + ], + [ + 281, + 31 + ], + [ + 303, + 10 + ], + [ + 282, + 51 + ] + ], + [ + [ + 318, + 19 + ], + [ + 299, + 11 + ], + [ + 307, + 0 + ], + [ + 328, + 37 + ], + [ + 329, + 7 + ], + [ + 348, + 45 + ] + ], + [ + [ + 311, + 0 + ], + [ + 338, + 1 + ], + [ + 312, + 0 + ], + [ + 310, + 9 + ], + [ + 283, + 0 + ], + [ + 284, + 0 + ] + ], + [ + [ + 294, + 11 + ], + [ + 285, + 28 + ], + [ + 293, + 0 + ], + [ + 294, + 40 + ], + [ + 301, + 0 + ], + [ + 302, + 24 + ] + ] + ] + }, + { + "task_id": "0003352", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_105422061350#441", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003352/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 51 + ], + [ + 287, + 26 + ], + [ + 301, + 33 + ], + [ + 268, + 69 + ], + [ + 298, + 56 + ], + [ + 267, + 91 + ] + ], + [ + [ + 269, + 30 + ], + [ + 270, + 24 + ], + [ + 294, + 9 + ], + [ + 246, + 50 + ], + [ + 294, + 14 + ], + [ + 244, + 58 + ] + ], + [ + [ + 283, + 58 + ], + [ + 287, + 28 + ], + [ + 265, + 64 + ], + [ + 300, + 53 + ], + [ + 262, + 93 + ], + [ + 296, + 80 + ] + ], + [ + [ + 268, + 76 + ], + [ + 294, + 93 + ], + [ + 250, + 88 + ], + [ + 284, + 65 + ], + [ + 224, + 71 + ], + [ + 260, + 50 + ] + ] + ] + }, + { + "task_id": "0003353", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_105422061350#658", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003353/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 204, + 167 + ], + [ + 229, + 168 + ], + [ + 190, + 149 + ], + [ + 217, + 183 + ], + [ + 165, + 148 + ], + [ + 194, + 181 + ] + ], + [ + [ + 172, + 135 + ], + [ + 186, + 118 + ], + [ + 147, + 120 + ], + [ + 196, + 150 + ], + [ + 135, + 136 + ], + [ + 183, + 165 + ] + ], + [ + [ + 177, + 135 + ], + [ + 152, + 148 + ], + [ + 192, + 159 + ], + [ + 162, + 110 + ], + [ + 216, + 146 + ], + [ + 188, + 97 + ] + ], + [ + [ + 185, + 141 + ], + [ + 183, + 123 + ], + [ + 200, + 121 + ], + [ + 169, + 163 + ], + [ + 202, + 138 + ], + [ + 173, + 178 + ] + ] + ] + }, + { + "task_id": "0003354", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0001_cfg_0005_105422061350#931", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003354/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 235, + 116 + ], + [ + 222, + 89 + ], + [ + 228, + 136 + ], + [ + 241, + 98 + ], + [ + 240, + 161 + ], + [ + 252, + 122 + ] + ], + [ + [ + 244, + 89 + ], + [ + 218, + 106 + ], + [ + 264, + 110 + ], + [ + 223, + 66 + ], + [ + 290, + 93 + ], + [ + 251, + 46 + ] + ], + [ + [ + 259, + 92 + ], + [ + 244, + 63 + ], + [ + 284, + 79 + ], + [ + 236, + 104 + ], + [ + 299, + 107 + ], + [ + 251, + 130 + ] + ], + [ + [ + 272, + 92 + ], + [ + 273, + 69 + ], + [ + 287, + 70 + ], + [ + 256, + 114 + ], + [ + 286, + 90 + ], + [ + 256, + 133 + ] + ] + ] + }, + { + "task_id": "0003358", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0003_cfg_0005_105422061350#571", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003358/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 148 + ], + [ + 295, + 172 + ], + [ + 330, + 158 + ], + [ + 283, + 139 + ], + [ + 341, + 134 + ], + [ + 294, + 117 + ] + ], + [ + [ + 274, + 115 + ], + [ + 246, + 108 + ], + [ + 265, + 136 + ], + [ + 284, + 93 + ], + [ + 291, + 143 + ], + [ + 311, + 101 + ] + ], + [ + [ + 276, + 117 + ], + [ + 292, + 103 + ], + [ + 299, + 120 + ], + [ + 256, + 115 + ], + [ + 282, + 136 + ], + [ + 239, + 130 + ] + ], + [ + [ + 284, + 139 + ], + [ + 290, + 120 + ], + [ + 308, + 153 + ], + [ + 261, + 124 + ], + [ + 302, + 169 + ], + [ + 257, + 141 + ] + ] + ] + }, + { + "task_id": "0003359", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0003_cfg_0005_105422061350#703", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003359/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 312, + 126 + ], + [ + 299, + 131 + ], + [ + 304, + 151 + ], + [ + 320, + 100 + ], + [ + 316, + 146 + ], + [ + 332, + 98 + ] + ], + [ + [ + 296, + 126 + ], + [ + 301, + 108 + ], + [ + 318, + 142 + ], + [ + 274, + 111 + ], + [ + 312, + 158 + ], + [ + 270, + 128 + ] + ], + [ + [ + 321, + 113 + ], + [ + 302, + 96 + ], + [ + 300, + 127 + ], + [ + 343, + 98 + ], + [ + 318, + 142 + ], + [ + 361, + 115 + ] + ], + [ + [ + 310, + 112 + ], + [ + 295, + 109 + ], + [ + 295, + 93 + ], + [ + 326, + 131 + ], + [ + 310, + 96 + ], + [ + 339, + 133 + ] + ] + ] + }, + { + "task_id": "0003361", + "id": "RH20T_cfg1_task_0105_user_0010_scene_0005_cfg_0001_038522063145#86", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003361/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 483, + 107 + ], + [ + 489, + 84 + ], + [ + 468, + 100 + ], + [ + 500, + 114 + ], + [ + 463, + 121 + ], + [ + 493, + 136 + ] + ], + [ + [ + 486, + 139 + ], + [ + 492, + 157 + ], + [ + 483, + 127 + ], + [ + 489, + 151 + ], + [ + 477, + 110 + ], + [ + 484, + 135 + ] + ], + [ + [ + 496, + 152 + ], + [ + 500, + 139 + ], + [ + 474, + 147 + ], + [ + 519, + 157 + ], + [ + 469, + 160 + ], + [ + 515, + 171 + ] + ], + [ + [ + 480, + 138 + ], + [ + 487, + 120 + ], + [ + 503, + 146 + ], + [ + 458, + 130 + ], + [ + 495, + 163 + ], + [ + 452, + 148 + ] + ] + ] + }, + { + "task_id": "0003362", + "id": "RH20T_cfg7_task_0020_user_0014_scene_0010_cfg_0007_037522061512#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003362/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 230, + 168 + ], + [ + 243, + 150 + ], + [ + 238, + 179 + ], + [ + 222, + 156 + ], + [ + 226, + 198 + ], + [ + 208, + 176 + ] + ], + [ + [ + 223, + 162 + ], + [ + 216, + 149 + ], + [ + 218, + 180 + ], + [ + 227, + 142 + ], + [ + 225, + 194 + ], + [ + 235, + 155 + ] + ], + [ + [ + 230, + 167 + ], + [ + 224, + 160 + ], + [ + 254, + 168 + ], + [ + 206, + 166 + ], + [ + 258, + 174 + ], + [ + 213, + 173 + ] + ], + [ + [ + 235, + 148 + ], + [ + 212, + 139 + ], + [ + 243, + 126 + ], + [ + 227, + 169 + ], + [ + 264, + 134 + ], + [ + 248, + 177 + ] + ] + ] + }, + { + "task_id": "0003363", + "id": "RH20T_cfg6_task_0042_user_0014_scene_0004_cfg_0006_104122063633#117", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003363/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 201, + 94 + ], + [ + 175, + 91 + ], + [ + 199, + 85 + ], + [ + 202, + 103 + ], + [ + 226, + 89 + ], + [ + 226, + 106 + ] + ], + [ + [ + 230, + 125 + ], + [ + 212, + 135 + ], + [ + 222, + 103 + ], + [ + 239, + 146 + ], + [ + 241, + 92 + ], + [ + 258, + 136 + ] + ], + [ + [ + 227, + 89 + ], + [ + 205, + 99 + ], + [ + 221, + 74 + ], + [ + 233, + 105 + ], + [ + 242, + 64 + ], + [ + 256, + 94 + ] + ], + [ + [ + 225, + 117 + ], + [ + 223, + 98 + ], + [ + 227, + 101 + ], + [ + 222, + 134 + ], + [ + 229, + 118 + ], + [ + 224, + 151 + ] + ] + ] + }, + { + "task_id": "0003364", + "id": "RH20T_cfg6_task_0042_user_0014_scene_0004_cfg_0006_104122063633#176", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003364/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 218, + 112 + ], + [ + 216, + 116 + ], + [ + 238, + 129 + ], + [ + 198, + 94 + ], + [ + 239, + 125 + ], + [ + 201, + 92 + ] + ], + [ + [ + 208, + 117 + ], + [ + 197, + 103 + ], + [ + 228, + 105 + ], + [ + 188, + 129 + ], + [ + 240, + 120 + ], + [ + 200, + 144 + ] + ], + [ + [ + 221, + 112 + ], + [ + 196, + 115 + ], + [ + 216, + 112 + ], + [ + 227, + 112 + ], + [ + 242, + 108 + ], + [ + 251, + 109 + ] + ], + [ + [ + 221, + 117 + ], + [ + 219, + 102 + ], + [ + 224, + 98 + ], + [ + 218, + 137 + ], + [ + 226, + 112 + ], + [ + 220, + 150 + ] + ] + ] + }, + { + "task_id": "0003369", + "id": "RH20T_cfg4_task_0103_user_0007_scene_0003_cfg_0004_036422060909#96", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003369/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 320, + 147 + ], + [ + 301, + 133 + ], + [ + 339, + 135 + ], + [ + 299, + 159 + ], + [ + 358, + 148 + ], + [ + 319, + 174 + ] + ], + [ + [ + 331, + 148 + ], + [ + 342, + 171 + ], + [ + 350, + 143 + ], + [ + 314, + 153 + ], + [ + 338, + 119 + ], + [ + 302, + 131 + ] + ], + [ + [ + 347, + 154 + ], + [ + 351, + 137 + ], + [ + 365, + 141 + ], + [ + 329, + 168 + ], + [ + 361, + 157 + ], + [ + 327, + 183 + ] + ], + [ + [ + 339, + 149 + ], + [ + 349, + 173 + ], + [ + 318, + 157 + ], + [ + 359, + 141 + ], + [ + 308, + 131 + ], + [ + 350, + 117 + ] + ] + ] + }, + { + "task_id": "0003370", + "id": "RH20T_cfg2_task_0071_user_0007_scene_0006_cfg_0002_105422061350#223", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003370/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 195 + ], + [ + 293, + 211 + ], + [ + 300, + 176 + ], + [ + 330, + 216 + ], + [ + 321, + 160 + ], + [ + 353, + 199 + ] + ], + [ + [ + 291, + 213 + ], + [ + 299, + 226 + ], + [ + 293, + 236 + ], + [ + 289, + 191 + ], + [ + 284, + 222 + ], + [ + 281, + 176 + ] + ], + [ + [ + 307, + 223 + ], + [ + 289, + 238 + ], + [ + 314, + 245 + ], + [ + 300, + 200 + ], + [ + 331, + 231 + ], + [ + 318, + 186 + ] + ], + [ + [ + 296, + 206 + ], + [ + 298, + 191 + ], + [ + 318, + 220 + ], + [ + 274, + 192 + ], + [ + 314, + 233 + ], + [ + 273, + 206 + ] + ] + ] + }, + { + "task_id": "0003371", + "id": "RH20T_cfg2_task_0071_user_0007_scene_0006_cfg_0002_105422061350#448", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003371/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 208 + ], + [ + 323, + 234 + ], + [ + 347, + 206 + ], + [ + 292, + 211 + ], + [ + 344, + 179 + ], + [ + 289, + 185 + ] + ], + [ + [ + 280, + 213 + ], + [ + 287, + 194 + ], + [ + 283, + 232 + ], + [ + 278, + 195 + ], + [ + 276, + 250 + ], + [ + 271, + 212 + ] + ], + [ + [ + 294, + 212 + ], + [ + 295, + 198 + ], + [ + 313, + 229 + ], + [ + 276, + 194 + ], + [ + 310, + 241 + ], + [ + 275, + 207 + ] + ], + [ + [ + 294, + 223 + ], + [ + 314, + 208 + ], + [ + 309, + 243 + ], + [ + 279, + 203 + ], + [ + 289, + 259 + ], + [ + 258, + 218 + ] + ] + ] + }, + { + "task_id": "0003372", + "id": "RH20T_cfg3_task_0005_user_0016_scene_0003_cfg_0003_104122063550#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003372/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 344, + 237 + ], + [ + 332, + 254 + ], + [ + 368, + 246 + ], + [ + 321, + 227 + ], + [ + 379, + 230 + ], + [ + 333, + 212 + ] + ], + [ + [ + 331, + 217 + ], + [ + 331, + 205 + ], + [ + 340, + 241 + ], + [ + 323, + 194 + ], + [ + 339, + 251 + ], + [ + 323, + 205 + ] + ], + [ + [ + 314, + 236 + ], + [ + 304, + 250 + ], + [ + 319, + 216 + ], + [ + 309, + 257 + ], + [ + 330, + 200 + ], + [ + 321, + 242 + ] + ], + [ + [ + 359, + 231 + ], + [ + 364, + 247 + ], + [ + 383, + 219 + ], + [ + 335, + 243 + ], + [ + 380, + 202 + ], + [ + 330, + 226 + ] + ] + ] + }, + { + "task_id": "0003373", + "id": "RH20T_cfg5_task_0118_user_0010_scene_0004_cfg_0005_037522062165#168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the water tap'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003373/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 188, + 165 + ], + [ + 161, + 178 + ], + [ + 171, + 144 + ], + [ + 206, + 186 + ], + [ + 197, + 132 + ], + [ + 232, + 172 + ] + ], + [ + [ + 164, + 162 + ], + [ + 159, + 143 + ], + [ + 137, + 147 + ], + [ + 192, + 178 + ], + [ + 143, + 165 + ], + [ + 194, + 195 + ] + ], + [ + [ + 155, + 135 + ], + [ + 136, + 148 + ], + [ + 179, + 134 + ], + [ + 127, + 135 + ], + [ + 199, + 121 + ], + [ + 147, + 120 + ] + ], + [ + [ + 171, + 152 + ], + [ + 186, + 145 + ], + [ + 146, + 169 + ], + [ + 194, + 136 + ], + [ + 132, + 174 + ], + [ + 179, + 142 + ] + ] + ] + }, + { + "task_id": "0003374", + "id": "RH20T_cfg5_task_0004_user_0014_scene_0007_cfg_0005_104122063678#255", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003374/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 401, + 252 + ], + [ + 406, + 236 + ], + [ + 413, + 273 + ], + [ + 389, + 232 + ], + [ + 407, + 287 + ], + [ + 384, + 246 + ] + ], + [ + [ + 432, + 244 + ], + [ + 410, + 227 + ], + [ + 440, + 233 + ], + [ + 424, + 255 + ], + [ + 464, + 251 + ], + [ + 445, + 272 + ] + ], + [ + [ + 416, + 257 + ], + [ + 439, + 245 + ], + [ + 411, + 261 + ], + [ + 422, + 253 + ], + [ + 388, + 272 + ], + [ + 397, + 266 + ] + ], + [ + [ + 399, + 273 + ], + [ + 377, + 281 + ], + [ + 412, + 295 + ], + [ + 386, + 252 + ], + [ + 433, + 287 + ], + [ + 407, + 245 + ] + ] + ] + }, + { + "task_id": "0003377", + "id": "RH20T_cfg4_task_0005_user_0010_scene_0004_cfg_0004_036422060909#103", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003377/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 374, + 94 + ], + [ + 371, + 105 + ], + [ + 399, + 93 + ], + [ + 349, + 95 + ], + [ + 401, + 83 + ], + [ + 353, + 85 + ] + ], + [ + [ + 353, + 91 + ], + [ + 357, + 73 + ], + [ + 371, + 81 + ], + [ + 334, + 101 + ], + [ + 366, + 98 + ], + [ + 330, + 118 + ] + ], + [ + [ + 332, + 83 + ], + [ + 339, + 75 + ], + [ + 310, + 94 + ], + [ + 354, + 71 + ], + [ + 305, + 100 + ], + [ + 347, + 79 + ] + ], + [ + [ + 371, + 90 + ], + [ + 366, + 115 + ], + [ + 346, + 85 + ], + [ + 397, + 96 + ], + [ + 351, + 59 + ], + [ + 403, + 69 + ] + ] + ] + }, + { + "task_id": "0003378", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_038522062547#156", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003378/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 437, + 21 + ], + [ + 459, + 19 + ], + [ + 431, + 0 + ], + [ + 444, + 44 + ], + [ + 408, + 2 + ], + [ + 422, + 47 + ] + ], + [ + [ + 453, + 70 + ], + [ + 468, + 63 + ], + [ + 443, + 90 + ], + [ + 463, + 50 + ], + [ + 429, + 95 + ], + [ + 448, + 57 + ] + ], + [ + [ + 419, + 63 + ], + [ + 443, + 68 + ], + [ + 425, + 38 + ], + [ + 413, + 87 + ], + [ + 401, + 34 + ], + [ + 390, + 82 + ] + ], + [ + [ + 431, + 48 + ], + [ + 432, + 26 + ], + [ + 439, + 62 + ], + [ + 423, + 35 + ], + [ + 438, + 83 + ], + [ + 422, + 56 + ] + ] + ] + }, + { + "task_id": "0003382", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0004_cfg_0006_104122063633#249", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003382/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 147, + 166 + ], + [ + 128, + 168 + ], + [ + 149, + 188 + ], + [ + 145, + 144 + ], + [ + 169, + 187 + ], + [ + 164, + 143 + ] + ], + [ + [ + 165, + 131 + ], + [ + 181, + 131 + ], + [ + 162, + 107 + ], + [ + 168, + 154 + ], + [ + 147, + 108 + ], + [ + 153, + 153 + ] + ], + [ + [ + 151, + 146 + ], + [ + 158, + 143 + ], + [ + 152, + 168 + ], + [ + 150, + 124 + ], + [ + 145, + 173 + ], + [ + 142, + 127 + ] + ], + [ + [ + 163, + 133 + ], + [ + 147, + 117 + ], + [ + 154, + 150 + ], + [ + 172, + 118 + ], + [ + 170, + 166 + ], + [ + 186, + 134 + ] + ] + ] + }, + { + "task_id": "0003383", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0004_cfg_0006_104122063633#373", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003383/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 171, + 119 + ], + [ + 184, + 136 + ], + [ + 188, + 114 + ], + [ + 152, + 125 + ], + [ + 175, + 98 + ], + [ + 138, + 107 + ] + ], + [ + [ + 178, + 142 + ], + [ + 200, + 143 + ], + [ + 173, + 131 + ], + [ + 182, + 151 + ], + [ + 149, + 130 + ], + [ + 160, + 150 + ] + ], + [ + [ + 161, + 131 + ], + [ + 158, + 114 + ], + [ + 164, + 118 + ], + [ + 157, + 146 + ], + [ + 166, + 134 + ], + [ + 160, + 162 + ] + ], + [ + [ + 170, + 115 + ], + [ + 186, + 119 + ], + [ + 181, + 98 + ], + [ + 160, + 132 + ], + [ + 164, + 92 + ], + [ + 142, + 128 + ] + ] + ] + }, + { + "task_id": "0003391", + "id": "RH20T_cfg1_task_0065_user_0014_scene_0004_cfg_0001_750612070853#81", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003391/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 251, + 178 + ], + [ + 243, + 147 + ], + [ + 228, + 164 + ], + [ + 276, + 195 + ], + [ + 236, + 191 + ], + [ + 282, + 224 + ] + ], + [ + [ + 266, + 162 + ], + [ + 278, + 140 + ], + [ + 291, + 138 + ], + [ + 243, + 184 + ], + [ + 277, + 163 + ], + [ + 228, + 209 + ] + ], + [ + [ + 280, + 167 + ], + [ + 299, + 149 + ], + [ + 316, + 172 + ], + [ + 246, + 163 + ], + [ + 295, + 194 + ], + [ + 222, + 182 + ] + ], + [ + [ + 284, + 200 + ], + [ + 274, + 162 + ], + [ + 304, + 198 + ], + [ + 267, + 201 + ], + [ + 315, + 238 + ], + [ + 277, + 237 + ] + ] + ] + }, + { + "task_id": "0003392", + "id": "RH20T_cfg1_task_0051_user_0015_scene_0002_cfg_0001_038522063145#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the dish off the dish rack'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003392/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 473, + 116 + ], + [ + 475, + 104 + ], + [ + 466, + 139 + ], + [ + 480, + 93 + ], + [ + 465, + 149 + ], + [ + 478, + 105 + ] + ], + [ + [ + 496, + 141 + ], + [ + 489, + 159 + ], + [ + 515, + 130 + ], + [ + 479, + 151 + ], + [ + 522, + 112 + ], + [ + 486, + 134 + ] + ], + [ + [ + 475, + 134 + ], + [ + 471, + 144 + ], + [ + 479, + 111 + ], + [ + 470, + 157 + ], + [ + 483, + 99 + ], + [ + 474, + 147 + ] + ], + [ + [ + 480, + 122 + ], + [ + 499, + 103 + ], + [ + 499, + 129 + ], + [ + 461, + 114 + ], + [ + 481, + 145 + ], + [ + 444, + 132 + ] + ] + ] + }, + { + "task_id": "0003393", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0006_cfg_0004_104122062295#94", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003393/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 335, + 132 + ], + [ + 368, + 101 + ], + [ + 362, + 152 + ], + [ + 311, + 114 + ], + [ + 326, + 186 + ], + [ + 280, + 144 + ] + ], + [ + [ + 307, + 106 + ], + [ + 315, + 59 + ], + [ + 274, + 89 + ], + [ + 345, + 125 + ], + [ + 269, + 130 + ], + [ + 335, + 169 + ] + ], + [ + [ + 341, + 159 + ], + [ + 378, + 189 + ], + [ + 317, + 193 + ], + [ + 367, + 121 + ], + [ + 285, + 164 + ], + [ + 331, + 92 + ] + ], + [ + [ + 278, + 147 + ], + [ + 245, + 171 + ], + [ + 250, + 107 + ], + [ + 306, + 186 + ], + [ + 280, + 89 + ], + [ + 332, + 163 + ] + ] + ] + }, + { + "task_id": "0003394", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0006_cfg_0004_104122062295#196", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003394/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 233, + 82 + ], + [ + 259, + 124 + ], + [ + 231, + 56 + ], + [ + 235, + 103 + ], + [ + 203, + 12 + ], + [ + 213, + 66 + ] + ], + [ + [ + 293, + 140 + ], + [ + 275, + 172 + ], + [ + 297, + 101 + ], + [ + 290, + 174 + ], + [ + 314, + 74 + ], + [ + 305, + 146 + ] + ], + [ + [ + 290, + 32 + ], + [ + 303, + 78 + ], + [ + 324, + 18 + ], + [ + 261, + 43 + ], + [ + 309, + 0 + ], + [ + 248, + 0 + ] + ], + [ + [ + 276, + 86 + ], + [ + 289, + 40 + ], + [ + 241, + 65 + ], + [ + 313, + 110 + ], + [ + 232, + 106 + ], + [ + 299, + 152 + ] + ] + ] + }, + { + "task_id": "0003395", + "id": "RH20T_cfg2_task_0041_user_0007_scene_0009_cfg_0002_104122063678#485", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the tabletop with a sponge'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003395/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 431, + 360 + ], + [ + 418, + 360 + ], + [ + 411, + 360 + ], + [ + 451, + 357 + ], + [ + 425, + 360 + ], + [ + 467, + 360 + ] + ], + [ + [ + 388, + 360 + ], + [ + 396, + 360 + ], + [ + 364, + 360 + ], + [ + 413, + 360 + ], + [ + 359, + 360 + ], + [ + 404, + 360 + ] + ], + [ + [ + 437, + 360 + ], + [ + 447, + 342 + ], + [ + 467, + 360 + ], + [ + 408, + 348 + ], + [ + 456, + 360 + ], + [ + 395, + 360 + ] + ], + [ + [ + 416, + 360 + ], + [ + 422, + 360 + ], + [ + 433, + 360 + ], + [ + 400, + 351 + ], + [ + 426, + 360 + ], + [ + 396, + 360 + ] + ] + ] + }, + { + "task_id": "0003398", + "id": "RH20T_cfg3_task_0058_user_0010_scene_0001_cfg_0003_104122062823#309", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003398/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 425, + 214 + ], + [ + 430, + 231 + ], + [ + 403, + 248 + ], + [ + 446, + 182 + ], + [ + 395, + 231 + ], + [ + 442, + 160 + ] + ], + [ + [ + 432, + 169 + ], + [ + 419, + 178 + ], + [ + 462, + 201 + ], + [ + 402, + 138 + ], + [ + 471, + 191 + ], + [ + 415, + 133 + ] + ], + [ + [ + 405, + 181 + ], + [ + 418, + 149 + ], + [ + 371, + 164 + ], + [ + 439, + 198 + ], + [ + 356, + 197 + ], + [ + 427, + 233 + ] + ], + [ + [ + 409, + 202 + ], + [ + 425, + 180 + ], + [ + 448, + 213 + ], + [ + 370, + 191 + ], + [ + 432, + 232 + ], + [ + 359, + 212 + ] + ] + ] + }, + { + "task_id": "0003400", + "id": "RH20T_cfg4_task_0011_user_0016_scene_0001_cfg_0004_104122062823#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003400/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 174, + 7 + ], + [ + 181, + 46 + ], + [ + 212, + 0 + ], + [ + 139, + 25 + ], + [ + 204, + 0 + ], + [ + 134, + 0 + ] + ], + [ + [ + 184, + 18 + ], + [ + 176, + 0 + ], + [ + 210, + 1 + ], + [ + 155, + 38 + ], + [ + 216, + 37 + ], + [ + 165, + 75 + ] + ], + [ + [ + 181, + 0 + ], + [ + 193, + 37 + ], + [ + 215, + 0 + ], + [ + 151, + 19 + ], + [ + 203, + 0 + ], + [ + 141, + 0 + ] + ], + [ + [ + 153, + 9 + ], + [ + 112, + 4 + ], + [ + 138, + 0 + ], + [ + 166, + 40 + ], + [ + 184, + 0 + ], + [ + 205, + 45 + ] + ] + ] + }, + { + "task_id": "0003401", + "id": "RH20T_cfg7_task_0038_user_0020_scene_0006_cfg_0007_104122061330#186", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003401/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 350, + 160 + ], + [ + 346, + 146 + ], + [ + 337, + 183 + ], + [ + 361, + 138 + ], + [ + 342, + 195 + ], + [ + 364, + 152 + ] + ], + [ + [ + 367, + 175 + ], + [ + 350, + 153 + ], + [ + 351, + 186 + ], + [ + 385, + 163 + ], + [ + 367, + 207 + ], + [ + 403, + 186 + ] + ], + [ + [ + 328, + 163 + ], + [ + 315, + 167 + ], + [ + 313, + 181 + ], + [ + 345, + 143 + ], + [ + 326, + 177 + ], + [ + 357, + 140 + ] + ], + [ + [ + 341, + 145 + ], + [ + 322, + 127 + ], + [ + 325, + 163 + ], + [ + 358, + 127 + ], + [ + 344, + 180 + ], + [ + 378, + 145 + ] + ] + ] + }, + { + "task_id": "0003403", + "id": "RH20T_cfg3_task_0026_user_0010_scene_0001_cfg_0003_036422060909#50", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push and close the drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003403/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 329, + 128 + ], + [ + 330, + 106 + ], + [ + 352, + 114 + ], + [ + 306, + 141 + ], + [ + 350, + 134 + ], + [ + 307, + 161 + ] + ], + [ + [ + 340, + 138 + ], + [ + 361, + 134 + ], + [ + 316, + 141 + ], + [ + 362, + 135 + ], + [ + 297, + 143 + ], + [ + 342, + 138 + ] + ], + [ + [ + 306, + 145 + ], + [ + 305, + 148 + ], + [ + 322, + 169 + ], + [ + 289, + 120 + ], + [ + 322, + 165 + ], + [ + 291, + 119 + ] + ], + [ + [ + 322, + 130 + ], + [ + 332, + 99 + ], + [ + 305, + 118 + ], + [ + 341, + 142 + ], + [ + 297, + 146 + ], + [ + 331, + 172 + ] + ] + ] + }, + { + "task_id": "0003404", + "id": "RH20T_cfg3_task_0008_user_0016_scene_0009_cfg_0003_104122062295#190", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003404/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 215, + 360 + ], + [ + 226, + 345 + ], + [ + 186, + 342 + ], + [ + 249, + 360 + ], + [ + 179, + 360 + ], + [ + 236, + 360 + ] + ], + [ + [ + 159, + 348 + ], + [ + 143, + 360 + ], + [ + 99, + 331 + ], + [ + 213, + 360 + ], + [ + 118, + 281 + ], + [ + 221, + 314 + ] + ], + [ + [ + 266, + 319 + ], + [ + 230, + 317 + ], + [ + 271, + 360 + ], + [ + 260, + 263 + ], + [ + 301, + 360 + ], + [ + 288, + 269 + ] + ], + [ + [ + 254, + 360 + ], + [ + 210, + 360 + ], + [ + 207, + 360 + ], + [ + 294, + 360 + ], + [ + 254, + 315 + ], + [ + 334, + 360 + ] + ] + ] + }, + { + "task_id": "0003407", + "id": "RH20T_cfg4_task_0071_user_0015_scene_0010_cfg_0004_104122062295#110", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003407/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 245, + 130 + ], + [ + 249, + 98 + ], + [ + 213, + 121 + ], + [ + 278, + 138 + ], + [ + 211, + 150 + ], + [ + 273, + 167 + ] + ], + [ + [ + 253, + 119 + ], + [ + 284, + 126 + ], + [ + 259, + 85 + ], + [ + 246, + 152 + ], + [ + 228, + 79 + ], + [ + 216, + 145 + ] + ], + [ + [ + 243, + 172 + ], + [ + 270, + 165 + ], + [ + 259, + 156 + ], + [ + 229, + 186 + ], + [ + 228, + 163 + ], + [ + 200, + 193 + ] + ], + [ + [ + 270, + 116 + ], + [ + 263, + 84 + ], + [ + 238, + 120 + ], + [ + 299, + 113 + ], + [ + 245, + 154 + ], + [ + 305, + 144 + ] + ] + ] + }, + { + "task_id": "0003408", + "id": "RH20T_cfg5_task_0010_user_0010_scene_0009_cfg_0005_104122063678#358", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003408/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 351, + 173 + ], + [ + 358, + 170 + ], + [ + 338, + 154 + ], + [ + 364, + 193 + ], + [ + 330, + 157 + ], + [ + 357, + 198 + ] + ], + [ + [ + 324, + 174 + ], + [ + 335, + 158 + ], + [ + 306, + 159 + ], + [ + 342, + 188 + ], + [ + 294, + 175 + ], + [ + 332, + 206 + ] + ], + [ + [ + 348, + 183 + ], + [ + 332, + 168 + ], + [ + 351, + 178 + ], + [ + 344, + 190 + ], + [ + 366, + 193 + ], + [ + 360, + 206 + ] + ], + [ + [ + 341, + 179 + ], + [ + 345, + 163 + ], + [ + 343, + 196 + ], + [ + 338, + 162 + ], + [ + 340, + 210 + ], + [ + 335, + 176 + ] + ] + ] + }, + { + "task_id": "0003409", + "id": "RH20T_cfg3_task_0007_user_0016_scene_0002_cfg_0003_104122063550#193", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003409/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 378, + 198 + ], + [ + 358, + 207 + ], + [ + 378, + 173 + ], + [ + 379, + 222 + ], + [ + 397, + 166 + ], + [ + 397, + 213 + ] + ], + [ + [ + 379, + 220 + ], + [ + 393, + 243 + ], + [ + 363, + 233 + ], + [ + 397, + 207 + ], + [ + 350, + 211 + ], + [ + 383, + 184 + ] + ], + [ + [ + 370, + 227 + ], + [ + 373, + 219 + ], + [ + 375, + 254 + ], + [ + 365, + 202 + ], + [ + 372, + 259 + ], + [ + 363, + 210 + ] + ], + [ + [ + 361, + 201 + ], + [ + 337, + 199 + ], + [ + 346, + 208 + ], + [ + 374, + 195 + ], + [ + 373, + 212 + ], + [ + 399, + 197 + ] + ] + ] + }, + { + "task_id": "0003414", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0004_cfg_0003_104122062295#288", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003414/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 235, + 269 + ], + [ + 276, + 261 + ], + [ + 260, + 250 + ], + [ + 214, + 284 + ], + [ + 212, + 257 + ], + [ + 169, + 293 + ] + ], + [ + [ + 260, + 287 + ], + [ + 245, + 260 + ], + [ + 233, + 258 + ], + [ + 292, + 321 + ], + [ + 247, + 282 + ], + [ + 304, + 343 + ] + ], + [ + [ + 296, + 261 + ], + [ + 259, + 233 + ], + [ + 263, + 251 + ], + [ + 324, + 268 + ], + [ + 309, + 286 + ], + [ + 365, + 298 + ] + ], + [ + [ + 298, + 267 + ], + [ + 310, + 231 + ], + [ + 265, + 237 + ], + [ + 334, + 302 + ], + [ + 256, + 269 + ], + [ + 320, + 334 + ] + ] + ] + }, + { + "task_id": "0003416", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0002_cfg_0006_037522061512#231", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003416/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 264, + 198 + ], + [ + 265, + 175 + ], + [ + 278, + 188 + ], + [ + 249, + 209 + ], + [ + 277, + 209 + ], + [ + 249, + 232 + ] + ], + [ + [ + 256, + 206 + ], + [ + 238, + 204 + ], + [ + 240, + 201 + ], + [ + 271, + 211 + ], + [ + 259, + 203 + ], + [ + 289, + 213 + ] + ], + [ + [ + 244, + 209 + ], + [ + 223, + 204 + ], + [ + 257, + 210 + ], + [ + 229, + 209 + ], + [ + 278, + 215 + ], + [ + 252, + 215 + ] + ], + [ + [ + 284, + 213 + ], + [ + 261, + 200 + ], + [ + 282, + 216 + ], + [ + 285, + 210 + ], + [ + 303, + 227 + ], + [ + 309, + 223 + ] + ] + ] + }, + { + "task_id": "0003419", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0002_cfg_0006_037522061512#710", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003419/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 240, + 196 + ], + [ + 220, + 186 + ], + [ + 226, + 214 + ], + [ + 253, + 178 + ], + [ + 247, + 224 + ], + [ + 274, + 188 + ] + ], + [ + [ + 225, + 184 + ], + [ + 242, + 179 + ], + [ + 224, + 162 + ], + [ + 226, + 206 + ], + [ + 205, + 166 + ], + [ + 208, + 213 + ] + ], + [ + [ + 194, + 193 + ], + [ + 211, + 210 + ], + [ + 185, + 187 + ], + [ + 203, + 198 + ], + [ + 168, + 170 + ], + [ + 187, + 182 + ] + ], + [ + [ + 215, + 184 + ], + [ + 215, + 164 + ], + [ + 230, + 171 + ], + [ + 199, + 198 + ], + [ + 229, + 188 + ], + [ + 200, + 215 + ] + ] + ] + }, + { + "task_id": "0003421", + "id": "RH20T_cfg4_task_0034_user_0016_scene_0001_cfg_0004_104122062295#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003421/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 238, + 256 + ], + [ + 237, + 211 + ], + [ + 177, + 249 + ], + [ + 293, + 263 + ], + [ + 171, + 304 + ], + [ + 298, + 314 + ] + ], + [ + [ + 218, + 231 + ], + [ + 251, + 215 + ], + [ + 188, + 197 + ], + [ + 250, + 268 + ], + [ + 148, + 213 + ], + [ + 213, + 291 + ] + ], + [ + [ + 206, + 247 + ], + [ + 214, + 200 + ], + [ + 158, + 231 + ], + [ + 256, + 263 + ], + [ + 153, + 275 + ], + [ + 246, + 307 + ] + ], + [ + [ + 178, + 300 + ], + [ + 149, + 351 + ], + [ + 128, + 275 + ], + [ + 227, + 325 + ], + [ + 157, + 234 + ], + [ + 247, + 281 + ] + ] + ] + }, + { + "task_id": "0003422", + "id": "RH20T_cfg4_task_0034_user_0016_scene_0001_cfg_0004_104122062295#223", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003422/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 329, + 186 + ], + [ + 309, + 208 + ], + [ + 352, + 217 + ], + [ + 307, + 157 + ], + [ + 377, + 194 + ], + [ + 328, + 133 + ] + ], + [ + [ + 346, + 206 + ], + [ + 332, + 220 + ], + [ + 329, + 169 + ], + [ + 362, + 243 + ], + [ + 342, + 160 + ], + [ + 373, + 229 + ] + ], + [ + [ + 287, + 160 + ], + [ + 288, + 131 + ], + [ + 243, + 159 + ], + [ + 329, + 162 + ], + [ + 238, + 191 + ], + [ + 330, + 195 + ] + ], + [ + [ + 313, + 193 + ], + [ + 321, + 157 + ], + [ + 272, + 191 + ], + [ + 351, + 195 + ], + [ + 266, + 225 + ], + [ + 341, + 226 + ] + ] + ] + }, + { + "task_id": "0003424", + "id": "RH20T_cfg4_task_0034_user_0016_scene_0001_cfg_0004_104122062295#430", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003424/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 240, + 307 + ], + [ + 243, + 260 + ], + [ + 203, + 281 + ], + [ + 282, + 335 + ], + [ + 201, + 322 + ], + [ + 277, + 360 + ] + ], + [ + [ + 214, + 300 + ], + [ + 187, + 332 + ], + [ + 261, + 317 + ], + [ + 169, + 283 + ], + [ + 279, + 288 + ], + [ + 195, + 258 + ] + ], + [ + [ + 283, + 338 + ], + [ + 311, + 297 + ], + [ + 241, + 310 + ], + [ + 328, + 360 + ], + [ + 207, + 353 + ], + [ + 299, + 360 + ] + ], + [ + [ + 237, + 298 + ], + [ + 261, + 269 + ], + [ + 265, + 345 + ], + [ + 208, + 249 + ], + [ + 239, + 360 + ], + [ + 174, + 280 + ] + ] + ] + }, + { + "task_id": "0003430", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0005_cfg_0001_039422060546#142", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003430/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 298, + 233 + ], + [ + 295, + 230 + ], + [ + 322, + 258 + ], + [ + 275, + 208 + ], + [ + 323, + 259 + ], + [ + 279, + 212 + ] + ], + [ + [ + 269, + 237 + ], + [ + 261, + 217 + ], + [ + 262, + 270 + ], + [ + 276, + 207 + ], + [ + 270, + 286 + ], + [ + 282, + 225 + ] + ], + [ + [ + 284, + 246 + ], + [ + 282, + 221 + ], + [ + 264, + 263 + ], + [ + 306, + 227 + ], + [ + 264, + 291 + ], + [ + 309, + 254 + ] + ], + [ + [ + 237, + 226 + ], + [ + 273, + 225 + ], + [ + 240, + 193 + ], + [ + 234, + 261 + ], + [ + 206, + 194 + ], + [ + 197, + 261 + ] + ] + ] + }, + { + "task_id": "0003432", + "id": "RH20T_cfg5_task_0081_user_0007_scene_0010_cfg_0005_105422061350#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003432/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 184, + 89 + ], + [ + 211, + 81 + ], + [ + 172, + 59 + ], + [ + 195, + 118 + ], + [ + 142, + 67 + ], + [ + 167, + 127 + ] + ], + [ + [ + 183, + 143 + ], + [ + 175, + 116 + ], + [ + 168, + 137 + ], + [ + 197, + 149 + ], + [ + 176, + 167 + ], + [ + 205, + 175 + ] + ], + [ + [ + 182, + 126 + ], + [ + 171, + 151 + ], + [ + 153, + 119 + ], + [ + 211, + 133 + ], + [ + 164, + 95 + ], + [ + 221, + 109 + ] + ], + [ + [ + 183, + 121 + ], + [ + 185, + 101 + ], + [ + 213, + 126 + ], + [ + 152, + 115 + ], + [ + 210, + 143 + ], + [ + 152, + 133 + ] + ] + ] + }, + { + "task_id": "0003433", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0008_cfg_0006_104122060811#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003433/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 377, + 151 + ], + [ + 382, + 141 + ], + [ + 391, + 153 + ], + [ + 364, + 150 + ], + [ + 385, + 164 + ], + [ + 358, + 161 + ] + ], + [ + [ + 359, + 162 + ], + [ + 362, + 149 + ], + [ + 367, + 157 + ], + [ + 350, + 166 + ], + [ + 364, + 170 + ], + [ + 347, + 179 + ] + ], + [ + [ + 384, + 161 + ], + [ + 387, + 155 + ], + [ + 378, + 147 + ], + [ + 391, + 174 + ], + [ + 375, + 153 + ], + [ + 389, + 180 + ] + ], + [ + [ + 370, + 165 + ], + [ + 383, + 158 + ], + [ + 377, + 172 + ], + [ + 363, + 159 + ], + [ + 365, + 179 + ], + [ + 350, + 166 + ] + ] + ] + }, + { + "task_id": "0003434", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0008_cfg_0006_104122060811#238", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003434/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 367, + 164 + ], + [ + 380, + 157 + ], + [ + 373, + 170 + ], + [ + 360, + 158 + ], + [ + 360, + 177 + ], + [ + 347, + 165 + ] + ], + [ + [ + 373, + 163 + ], + [ + 361, + 164 + ], + [ + 380, + 174 + ], + [ + 365, + 153 + ], + [ + 391, + 173 + ], + [ + 377, + 152 + ] + ], + [ + [ + 353, + 173 + ], + [ + 366, + 174 + ], + [ + 350, + 188 + ], + [ + 356, + 159 + ], + [ + 338, + 187 + ], + [ + 343, + 158 + ] + ], + [ + [ + 380, + 159 + ], + [ + 384, + 149 + ], + [ + 373, + 167 + ], + [ + 387, + 151 + ], + [ + 369, + 178 + ], + [ + 383, + 162 + ] + ] + ] + }, + { + "task_id": "0003435", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0008_cfg_0006_104122060811#306", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003435/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 366, + 144 + ], + [ + 372, + 131 + ], + [ + 356, + 138 + ], + [ + 377, + 150 + ], + [ + 350, + 152 + ], + [ + 371, + 163 + ] + ], + [ + [ + 364, + 156 + ], + [ + 367, + 168 + ], + [ + 368, + 148 + ], + [ + 359, + 165 + ], + [ + 365, + 136 + ], + [ + 356, + 153 + ] + ], + [ + [ + 358, + 153 + ], + [ + 371, + 145 + ], + [ + 364, + 160 + ], + [ + 351, + 145 + ], + [ + 351, + 166 + ], + [ + 338, + 152 + ] + ], + [ + [ + 361, + 133 + ], + [ + 351, + 134 + ], + [ + 351, + 128 + ], + [ + 371, + 139 + ], + [ + 362, + 127 + ], + [ + 382, + 139 + ] + ] + ] + }, + { + "task_id": "0003436", + "id": "RH20T_cfg6_task_0070_user_0020_scene_0008_cfg_0006_104122060811#411", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003436/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 352, + 154 + ], + [ + 358, + 167 + ], + [ + 350, + 152 + ], + [ + 353, + 156 + ], + [ + 344, + 139 + ], + [ + 347, + 143 + ] + ], + [ + [ + 354, + 152 + ], + [ + 366, + 145 + ], + [ + 360, + 159 + ], + [ + 347, + 146 + ], + [ + 347, + 166 + ], + [ + 334, + 153 + ] + ], + [ + [ + 349, + 135 + ], + [ + 357, + 145 + ], + [ + 339, + 145 + ], + [ + 359, + 124 + ], + [ + 331, + 135 + ], + [ + 351, + 115 + ] + ], + [ + [ + 347, + 164 + ], + [ + 338, + 152 + ], + [ + 335, + 173 + ], + [ + 359, + 156 + ], + [ + 344, + 185 + ], + [ + 368, + 168 + ] + ] + ] + }, + { + "task_id": "0003438", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0006_cfg_0005_104122063678#459", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003438/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 330, + 159 + ], + [ + 330, + 173 + ], + [ + 339, + 177 + ], + [ + 321, + 142 + ], + [ + 340, + 162 + ], + [ + 321, + 127 + ] + ], + [ + [ + 316, + 170 + ], + [ + 294, + 165 + ], + [ + 321, + 148 + ], + [ + 312, + 192 + ], + [ + 343, + 152 + ], + [ + 335, + 196 + ] + ], + [ + [ + 321, + 161 + ], + [ + 336, + 155 + ], + [ + 337, + 176 + ], + [ + 306, + 147 + ], + [ + 322, + 182 + ], + [ + 290, + 153 + ] + ], + [ + [ + 331, + 175 + ], + [ + 338, + 158 + ], + [ + 333, + 192 + ], + [ + 329, + 160 + ], + [ + 326, + 208 + ], + [ + 323, + 176 + ] + ] + ] + }, + { + "task_id": "0003443", + "id": "RH20T_cfg4_task_0018_user_0016_scene_0001_cfg_0004_038522062288#313", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003443/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 177, + 191 + ], + [ + 208, + 199 + ], + [ + 167, + 216 + ], + [ + 186, + 168 + ], + [ + 135, + 207 + ], + [ + 157, + 160 + ] + ], + [ + [ + 187, + 259 + ], + [ + 164, + 243 + ], + [ + 163, + 283 + ], + [ + 208, + 238 + ], + [ + 188, + 301 + ], + [ + 231, + 254 + ] + ], + [ + [ + 188, + 221 + ], + [ + 182, + 206 + ], + [ + 164, + 243 + ], + [ + 209, + 201 + ], + [ + 170, + 257 + ], + [ + 212, + 215 + ] + ], + [ + [ + 191, + 232 + ], + [ + 203, + 259 + ], + [ + 203, + 231 + ], + [ + 176, + 233 + ], + [ + 192, + 206 + ], + [ + 163, + 204 + ] + ] + ] + }, + { + "task_id": "0003445", + "id": "RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_037522061512#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003445/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 311, + 141 + ], + [ + 331, + 142 + ], + [ + 317, + 151 + ], + [ + 304, + 130 + ], + [ + 297, + 150 + ], + [ + 283, + 129 + ] + ], + [ + [ + 294, + 140 + ], + [ + 287, + 162 + ], + [ + 302, + 145 + ], + [ + 287, + 136 + ], + [ + 310, + 122 + ], + [ + 294, + 115 + ] + ], + [ + [ + 297, + 142 + ], + [ + 294, + 126 + ], + [ + 307, + 128 + ], + [ + 285, + 157 + ], + [ + 310, + 143 + ], + [ + 288, + 172 + ] + ], + [ + [ + 308, + 141 + ], + [ + 325, + 149 + ], + [ + 315, + 120 + ], + [ + 301, + 161 + ], + [ + 299, + 112 + ], + [ + 286, + 152 + ] + ] + ] + }, + { + "task_id": "0003447", + "id": "RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_037522061512#191", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003447/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 277, + 144 + ], + [ + 291, + 155 + ], + [ + 268, + 134 + ], + [ + 288, + 154 + ], + [ + 253, + 123 + ], + [ + 273, + 142 + ] + ], + [ + [ + 269, + 129 + ], + [ + 265, + 111 + ], + [ + 280, + 117 + ], + [ + 256, + 142 + ], + [ + 283, + 134 + ], + [ + 260, + 159 + ] + ], + [ + [ + 282, + 128 + ], + [ + 264, + 120 + ], + [ + 294, + 124 + ], + [ + 268, + 131 + ], + [ + 312, + 132 + ], + [ + 287, + 139 + ] + ], + [ + [ + 261, + 137 + ], + [ + 282, + 144 + ], + [ + 254, + 157 + ], + [ + 269, + 117 + ], + [ + 234, + 150 + ], + [ + 248, + 110 + ] + ] + ] + }, + { + "task_id": "0003449", + "id": "RH20T_cfg5_task_0085_user_0007_scene_0007_cfg_0005_105422061350#264", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the bubble ring from the assembled bubble ring and ball'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003449/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 212, + 79 + ], + [ + 196, + 55 + ], + [ + 189, + 94 + ], + [ + 235, + 65 + ], + [ + 204, + 116 + ], + [ + 249, + 88 + ] + ], + [ + [ + 208, + 132 + ], + [ + 228, + 120 + ], + [ + 193, + 110 + ], + [ + 223, + 155 + ], + [ + 171, + 122 + ], + [ + 202, + 168 + ] + ], + [ + [ + 230, + 114 + ], + [ + 229, + 96 + ], + [ + 256, + 120 + ], + [ + 204, + 107 + ], + [ + 256, + 136 + ], + [ + 206, + 124 + ] + ], + [ + [ + 234, + 94 + ], + [ + 251, + 85 + ], + [ + 254, + 109 + ], + [ + 214, + 79 + ], + [ + 236, + 119 + ], + [ + 194, + 87 + ] + ] + ] + }, + { + "task_id": "0003450", + "id": "RH20T_cfg7_task_0015_user_0020_scene_0007_cfg_0007_104122061602#246", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003450/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 405, + 139 + ], + [ + 393, + 143 + ], + [ + 425, + 145 + ], + [ + 385, + 134 + ], + [ + 436, + 141 + ], + [ + 397, + 131 + ] + ], + [ + [ + 397, + 123 + ], + [ + 380, + 132 + ], + [ + 405, + 145 + ], + [ + 389, + 101 + ], + [ + 421, + 137 + ], + [ + 406, + 94 + ] + ], + [ + [ + 397, + 117 + ], + [ + 419, + 111 + ], + [ + 400, + 97 + ], + [ + 395, + 135 + ], + [ + 377, + 104 + ], + [ + 375, + 141 + ] + ], + [ + [ + 413, + 127 + ], + [ + 421, + 149 + ], + [ + 404, + 128 + ], + [ + 424, + 126 + ], + [ + 397, + 106 + ], + [ + 416, + 102 + ] + ] + ] + }, + { + "task_id": "0003453", + "id": "RH20T_cfg6_task_0047_user_0014_scene_0002_cfg_0006_104122060811#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003453/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 459, + 110 + ], + [ + 475, + 106 + ], + [ + 455, + 96 + ], + [ + 463, + 125 + ], + [ + 440, + 100 + ], + [ + 448, + 129 + ] + ], + [ + [ + 463, + 114 + ], + [ + 467, + 99 + ], + [ + 477, + 117 + ], + [ + 448, + 111 + ], + [ + 473, + 132 + ], + [ + 445, + 125 + ] + ], + [ + [ + 443, + 127 + ], + [ + 433, + 135 + ], + [ + 433, + 116 + ], + [ + 452, + 138 + ], + [ + 443, + 107 + ], + [ + 462, + 130 + ] + ], + [ + [ + 444, + 109 + ], + [ + 443, + 103 + ], + [ + 458, + 102 + ], + [ + 431, + 117 + ], + [ + 459, + 108 + ], + [ + 432, + 122 + ] + ] + ] + }, + { + "task_id": "0003454", + "id": "RH20T_cfg1_task_0005_user_0008_scene_0001_cfg_0001_750612070851#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003454/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 219, + 233 + ], + [ + 213, + 216 + ], + [ + 182, + 234 + ], + [ + 257, + 231 + ], + [ + 190, + 248 + ], + [ + 260, + 245 + ] + ], + [ + [ + 187, + 222 + ], + [ + 188, + 260 + ], + [ + 208, + 214 + ], + [ + 167, + 230 + ], + [ + 206, + 176 + ], + [ + 166, + 195 + ] + ], + [ + [ + 240, + 221 + ], + [ + 240, + 189 + ], + [ + 209, + 234 + ], + [ + 272, + 206 + ], + [ + 207, + 269 + ], + [ + 273, + 242 + ] + ], + [ + [ + 210, + 229 + ], + [ + 229, + 256 + ], + [ + 176, + 247 + ], + [ + 243, + 212 + ], + [ + 160, + 220 + ], + [ + 225, + 188 + ] + ] + ] + }, + { + "task_id": "0003458", + "id": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_750612070853#590", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003458/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 460, + 353 + ], + [ + 468, + 306 + ], + [ + 428, + 360 + ], + [ + 498, + 339 + ], + [ + 418, + 360 + ], + [ + 492, + 360 + ] + ], + [ + [ + 410, + 264 + ], + [ + 410, + 250 + ], + [ + 420, + 317 + ], + [ + 401, + 216 + ], + [ + 420, + 325 + ], + [ + 403, + 232 + ] + ], + [ + [ + 457, + 304 + ], + [ + 447, + 259 + ], + [ + 404, + 307 + ], + [ + 516, + 302 + ], + [ + 415, + 347 + ], + [ + 524, + 345 + ] + ], + [ + [ + 486, + 238 + ], + [ + 432, + 231 + ], + [ + 490, + 216 + ], + [ + 483, + 256 + ], + [ + 554, + 224 + ], + [ + 535, + 264 + ] + ] + ] + }, + { + "task_id": "0003461", + "id": "RH20T_cfg5_task_0015_user_0007_scene_0006_cfg_0005_036422060215#208", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003461/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 286, + 117 + ], + [ + 290, + 98 + ], + [ + 306, + 126 + ], + [ + 267, + 107 + ], + [ + 302, + 143 + ], + [ + 264, + 125 + ] + ], + [ + [ + 288, + 96 + ], + [ + 282, + 115 + ], + [ + 292, + 86 + ], + [ + 284, + 106 + ], + [ + 298, + 67 + ], + [ + 289, + 88 + ] + ], + [ + [ + 287, + 131 + ], + [ + 306, + 122 + ], + [ + 279, + 123 + ], + [ + 293, + 138 + ], + [ + 259, + 131 + ], + [ + 275, + 146 + ] + ], + [ + [ + 294, + 147 + ], + [ + 284, + 134 + ], + [ + 294, + 163 + ], + [ + 294, + 130 + ], + [ + 303, + 177 + ], + [ + 304, + 145 + ] + ] + ] + }, + { + "task_id": "0003462", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0001_cfg_0007_037522061512#202", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003462/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 236, + 137 + ], + [ + 227, + 139 + ], + [ + 253, + 150 + ], + [ + 218, + 123 + ], + [ + 263, + 149 + ], + [ + 227, + 121 + ] + ], + [ + [ + 243, + 115 + ], + [ + 256, + 116 + ], + [ + 239, + 137 + ], + [ + 248, + 93 + ], + [ + 225, + 137 + ], + [ + 233, + 91 + ] + ], + [ + [ + 237, + 129 + ], + [ + 222, + 118 + ], + [ + 220, + 140 + ], + [ + 253, + 119 + ], + [ + 236, + 152 + ], + [ + 269, + 130 + ] + ], + [ + [ + 245, + 100 + ], + [ + 231, + 91 + ], + [ + 229, + 115 + ], + [ + 260, + 86 + ], + [ + 244, + 125 + ], + [ + 275, + 95 + ] + ] + ] + }, + { + "task_id": "0003464", + "id": "RH20T_cfg4_task_0217_user_0010_scene_0003_cfg_0004_104122062823#114", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003464/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 313, + 46 + ], + [ + 310, + 37 + ], + [ + 286, + 61 + ], + [ + 341, + 30 + ], + [ + 288, + 71 + ], + [ + 347, + 39 + ] + ], + [ + [ + 286, + 56 + ], + [ + 281, + 75 + ], + [ + 306, + 39 + ], + [ + 268, + 72 + ], + [ + 309, + 21 + ], + [ + 273, + 54 + ] + ], + [ + [ + 295, + 30 + ], + [ + 292, + 1 + ], + [ + 324, + 31 + ], + [ + 266, + 29 + ], + [ + 325, + 57 + ], + [ + 270, + 55 + ] + ], + [ + [ + 302, + 35 + ], + [ + 297, + 34 + ], + [ + 283, + 57 + ], + [ + 321, + 12 + ], + [ + 289, + 57 + ], + [ + 325, + 14 + ] + ] + ] + }, + { + "task_id": "0003465", + "id": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_104122061602#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003465/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 425, + 172 + ], + [ + 420, + 165 + ], + [ + 446, + 161 + ], + [ + 405, + 184 + ], + [ + 452, + 169 + ], + [ + 409, + 192 + ] + ], + [ + [ + 411, + 157 + ], + [ + 433, + 150 + ], + [ + 418, + 179 + ], + [ + 403, + 136 + ], + [ + 395, + 186 + ], + [ + 381, + 143 + ] + ], + [ + [ + 410, + 170 + ], + [ + 393, + 167 + ], + [ + 427, + 173 + ], + [ + 395, + 168 + ], + [ + 444, + 177 + ], + [ + 412, + 171 + ] + ], + [ + [ + 415, + 146 + ], + [ + 423, + 132 + ], + [ + 429, + 164 + ], + [ + 402, + 129 + ], + [ + 422, + 176 + ], + [ + 396, + 143 + ] + ] + ] + }, + { + "task_id": "0003466", + "id": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_104122061602#302", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003466/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 409, + 133 + ], + [ + 417, + 122 + ], + [ + 419, + 151 + ], + [ + 398, + 115 + ], + [ + 411, + 160 + ], + [ + 391, + 126 + ] + ], + [ + [ + 431, + 141 + ], + [ + 414, + 149 + ], + [ + 422, + 121 + ], + [ + 440, + 162 + ], + [ + 439, + 114 + ], + [ + 457, + 154 + ] + ], + [ + [ + 402, + 124 + ], + [ + 399, + 134 + ], + [ + 400, + 144 + ], + [ + 404, + 106 + ], + [ + 403, + 134 + ], + [ + 407, + 94 + ] + ], + [ + [ + 428, + 139 + ], + [ + 417, + 152 + ], + [ + 441, + 157 + ], + [ + 416, + 121 + ], + [ + 451, + 145 + ], + [ + 427, + 109 + ] + ] + ] + }, + { + "task_id": "0003468", + "id": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_104122061602#597", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003468/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 532, + 133 + ], + [ + 551, + 116 + ], + [ + 552, + 155 + ], + [ + 514, + 112 + ], + [ + 533, + 170 + ], + [ + 498, + 128 + ] + ], + [ + [ + 542, + 149 + ], + [ + 528, + 164 + ], + [ + 567, + 167 + ], + [ + 517, + 132 + ], + [ + 580, + 152 + ], + [ + 531, + 119 + ] + ], + [ + [ + 545, + 153 + ], + [ + 528, + 174 + ], + [ + 571, + 161 + ], + [ + 521, + 146 + ], + [ + 589, + 139 + ], + [ + 537, + 127 + ] + ], + [ + [ + 521, + 145 + ], + [ + 505, + 132 + ], + [ + 514, + 126 + ], + [ + 527, + 162 + ], + [ + 532, + 140 + ], + [ + 544, + 176 + ] + ] + ] + }, + { + "task_id": "0003469", + "id": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_104122061602#633", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003469/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 511, + 138 + ], + [ + 483, + 151 + ], + [ + 504, + 135 + ], + [ + 519, + 142 + ], + [ + 532, + 123 + ], + [ + 550, + 129 + ] + ], + [ + [ + 524, + 164 + ], + [ + 519, + 156 + ], + [ + 552, + 149 + ], + [ + 498, + 178 + ], + [ + 560, + 156 + ], + [ + 501, + 187 + ] + ], + [ + [ + 519, + 151 + ], + [ + 533, + 141 + ], + [ + 498, + 132 + ], + [ + 541, + 171 + ], + [ + 482, + 142 + ], + [ + 526, + 182 + ] + ], + [ + [ + 524, + 131 + ], + [ + 546, + 116 + ], + [ + 543, + 154 + ], + [ + 506, + 109 + ], + [ + 522, + 167 + ], + [ + 488, + 124 + ] + ] + ] + }, + { + "task_id": "0003471", + "id": "RH20T_cfg1_task_0065_user_0014_scene_0004_cfg_0001_038522063145#104", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003471/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 492, + 177 + ], + [ + 500, + 160 + ], + [ + 514, + 189 + ], + [ + 471, + 166 + ], + [ + 505, + 206 + ], + [ + 464, + 182 + ] + ], + [ + [ + 476, + 166 + ], + [ + 459, + 180 + ], + [ + 477, + 182 + ], + [ + 474, + 152 + ], + [ + 496, + 167 + ], + [ + 490, + 138 + ] + ], + [ + [ + 476, + 176 + ], + [ + 487, + 156 + ], + [ + 490, + 174 + ], + [ + 461, + 179 + ], + [ + 480, + 193 + ], + [ + 450, + 198 + ] + ], + [ + [ + 475, + 183 + ], + [ + 461, + 193 + ], + [ + 488, + 204 + ], + [ + 462, + 163 + ], + [ + 500, + 194 + ], + [ + 475, + 154 + ] + ] + ] + }, + { + "task_id": "0003474", + "id": "RH20T_cfg1_task_0032_user_0015_scene_0004_cfg_0001_035622060973#208", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003474/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 192, + 179 + ], + [ + 178, + 198 + ], + [ + 174, + 170 + ], + [ + 211, + 188 + ], + [ + 188, + 153 + ], + [ + 224, + 170 + ] + ], + [ + [ + 208, + 164 + ], + [ + 190, + 177 + ], + [ + 197, + 149 + ], + [ + 219, + 180 + ], + [ + 214, + 136 + ], + [ + 236, + 166 + ] + ], + [ + [ + 198, + 168 + ], + [ + 198, + 180 + ], + [ + 219, + 175 + ], + [ + 178, + 161 + ], + [ + 220, + 163 + ], + [ + 177, + 148 + ] + ], + [ + [ + 194, + 175 + ], + [ + 191, + 164 + ], + [ + 175, + 165 + ], + [ + 213, + 185 + ], + [ + 179, + 175 + ], + [ + 215, + 194 + ] + ] + ] + }, + { + "task_id": "0003476", + "id": "RH20T_cfg1_task_0004_user_0005_scene_0003_cfg_0001_819112070044#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003476/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 441, + 204 + ], + [ + 445, + 187 + ], + [ + 457, + 198 + ], + [ + 425, + 210 + ], + [ + 454, + 214 + ], + [ + 422, + 226 + ] + ], + [ + [ + 452, + 208 + ], + [ + 468, + 204 + ], + [ + 440, + 206 + ], + [ + 465, + 210 + ], + [ + 424, + 209 + ], + [ + 448, + 214 + ] + ], + [ + [ + 454, + 197 + ], + [ + 439, + 183 + ], + [ + 467, + 189 + ], + [ + 440, + 205 + ], + [ + 483, + 203 + ], + [ + 455, + 220 + ] + ], + [ + [ + 438, + 185 + ], + [ + 438, + 199 + ], + [ + 418, + 181 + ], + [ + 458, + 189 + ], + [ + 417, + 167 + ], + [ + 459, + 175 + ] + ] + ] + }, + { + "task_id": "0003480", + "id": "RH20T_cfg1_task_0061_user_0014_scene_0008_cfg_0001_750612070853#171", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003480/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 457, + 216 + ], + [ + 412, + 201 + ], + [ + 467, + 170 + ], + [ + 446, + 265 + ], + [ + 510, + 184 + ], + [ + 492, + 279 + ] + ], + [ + [ + 441, + 209 + ], + [ + 440, + 169 + ], + [ + 393, + 199 + ], + [ + 495, + 222 + ], + [ + 397, + 232 + ], + [ + 492, + 257 + ] + ], + [ + [ + 450, + 216 + ], + [ + 458, + 219 + ], + [ + 441, + 274 + ], + [ + 459, + 158 + ], + [ + 435, + 265 + ], + [ + 451, + 162 + ] + ], + [ + [ + 465, + 205 + ], + [ + 512, + 240 + ], + [ + 488, + 215 + ], + [ + 447, + 197 + ], + [ + 434, + 174 + ], + [ + 405, + 165 + ] + ] + ] + }, + { + "task_id": "0003482", + "id": "RH20T_cfg3_task_0008_user_0016_scene_0005_cfg_0003_036422060909#199", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003482/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 354, + 80 + ], + [ + 357, + 64 + ], + [ + 370, + 72 + ], + [ + 336, + 88 + ], + [ + 366, + 86 + ], + [ + 333, + 102 + ] + ], + [ + [ + 371, + 69 + ], + [ + 381, + 73 + ], + [ + 382, + 86 + ], + [ + 360, + 51 + ], + [ + 373, + 82 + ], + [ + 352, + 48 + ] + ], + [ + [ + 379, + 85 + ], + [ + 378, + 96 + ], + [ + 377, + 104 + ], + [ + 381, + 67 + ], + [ + 378, + 93 + ], + [ + 382, + 54 + ] + ], + [ + [ + 370, + 72 + ], + [ + 362, + 54 + ], + [ + 390, + 61 + ], + [ + 351, + 82 + ], + [ + 398, + 79 + ], + [ + 359, + 100 + ] + ] + ] + }, + { + "task_id": "0003484", + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_037522061512#145", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003484/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 137 + ], + [ + 270, + 116 + ], + [ + 278, + 135 + ], + [ + 253, + 140 + ], + [ + 274, + 154 + ], + [ + 248, + 161 + ] + ], + [ + [ + 271, + 123 + ], + [ + 278, + 102 + ], + [ + 293, + 129 + ], + [ + 250, + 118 + ], + [ + 286, + 150 + ], + [ + 243, + 139 + ] + ], + [ + [ + 283, + 147 + ], + [ + 294, + 134 + ], + [ + 301, + 147 + ], + [ + 264, + 147 + ], + [ + 290, + 159 + ], + [ + 254, + 160 + ] + ], + [ + [ + 277, + 143 + ], + [ + 269, + 159 + ], + [ + 258, + 133 + ], + [ + 297, + 154 + ], + [ + 266, + 118 + ], + [ + 303, + 139 + ] + ] + ] + }, + { + "task_id": "0003485", + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_037522061512#180", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003485/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 276, + 127 + ], + [ + 276, + 104 + ], + [ + 278, + 126 + ], + [ + 274, + 127 + ], + [ + 279, + 149 + ], + [ + 275, + 149 + ] + ], + [ + [ + 252, + 131 + ], + [ + 242, + 127 + ], + [ + 231, + 130 + ], + [ + 272, + 132 + ], + [ + 242, + 135 + ], + [ + 283, + 137 + ] + ], + [ + [ + 255, + 157 + ], + [ + 247, + 177 + ], + [ + 245, + 157 + ], + [ + 264, + 157 + ], + [ + 253, + 136 + ], + [ + 272, + 138 + ] + ], + [ + [ + 272, + 146 + ], + [ + 276, + 124 + ], + [ + 284, + 144 + ], + [ + 259, + 148 + ], + [ + 281, + 165 + ], + [ + 255, + 171 + ] + ] + ] + }, + { + "task_id": "0003487", + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_037522061512#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003487/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 317, + 154 + ], + [ + 339, + 158 + ], + [ + 311, + 149 + ], + [ + 323, + 158 + ], + [ + 288, + 145 + ], + [ + 302, + 154 + ] + ], + [ + [ + 301, + 154 + ], + [ + 306, + 155 + ], + [ + 277, + 157 + ], + [ + 327, + 152 + ], + [ + 271, + 156 + ], + [ + 323, + 151 + ] + ], + [ + [ + 286, + 172 + ], + [ + 290, + 148 + ], + [ + 298, + 170 + ], + [ + 273, + 174 + ], + [ + 294, + 191 + ], + [ + 269, + 198 + ] + ], + [ + [ + 291, + 164 + ], + [ + 305, + 147 + ], + [ + 283, + 153 + ], + [ + 299, + 175 + ], + [ + 268, + 172 + ], + [ + 285, + 192 + ] + ] + ] + }, + { + "task_id": "0003488", + "id": "RH20T_cfg4_task_0091_user_0014_scene_0007_cfg_0004_104422070011#116", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003488/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 221, + 57 + ], + [ + 225, + 39 + ], + [ + 195, + 40 + ], + [ + 245, + 73 + ], + [ + 192, + 57 + ], + [ + 240, + 88 + ] + ], + [ + [ + 229, + 52 + ], + [ + 240, + 57 + ], + [ + 222, + 81 + ], + [ + 236, + 22 + ], + [ + 213, + 75 + ], + [ + 226, + 19 + ] + ], + [ + [ + 223, + 49 + ], + [ + 193, + 41 + ], + [ + 231, + 21 + ], + [ + 216, + 75 + ], + [ + 260, + 30 + ], + [ + 243, + 82 + ] + ], + [ + [ + 229, + 82 + ], + [ + 235, + 68 + ], + [ + 218, + 105 + ], + [ + 240, + 61 + ], + [ + 213, + 117 + ], + [ + 234, + 75 + ] + ] + ] + }, + { + "task_id": "0003489", + "id": "RH20T_cfg5_task_0039_user_0007_scene_0003_cfg_0005_037522062165#254", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003489/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 292, + 239 + ], + [ + 308, + 256 + ], + [ + 293, + 263 + ], + [ + 291, + 218 + ], + [ + 275, + 245 + ], + [ + 275, + 200 + ] + ], + [ + [ + 294, + 290 + ], + [ + 290, + 315 + ], + [ + 299, + 268 + ], + [ + 290, + 311 + ], + [ + 302, + 247 + ], + [ + 293, + 289 + ] + ], + [ + [ + 299, + 258 + ], + [ + 302, + 242 + ], + [ + 276, + 240 + ], + [ + 323, + 276 + ], + [ + 275, + 255 + ], + [ + 320, + 290 + ] + ], + [ + [ + 278, + 272 + ], + [ + 261, + 296 + ], + [ + 304, + 280 + ], + [ + 254, + 264 + ], + [ + 319, + 257 + ], + [ + 270, + 243 + ] + ] + ] + }, + { + "task_id": "0003490", + "id": "RH20T_cfg1_task_0004_user_0001_scene_0005_cfg_0001_819112070044#496", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003490/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 425, + 161 + ], + [ + 436, + 147 + ], + [ + 443, + 162 + ], + [ + 407, + 160 + ], + [ + 433, + 175 + ], + [ + 398, + 173 + ] + ], + [ + [ + 426, + 182 + ], + [ + 445, + 183 + ], + [ + 414, + 181 + ], + [ + 438, + 183 + ], + [ + 395, + 181 + ], + [ + 420, + 183 + ] + ], + [ + [ + 419, + 177 + ], + [ + 398, + 179 + ], + [ + 421, + 197 + ], + [ + 418, + 156 + ], + [ + 442, + 195 + ], + [ + 440, + 154 + ] + ], + [ + [ + 450, + 146 + ], + [ + 455, + 131 + ], + [ + 463, + 135 + ], + [ + 437, + 157 + ], + [ + 458, + 149 + ], + [ + 432, + 170 + ] + ] + ] + }, + { + "task_id": "0003492", + "id": "RH20T_cfg5_task_0018_user_0010_scene_0009_cfg_0005_105422061350#356", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003492/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 202, + 159 + ], + [ + 193, + 173 + ], + [ + 193, + 180 + ], + [ + 212, + 136 + ], + [ + 202, + 166 + ], + [ + 220, + 124 + ] + ], + [ + [ + 237, + 125 + ], + [ + 240, + 119 + ], + [ + 213, + 140 + ], + [ + 261, + 110 + ], + [ + 208, + 148 + ], + [ + 260, + 116 + ] + ], + [ + [ + 210, + 134 + ], + [ + 208, + 115 + ], + [ + 237, + 141 + ], + [ + 182, + 126 + ], + [ + 237, + 157 + ], + [ + 185, + 144 + ] + ], + [ + [ + 219, + 135 + ], + [ + 201, + 112 + ], + [ + 203, + 150 + ], + [ + 237, + 119 + ], + [ + 221, + 171 + ], + [ + 256, + 143 + ] + ] + ] + }, + { + "task_id": "0003495", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0002_cfg_0006_104122061018#333", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003495/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 400, + 241 + ], + [ + 365, + 253 + ], + [ + 381, + 225 + ], + [ + 422, + 261 + ], + [ + 413, + 215 + ], + [ + 458, + 248 + ] + ], + [ + [ + 370, + 213 + ], + [ + 353, + 180 + ], + [ + 362, + 229 + ], + [ + 379, + 194 + ], + [ + 378, + 263 + ], + [ + 398, + 232 + ] + ], + [ + [ + 331, + 190 + ], + [ + 338, + 171 + ], + [ + 366, + 175 + ], + [ + 298, + 204 + ], + [ + 360, + 196 + ], + [ + 287, + 227 + ] + ], + [ + [ + 332, + 218 + ], + [ + 309, + 249 + ], + [ + 357, + 233 + ], + [ + 310, + 205 + ], + [ + 381, + 200 + ], + [ + 332, + 177 + ] + ] + ] + }, + { + "task_id": "0003496", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0002_cfg_0003_104122062295#142", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003496/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 161, + 94 + ], + [ + 165, + 73 + ], + [ + 168, + 118 + ], + [ + 153, + 66 + ], + [ + 164, + 141 + ], + [ + 147, + 89 + ] + ], + [ + [ + 148, + 100 + ], + [ + 149, + 67 + ], + [ + 132, + 83 + ], + [ + 165, + 119 + ], + [ + 132, + 112 + ], + [ + 163, + 149 + ] + ], + [ + [ + 167, + 114 + ], + [ + 177, + 146 + ], + [ + 141, + 130 + ], + [ + 190, + 101 + ], + [ + 128, + 94 + ], + [ + 181, + 68 + ] + ], + [ + [ + 162, + 110 + ], + [ + 177, + 84 + ], + [ + 155, + 127 + ], + [ + 171, + 91 + ], + [ + 139, + 155 + ], + [ + 154, + 122 + ] + ] + ] + }, + { + "task_id": "0003498", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0002_cfg_0003_104122062295#235", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003498/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 362, + 304 + ], + [ + 351, + 337 + ], + [ + 408, + 311 + ], + [ + 316, + 296 + ], + [ + 423, + 275 + ], + [ + 326, + 259 + ] + ], + [ + [ + 325, + 252 + ], + [ + 296, + 251 + ], + [ + 293, + 269 + ], + [ + 360, + 233 + ], + [ + 318, + 268 + ], + [ + 383, + 235 + ] + ], + [ + [ + 336, + 261 + ], + [ + 344, + 222 + ], + [ + 304, + 235 + ], + [ + 372, + 292 + ], + [ + 298, + 269 + ], + [ + 362, + 328 + ] + ], + [ + [ + 314, + 320 + ], + [ + 291, + 274 + ], + [ + 335, + 299 + ], + [ + 296, + 337 + ], + [ + 364, + 354 + ], + [ + 318, + 360 + ] + ] + ] + }, + { + "task_id": "0003499", + "id": "RH20T_cfg4_task_0091_user_0007_scene_0006_cfg_0004_038522062288#123", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003499/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 189, + 151 + ], + [ + 216, + 163 + ], + [ + 183, + 166 + ], + [ + 195, + 138 + ], + [ + 153, + 154 + ], + [ + 168, + 127 + ] + ], + [ + [ + 192, + 148 + ], + [ + 184, + 127 + ], + [ + 163, + 151 + ], + [ + 221, + 146 + ], + [ + 171, + 170 + ], + [ + 226, + 164 + ] + ], + [ + [ + 193, + 178 + ], + [ + 195, + 157 + ], + [ + 170, + 164 + ], + [ + 214, + 191 + ], + [ + 166, + 187 + ], + [ + 212, + 213 + ] + ], + [ + [ + 162, + 145 + ], + [ + 185, + 151 + ], + [ + 158, + 173 + ], + [ + 166, + 117 + ], + [ + 132, + 168 + ], + [ + 140, + 108 + ] + ] + ] + }, + { + "task_id": "0003500", + "id": "RH20T_cfg6_task_0029_user_0014_scene_0005_cfg_0006_104122061602#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003500/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 445, + 165 + ], + [ + 459, + 148 + ], + [ + 464, + 191 + ], + [ + 428, + 141 + ], + [ + 450, + 206 + ], + [ + 416, + 157 + ] + ], + [ + [ + 476, + 143 + ], + [ + 476, + 176 + ], + [ + 475, + 146 + ], + [ + 478, + 140 + ], + [ + 475, + 116 + ], + [ + 479, + 106 + ] + ], + [ + [ + 444, + 189 + ], + [ + 459, + 162 + ], + [ + 436, + 189 + ], + [ + 453, + 188 + ], + [ + 422, + 214 + ], + [ + 437, + 216 + ] + ], + [ + [ + 431, + 154 + ], + [ + 410, + 179 + ], + [ + 418, + 145 + ], + [ + 445, + 165 + ], + [ + 438, + 121 + ], + [ + 468, + 138 + ] + ] + ] + }, + { + "task_id": "0003502", + "id": "RH20T_cfg1_task_0064_user_0015_scene_0008_cfg_0001_038522063145#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003502/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 129 + ], + [ + 336, + 118 + ], + [ + 325, + 151 + ], + [ + 304, + 106 + ], + [ + 304, + 161 + ], + [ + 283, + 117 + ] + ], + [ + [ + 320, + 124 + ], + [ + 330, + 145 + ], + [ + 330, + 122 + ], + [ + 311, + 125 + ], + [ + 319, + 99 + ], + [ + 300, + 104 + ] + ], + [ + [ + 330, + 113 + ], + [ + 340, + 91 + ], + [ + 348, + 113 + ], + [ + 311, + 113 + ], + [ + 339, + 132 + ], + [ + 302, + 133 + ] + ], + [ + [ + 314, + 118 + ], + [ + 330, + 127 + ], + [ + 307, + 104 + ], + [ + 320, + 133 + ], + [ + 291, + 94 + ], + [ + 303, + 124 + ] + ] + ] + }, + { + "task_id": "0003506", + "id": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_104122063633#406", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003506/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 129, + 249 + ], + [ + 122, + 226 + ], + [ + 142, + 255 + ], + [ + 114, + 242 + ], + [ + 149, + 277 + ], + [ + 121, + 267 + ] + ], + [ + [ + 161, + 215 + ], + [ + 186, + 220 + ], + [ + 155, + 213 + ], + [ + 167, + 216 + ], + [ + 129, + 207 + ], + [ + 144, + 210 + ] + ], + [ + [ + 162, + 217 + ], + [ + 135, + 220 + ], + [ + 162, + 214 + ], + [ + 162, + 220 + ], + [ + 189, + 211 + ], + [ + 187, + 217 + ] + ], + [ + [ + 132, + 228 + ], + [ + 140, + 204 + ], + [ + 137, + 231 + ], + [ + 127, + 225 + ], + [ + 129, + 255 + ], + [ + 118, + 252 + ] + ] + ] + }, + { + "task_id": "0003508", + "id": "RH20T_cfg5_task_0016_user_0010_scene_0001_cfg_0005_104422070044#92", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003508/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 375, + 198 + ], + [ + 388, + 195 + ], + [ + 368, + 181 + ], + [ + 383, + 216 + ], + [ + 354, + 184 + ], + [ + 369, + 220 + ] + ], + [ + [ + 375, + 200 + ], + [ + 371, + 182 + ], + [ + 371, + 195 + ], + [ + 380, + 205 + ], + [ + 375, + 213 + ], + [ + 385, + 224 + ] + ], + [ + [ + 382, + 197 + ], + [ + 381, + 183 + ], + [ + 399, + 204 + ], + [ + 366, + 190 + ], + [ + 400, + 217 + ], + [ + 367, + 203 + ] + ], + [ + [ + 380, + 204 + ], + [ + 374, + 186 + ], + [ + 367, + 201 + ], + [ + 394, + 206 + ], + [ + 372, + 218 + ], + [ + 399, + 225 + ] + ] + ] + }, + { + "task_id": "0003509", + "id": "RH20T_cfg5_task_0016_user_0010_scene_0001_cfg_0005_104422070044#114", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003509/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 379, + 170 + ], + [ + 379, + 154 + ], + [ + 392, + 180 + ], + [ + 368, + 160 + ], + [ + 391, + 195 + ], + [ + 368, + 175 + ] + ], + [ + [ + 389, + 164 + ], + [ + 396, + 183 + ], + [ + 371, + 174 + ], + [ + 406, + 154 + ], + [ + 365, + 156 + ], + [ + 399, + 136 + ] + ], + [ + [ + 374, + 202 + ], + [ + 362, + 188 + ], + [ + 382, + 206 + ], + [ + 366, + 197 + ], + [ + 393, + 220 + ], + [ + 378, + 211 + ] + ], + [ + [ + 356, + 172 + ], + [ + 363, + 155 + ], + [ + 338, + 165 + ], + [ + 374, + 180 + ], + [ + 331, + 182 + ], + [ + 366, + 197 + ] + ] + ] + }, + { + "task_id": "0003510", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0006_cfg_0005_104122061850#181", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003510/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 356, + 163 + ], + [ + 374, + 143 + ], + [ + 352, + 147 + ], + [ + 360, + 181 + ], + [ + 336, + 165 + ], + [ + 342, + 200 + ] + ], + [ + [ + 356, + 176 + ], + [ + 355, + 161 + ], + [ + 342, + 197 + ], + [ + 369, + 156 + ], + [ + 344, + 210 + ], + [ + 369, + 170 + ] + ], + [ + [ + 346, + 176 + ], + [ + 322, + 167 + ], + [ + 336, + 199 + ], + [ + 356, + 152 + ], + [ + 360, + 208 + ], + [ + 381, + 162 + ] + ], + [ + [ + 356, + 185 + ], + [ + 378, + 176 + ], + [ + 361, + 210 + ], + [ + 352, + 158 + ], + [ + 339, + 220 + ], + [ + 328, + 167 + ] + ] + ] + }, + { + "task_id": "0003511", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0006_cfg_0005_104122061850#378", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003511/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 386, + 136 + ], + [ + 396, + 119 + ], + [ + 377, + 150 + ], + [ + 395, + 123 + ], + [ + 368, + 165 + ], + [ + 386, + 138 + ] + ], + [ + [ + 391, + 90 + ], + [ + 385, + 108 + ], + [ + 395, + 74 + ], + [ + 386, + 107 + ], + [ + 402, + 54 + ], + [ + 393, + 88 + ] + ], + [ + [ + 375, + 104 + ], + [ + 376, + 87 + ], + [ + 362, + 121 + ], + [ + 387, + 88 + ], + [ + 361, + 136 + ], + [ + 385, + 104 + ] + ], + [ + [ + 389, + 99 + ], + [ + 369, + 90 + ], + [ + 400, + 76 + ], + [ + 379, + 122 + ], + [ + 422, + 86 + ], + [ + 399, + 132 + ] + ] + ] + }, + { + "task_id": "0003513", + "id": "RH20T_cfg1_task_0030_user_0015_scene_0001_cfg_0001_035622060973#237", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour out the test tube'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003513/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 276, + 252 + ], + [ + 264, + 236 + ], + [ + 256, + 255 + ], + [ + 298, + 248 + ], + [ + 267, + 270 + ], + [ + 309, + 264 + ] + ], + [ + [ + 285, + 248 + ], + [ + 304, + 232 + ], + [ + 286, + 247 + ], + [ + 284, + 250 + ], + [ + 266, + 264 + ], + [ + 266, + 266 + ] + ], + [ + [ + 280, + 253 + ], + [ + 274, + 230 + ], + [ + 257, + 258 + ], + [ + 301, + 248 + ], + [ + 262, + 282 + ], + [ + 307, + 270 + ] + ], + [ + [ + 280, + 257 + ], + [ + 260, + 269 + ], + [ + 282, + 264 + ], + [ + 278, + 249 + ], + [ + 301, + 252 + ], + [ + 299, + 236 + ] + ] + ] + }, + { + "task_id": "0003514", + "id": "RH20T_cfg4_task_0013_user_0016_scene_0007_cfg_0004_104122063550#67", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003514/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 321, + 162 + ], + [ + 335, + 174 + ], + [ + 328, + 137 + ], + [ + 315, + 185 + ], + [ + 315, + 127 + ], + [ + 303, + 173 + ] + ], + [ + [ + 328, + 168 + ], + [ + 357, + 169 + ], + [ + 329, + 140 + ], + [ + 328, + 195 + ], + [ + 300, + 139 + ], + [ + 300, + 194 + ] + ], + [ + [ + 346, + 175 + ], + [ + 351, + 160 + ], + [ + 353, + 198 + ], + [ + 340, + 153 + ], + [ + 349, + 210 + ], + [ + 336, + 166 + ] + ], + [ + [ + 342, + 159 + ], + [ + 353, + 147 + ], + [ + 324, + 139 + ], + [ + 359, + 178 + ], + [ + 311, + 151 + ], + [ + 347, + 191 + ] + ] + ] + }, + { + "task_id": "0003518", + "id": "RH20T_cfg6_task_0006_user_0014_scene_0005_cfg_0006_104122064161#90", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003518/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 376, + 236 + ], + [ + 352, + 230 + ], + [ + 378, + 227 + ], + [ + 373, + 247 + ], + [ + 402, + 233 + ], + [ + 399, + 254 + ] + ], + [ + [ + 359, + 227 + ], + [ + 349, + 251 + ], + [ + 369, + 241 + ], + [ + 350, + 215 + ], + [ + 380, + 215 + ], + [ + 360, + 192 + ] + ], + [ + [ + 368, + 219 + ], + [ + 343, + 213 + ], + [ + 372, + 214 + ], + [ + 364, + 223 + ], + [ + 397, + 221 + ], + [ + 387, + 229 + ] + ], + [ + [ + 366, + 228 + ], + [ + 371, + 216 + ], + [ + 363, + 251 + ], + [ + 370, + 206 + ], + [ + 359, + 262 + ], + [ + 365, + 218 + ] + ] + ] + }, + { + "task_id": "0003519", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122061602#165", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003519/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 445, + 193 + ], + [ + 461, + 188 + ], + [ + 453, + 170 + ], + [ + 438, + 215 + ], + [ + 436, + 174 + ], + [ + 421, + 220 + ] + ], + [ + [ + 479, + 212 + ], + [ + 498, + 199 + ], + [ + 475, + 229 + ], + [ + 483, + 193 + ], + [ + 457, + 243 + ], + [ + 463, + 207 + ] + ], + [ + [ + 503, + 195 + ], + [ + 498, + 185 + ], + [ + 528, + 207 + ], + [ + 480, + 183 + ], + [ + 532, + 216 + ], + [ + 485, + 193 + ] + ], + [ + [ + 478, + 203 + ], + [ + 492, + 192 + ], + [ + 493, + 226 + ], + [ + 463, + 180 + ], + [ + 479, + 235 + ], + [ + 451, + 191 + ] + ] + ] + }, + { + "task_id": "0003520", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0007_cfg_0006_104122061602#367", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003520/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 444, + 233 + ], + [ + 456, + 251 + ], + [ + 470, + 221 + ], + [ + 420, + 245 + ], + [ + 458, + 205 + ], + [ + 411, + 229 + ] + ], + [ + [ + 477, + 226 + ], + [ + 493, + 219 + ], + [ + 490, + 252 + ], + [ + 464, + 200 + ], + [ + 475, + 257 + ], + [ + 450, + 208 + ] + ], + [ + [ + 464, + 247 + ], + [ + 477, + 223 + ], + [ + 481, + 258 + ], + [ + 446, + 235 + ], + [ + 468, + 282 + ], + [ + 433, + 260 + ] + ], + [ + [ + 487, + 216 + ], + [ + 509, + 239 + ], + [ + 475, + 231 + ], + [ + 500, + 200 + ], + [ + 455, + 210 + ], + [ + 478, + 178 + ] + ] + ] + }, + { + "task_id": "0003521", + "id": "RH20T_cfg1_task_0005_user_0008_scene_0006_cfg_0001_038522062547#185", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003521/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 480, + 259 + ], + [ + 483, + 236 + ], + [ + 507, + 250 + ], + [ + 453, + 268 + ], + [ + 503, + 272 + ], + [ + 451, + 289 + ] + ], + [ + [ + 429, + 231 + ], + [ + 453, + 228 + ], + [ + 409, + 216 + ], + [ + 448, + 245 + ], + [ + 388, + 218 + ], + [ + 426, + 246 + ] + ], + [ + [ + 456, + 244 + ], + [ + 464, + 243 + ], + [ + 462, + 273 + ], + [ + 450, + 215 + ], + [ + 455, + 272 + ], + [ + 444, + 217 + ] + ], + [ + [ + 428, + 264 + ], + [ + 401, + 268 + ], + [ + 426, + 239 + ], + [ + 431, + 291 + ], + [ + 454, + 234 + ], + [ + 460, + 287 + ] + ] + ] + }, + { + "task_id": "0003522", + "id": "RH20T_cfg4_task_0013_user_0010_scene_0002_cfg_0004_104122063550#104", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003522/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 307, + 108 + ], + [ + 298, + 114 + ], + [ + 329, + 107 + ], + [ + 284, + 110 + ], + [ + 341, + 100 + ], + [ + 293, + 103 + ] + ], + [ + [ + 302, + 79 + ], + [ + 280, + 82 + ], + [ + 311, + 93 + ], + [ + 293, + 65 + ], + [ + 333, + 90 + ], + [ + 314, + 62 + ] + ], + [ + [ + 290, + 113 + ], + [ + 311, + 108 + ], + [ + 285, + 88 + ], + [ + 295, + 137 + ], + [ + 263, + 94 + ], + [ + 273, + 144 + ] + ], + [ + [ + 302, + 99 + ], + [ + 301, + 89 + ], + [ + 300, + 121 + ], + [ + 305, + 77 + ], + [ + 301, + 129 + ], + [ + 307, + 87 + ] + ] + ] + }, + { + "task_id": "0003528", + "id": "RH20T_cfg7_task_0039_user_0014_scene_0008_cfg_0007_104122063633#106", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003528/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 209, + 127 + ], + [ + 207, + 111 + ], + [ + 211, + 113 + ], + [ + 206, + 143 + ], + [ + 212, + 129 + ], + [ + 208, + 158 + ] + ], + [ + [ + 225, + 131 + ], + [ + 244, + 123 + ], + [ + 234, + 146 + ], + [ + 216, + 118 + ], + [ + 214, + 155 + ], + [ + 197, + 127 + ] + ], + [ + [ + 226, + 122 + ], + [ + 245, + 132 + ], + [ + 227, + 123 + ], + [ + 225, + 121 + ], + [ + 207, + 114 + ], + [ + 203, + 110 + ] + ], + [ + [ + 223, + 141 + ], + [ + 205, + 154 + ], + [ + 228, + 150 + ], + [ + 217, + 132 + ], + [ + 245, + 138 + ], + [ + 236, + 118 + ] + ] + ] + }, + { + "task_id": "0003529", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0009_cfg_0006_104122061330#144", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003529/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 186 + ], + [ + 274, + 163 + ], + [ + 307, + 177 + ], + [ + 261, + 195 + ], + [ + 316, + 199 + ], + [ + 270, + 218 + ] + ], + [ + [ + 306, + 174 + ], + [ + 297, + 173 + ], + [ + 283, + 184 + ], + [ + 328, + 165 + ], + [ + 291, + 186 + ], + [ + 339, + 166 + ] + ], + [ + [ + 273, + 176 + ], + [ + 256, + 183 + ], + [ + 270, + 199 + ], + [ + 276, + 152 + ], + [ + 285, + 191 + ], + [ + 292, + 146 + ] + ], + [ + [ + 256, + 196 + ], + [ + 266, + 207 + ], + [ + 260, + 171 + ], + [ + 253, + 220 + ], + [ + 251, + 161 + ], + [ + 244, + 208 + ] + ] + ] + }, + { + "task_id": "0003530", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0009_cfg_0006_104122061330#346", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003530/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 331, + 204 + ], + [ + 331, + 226 + ], + [ + 316, + 217 + ], + [ + 347, + 190 + ], + [ + 317, + 198 + ], + [ + 347, + 170 + ] + ], + [ + [ + 326, + 201 + ], + [ + 334, + 228 + ], + [ + 353, + 194 + ], + [ + 300, + 208 + ], + [ + 346, + 167 + ], + [ + 293, + 182 + ] + ], + [ + [ + 329, + 210 + ], + [ + 324, + 187 + ], + [ + 316, + 227 + ], + [ + 340, + 194 + ], + [ + 321, + 249 + ], + [ + 343, + 216 + ] + ], + [ + [ + 329, + 228 + ], + [ + 319, + 219 + ], + [ + 324, + 205 + ], + [ + 335, + 253 + ], + [ + 333, + 214 + ], + [ + 344, + 260 + ] + ] + ] + }, + { + "task_id": "0003531", + "id": "RH20T_cfg4_task_0001_user_0010_scene_0004_cfg_0004_104122062823#168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003531/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 364, + 61 + ], + [ + 369, + 34 + ], + [ + 329, + 62 + ], + [ + 398, + 61 + ], + [ + 327, + 86 + ], + [ + 392, + 85 + ] + ], + [ + [ + 376, + 65 + ], + [ + 385, + 98 + ], + [ + 396, + 49 + ], + [ + 353, + 82 + ], + [ + 387, + 14 + ], + [ + 341, + 45 + ] + ], + [ + [ + 352, + 58 + ], + [ + 356, + 26 + ], + [ + 386, + 59 + ], + [ + 316, + 57 + ], + [ + 381, + 89 + ], + [ + 315, + 87 + ] + ], + [ + [ + 371, + 87 + ], + [ + 396, + 60 + ], + [ + 392, + 110 + ], + [ + 348, + 62 + ], + [ + 368, + 136 + ], + [ + 323, + 90 + ] + ] + ] + }, + { + "task_id": "0003534", + "id": "RH20T_cfg4_task_0016_user_0016_scene_0005_cfg_0004_104422070011#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003534/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 394, + 207 + ], + [ + 407, + 194 + ], + [ + 374, + 220 + ], + [ + 413, + 195 + ], + [ + 363, + 231 + ], + [ + 401, + 207 + ] + ], + [ + [ + 380, + 201 + ], + [ + 354, + 203 + ], + [ + 384, + 189 + ], + [ + 376, + 213 + ], + [ + 410, + 187 + ], + [ + 404, + 211 + ] + ], + [ + [ + 393, + 212 + ], + [ + 386, + 237 + ], + [ + 403, + 207 + ], + [ + 383, + 216 + ], + [ + 410, + 183 + ], + [ + 390, + 189 + ] + ], + [ + [ + 379, + 228 + ], + [ + 406, + 240 + ], + [ + 375, + 250 + ], + [ + 383, + 209 + ], + [ + 347, + 238 + ], + [ + 357, + 198 + ] + ] + ] + }, + { + "task_id": "0003536", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_039422060546#365", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003536/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 235, + 66 + ], + [ + 264, + 43 + ], + [ + 237, + 84 + ], + [ + 233, + 44 + ], + [ + 210, + 107 + ], + [ + 200, + 69 + ] + ], + [ + [ + 258, + 97 + ], + [ + 290, + 114 + ], + [ + 263, + 105 + ], + [ + 254, + 90 + ], + [ + 226, + 86 + ], + [ + 223, + 74 + ] + ], + [ + [ + 232, + 96 + ], + [ + 224, + 105 + ], + [ + 228, + 60 + ], + [ + 237, + 129 + ], + [ + 235, + 55 + ], + [ + 244, + 120 + ] + ], + [ + [ + 257, + 66 + ], + [ + 252, + 36 + ], + [ + 245, + 91 + ], + [ + 267, + 44 + ], + [ + 250, + 118 + ], + [ + 271, + 71 + ] + ] + ] + }, + { + "task_id": "0003537", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0004_cfg_0001_039422060546#568", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003537/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 238, + 74 + ], + [ + 234, + 43 + ], + [ + 230, + 105 + ], + [ + 245, + 46 + ], + [ + 234, + 131 + ], + [ + 248, + 73 + ] + ], + [ + [ + 253, + 67 + ], + [ + 257, + 29 + ], + [ + 235, + 75 + ], + [ + 274, + 57 + ], + [ + 231, + 112 + ], + [ + 270, + 99 + ] + ], + [ + [ + 194, + 57 + ], + [ + 152, + 54 + ], + [ + 194, + 12 + ], + [ + 194, + 99 + ], + [ + 237, + 16 + ], + [ + 235, + 103 + ] + ], + [ + [ + 214, + 108 + ], + [ + 174, + 113 + ], + [ + 222, + 137 + ], + [ + 207, + 82 + ], + [ + 261, + 132 + ], + [ + 243, + 78 + ] + ] + ] + }, + { + "task_id": "0003539", + "id": "RH20T_cfg6_task_0008_user_0014_scene_0007_cfg_0006_104122060811#159", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003539/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 384, + 141 + ], + [ + 369, + 140 + ], + [ + 387, + 143 + ], + [ + 380, + 140 + ], + [ + 401, + 144 + ], + [ + 396, + 141 + ] + ], + [ + [ + 389, + 136 + ], + [ + 392, + 128 + ], + [ + 403, + 132 + ], + [ + 376, + 140 + ], + [ + 400, + 139 + ], + [ + 374, + 147 + ] + ], + [ + [ + 392, + 127 + ], + [ + 378, + 131 + ], + [ + 389, + 115 + ], + [ + 395, + 138 + ], + [ + 403, + 111 + ], + [ + 410, + 135 + ] + ], + [ + [ + 378, + 129 + ], + [ + 378, + 122 + ], + [ + 378, + 116 + ], + [ + 378, + 142 + ], + [ + 377, + 123 + ], + [ + 377, + 150 + ] + ] + ] + }, + { + "task_id": "0003541", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_039422060546#261", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003541/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 303, + 182 + ], + [ + 310, + 149 + ], + [ + 335, + 187 + ], + [ + 269, + 176 + ], + [ + 328, + 218 + ], + [ + 263, + 208 + ] + ], + [ + [ + 287, + 170 + ], + [ + 319, + 177 + ], + [ + 294, + 138 + ], + [ + 280, + 202 + ], + [ + 263, + 132 + ], + [ + 250, + 195 + ] + ], + [ + [ + 287, + 190 + ], + [ + 280, + 169 + ], + [ + 298, + 214 + ], + [ + 276, + 167 + ], + [ + 303, + 232 + ], + [ + 282, + 186 + ] + ], + [ + [ + 277, + 206 + ], + [ + 245, + 214 + ], + [ + 270, + 173 + ], + [ + 284, + 240 + ], + [ + 302, + 165 + ], + [ + 317, + 232 + ] + ] + ] + }, + { + "task_id": "0003542", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_039422060546#379", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003542/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 241, + 127 + ], + [ + 266, + 142 + ], + [ + 236, + 107 + ], + [ + 246, + 145 + ], + [ + 212, + 92 + ], + [ + 224, + 130 + ] + ], + [ + [ + 283, + 126 + ], + [ + 283, + 125 + ], + [ + 253, + 123 + ], + [ + 312, + 129 + ], + [ + 251, + 125 + ], + [ + 314, + 131 + ] + ], + [ + [ + 263, + 134 + ], + [ + 256, + 114 + ], + [ + 262, + 159 + ], + [ + 265, + 111 + ], + [ + 269, + 176 + ], + [ + 272, + 130 + ] + ], + [ + [ + 270, + 134 + ], + [ + 268, + 167 + ], + [ + 244, + 127 + ], + [ + 293, + 140 + ], + [ + 246, + 92 + ], + [ + 295, + 109 + ] + ] + ] + }, + { + "task_id": "0003543", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_039422060546#561", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003543/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 230, + 200 + ], + [ + 221, + 179 + ], + [ + 251, + 217 + ], + [ + 209, + 184 + ], + [ + 258, + 235 + ], + [ + 218, + 202 + ] + ], + [ + [ + 208, + 204 + ], + [ + 190, + 226 + ], + [ + 192, + 183 + ], + [ + 224, + 227 + ], + [ + 210, + 160 + ], + [ + 243, + 203 + ] + ], + [ + [ + 268, + 217 + ], + [ + 299, + 226 + ], + [ + 277, + 188 + ], + [ + 259, + 245 + ], + [ + 245, + 178 + ], + [ + 228, + 237 + ] + ], + [ + [ + 211, + 192 + ], + [ + 239, + 176 + ], + [ + 227, + 219 + ], + [ + 195, + 165 + ], + [ + 199, + 236 + ], + [ + 165, + 182 + ] + ] + ] + }, + { + "task_id": "0003545", + "id": "RH20T_cfg5_task_0118_user_0007_scene_0010_cfg_0005_104122061850#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the water tap'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003545/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 399, + 184 + ], + [ + 423, + 179 + ], + [ + 404, + 212 + ], + [ + 393, + 157 + ], + [ + 381, + 216 + ], + [ + 371, + 163 + ] + ], + [ + [ + 422, + 139 + ], + [ + 417, + 156 + ], + [ + 440, + 157 + ], + [ + 403, + 120 + ], + [ + 443, + 141 + ], + [ + 408, + 105 + ] + ], + [ + [ + 410, + 157 + ], + [ + 407, + 140 + ], + [ + 387, + 173 + ], + [ + 434, + 140 + ], + [ + 391, + 187 + ], + [ + 436, + 156 + ] + ], + [ + [ + 396, + 129 + ], + [ + 397, + 157 + ], + [ + 371, + 132 + ], + [ + 423, + 125 + ], + [ + 370, + 105 + ], + [ + 422, + 95 + ] + ] + ] + }, + { + "task_id": "0003547", + "id": "RH20T_cfg6_task_0039_user_0014_scene_0005_cfg_0006_104122061018#86", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003547/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 102 + ], + [ + 379, + 79 + ], + [ + 357, + 78 + ], + [ + 386, + 127 + ], + [ + 351, + 99 + ], + [ + 378, + 146 + ] + ], + [ + [ + 349, + 87 + ], + [ + 317, + 95 + ], + [ + 362, + 118 + ], + [ + 336, + 59 + ], + [ + 394, + 110 + ], + [ + 367, + 52 + ] + ], + [ + [ + 351, + 85 + ], + [ + 371, + 65 + ], + [ + 327, + 70 + ], + [ + 377, + 101 + ], + [ + 306, + 90 + ], + [ + 356, + 122 + ] + ], + [ + [ + 385, + 66 + ], + [ + 389, + 81 + ], + [ + 407, + 86 + ], + [ + 364, + 47 + ], + [ + 404, + 69 + ], + [ + 358, + 28 + ] + ] + ] + }, + { + "task_id": "0003548", + "id": "RH20T_cfg6_task_0130_user_0014_scene_0004_cfg_0006_104122063633#226", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003548/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 182, + 207 + ], + [ + 198, + 223 + ], + [ + 192, + 191 + ], + [ + 171, + 223 + ], + [ + 175, + 175 + ], + [ + 153, + 206 + ] + ], + [ + [ + 167, + 205 + ], + [ + 175, + 181 + ], + [ + 170, + 199 + ], + [ + 164, + 211 + ], + [ + 164, + 221 + ], + [ + 157, + 236 + ] + ], + [ + [ + 194, + 193 + ], + [ + 192, + 178 + ], + [ + 219, + 193 + ], + [ + 167, + 193 + ], + [ + 223, + 209 + ], + [ + 168, + 209 + ] + ], + [ + [ + 161, + 217 + ], + [ + 170, + 199 + ], + [ + 137, + 205 + ], + [ + 184, + 229 + ], + [ + 129, + 223 + ], + [ + 175, + 245 + ] + ] + ] + }, + { + "task_id": "0003552", + "id": "RH20T_cfg4_task_0066_user_0015_scene_0007_cfg_0004_104122062823#385", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003552/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 255, + 16 + ], + [ + 250, + 0 + ], + [ + 275, + 24 + ], + [ + 235, + 9 + ], + [ + 278, + 46 + ], + [ + 240, + 31 + ] + ], + [ + [ + 240, + 0 + ], + [ + 261, + 12 + ], + [ + 243, + 0 + ], + [ + 237, + 1 + ], + [ + 220, + 0 + ], + [ + 217, + 0 + ] + ], + [ + [ + 269, + 31 + ], + [ + 262, + 7 + ], + [ + 275, + 23 + ], + [ + 263, + 39 + ], + [ + 282, + 49 + ], + [ + 270, + 62 + ] + ], + [ + [ + 272, + 33 + ], + [ + 275, + 7 + ], + [ + 297, + 37 + ], + [ + 246, + 29 + ], + [ + 293, + 63 + ], + [ + 244, + 55 + ] + ] + ] + }, + { + "task_id": "0003554", + "id": "RH20T_cfg4_task_0066_user_0015_scene_0007_cfg_0004_104122062823#829", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003554/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 130, + 65 + ], + [ + 152, + 48 + ], + [ + 118, + 79 + ], + [ + 144, + 50 + ], + [ + 95, + 97 + ], + [ + 120, + 69 + ] + ], + [ + [ + 171, + 86 + ], + [ + 144, + 97 + ], + [ + 183, + 116 + ], + [ + 158, + 56 + ], + [ + 206, + 105 + ], + [ + 184, + 48 + ] + ], + [ + [ + 144, + 39 + ], + [ + 158, + 16 + ], + [ + 164, + 33 + ], + [ + 121, + 47 + ], + [ + 151, + 53 + ], + [ + 107, + 70 + ] + ], + [ + [ + 150, + 74 + ], + [ + 138, + 46 + ], + [ + 173, + 82 + ], + [ + 128, + 66 + ], + [ + 182, + 109 + ], + [ + 139, + 92 + ] + ] + ] + }, + { + "task_id": "0003556", + "id": "RH20T_cfg4_task_0066_user_0015_scene_0007_cfg_0004_104122062823#1107", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003556/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 234, + 45 + ], + [ + 211, + 47 + ], + [ + 229, + 18 + ], + [ + 238, + 71 + ], + [ + 251, + 16 + ], + [ + 259, + 68 + ] + ], + [ + [ + 237, + 50 + ], + [ + 227, + 72 + ], + [ + 233, + 61 + ], + [ + 242, + 37 + ], + [ + 242, + 40 + ], + [ + 252, + 14 + ] + ], + [ + [ + 225, + 43 + ], + [ + 205, + 57 + ], + [ + 209, + 26 + ], + [ + 241, + 60 + ], + [ + 228, + 13 + ], + [ + 260, + 47 + ] + ], + [ + [ + 218, + 39 + ], + [ + 212, + 14 + ], + [ + 237, + 47 + ], + [ + 200, + 32 + ], + [ + 242, + 71 + ], + [ + 206, + 55 + ] + ] + ] + }, + { + "task_id": "0003557", + "id": "RH20T_cfg4_task_0066_user_0015_scene_0007_cfg_0004_104122062823#1223", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003557/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 243, + 37 + ], + [ + 238, + 12 + ], + [ + 265, + 46 + ], + [ + 223, + 27 + ], + [ + 269, + 69 + ], + [ + 228, + 50 + ] + ], + [ + [ + 256, + 56 + ], + [ + 271, + 64 + ], + [ + 280, + 46 + ], + [ + 231, + 66 + ], + [ + 266, + 40 + ], + [ + 218, + 58 + ] + ], + [ + [ + 272, + 39 + ], + [ + 261, + 14 + ], + [ + 247, + 47 + ], + [ + 297, + 31 + ], + [ + 259, + 73 + ], + [ + 308, + 56 + ] + ], + [ + [ + 229, + 34 + ], + [ + 206, + 48 + ], + [ + 215, + 10 + ], + [ + 243, + 58 + ], + [ + 238, + 0 + ], + [ + 266, + 45 + ] + ] + ] + }, + { + "task_id": "0003559", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122061602#214", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003559/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 411, + 207 + ], + [ + 400, + 205 + ], + [ + 432, + 188 + ], + [ + 391, + 226 + ], + [ + 441, + 191 + ], + [ + 402, + 227 + ] + ], + [ + [ + 446, + 204 + ], + [ + 461, + 178 + ], + [ + 470, + 219 + ], + [ + 421, + 188 + ], + [ + 455, + 245 + ], + [ + 406, + 215 + ] + ], + [ + [ + 438, + 186 + ], + [ + 458, + 187 + ], + [ + 443, + 214 + ], + [ + 432, + 158 + ], + [ + 425, + 211 + ], + [ + 414, + 158 + ] + ], + [ + [ + 449, + 174 + ], + [ + 465, + 149 + ], + [ + 471, + 190 + ], + [ + 430, + 160 + ], + [ + 455, + 214 + ], + [ + 416, + 183 + ] + ] + ] + }, + { + "task_id": "0003560", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122061602#337", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003560/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 425, + 218 + ], + [ + 408, + 193 + ], + [ + 433, + 212 + ], + [ + 418, + 223 + ], + [ + 452, + 239 + ], + [ + 434, + 247 + ] + ], + [ + [ + 458, + 181 + ], + [ + 448, + 204 + ], + [ + 482, + 180 + ], + [ + 433, + 182 + ], + [ + 493, + 156 + ], + [ + 442, + 157 + ] + ], + [ + [ + 416, + 204 + ], + [ + 395, + 214 + ], + [ + 430, + 191 + ], + [ + 404, + 215 + ], + [ + 451, + 182 + ], + [ + 424, + 205 + ] + ], + [ + [ + 441, + 183 + ], + [ + 461, + 202 + ], + [ + 433, + 204 + ], + [ + 450, + 161 + ], + [ + 415, + 186 + ], + [ + 430, + 143 + ] + ] + ] + }, + { + "task_id": "0003562", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122061602#687", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003562/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 440, + 151 + ], + [ + 437, + 138 + ], + [ + 470, + 154 + ], + [ + 410, + 149 + ], + [ + 471, + 166 + ], + [ + 414, + 161 + ] + ], + [ + [ + 470, + 184 + ], + [ + 441, + 195 + ], + [ + 467, + 163 + ], + [ + 472, + 203 + ], + [ + 496, + 151 + ], + [ + 499, + 192 + ] + ], + [ + [ + 438, + 177 + ], + [ + 455, + 170 + ], + [ + 450, + 203 + ], + [ + 426, + 151 + ], + [ + 433, + 208 + ], + [ + 411, + 158 + ] + ], + [ + [ + 456, + 199 + ], + [ + 427, + 191 + ], + [ + 459, + 175 + ], + [ + 453, + 225 + ], + [ + 488, + 183 + ], + [ + 484, + 234 + ] + ] + ] + }, + { + "task_id": "0003565", + "id": "RH20T_cfg4_task_0014_user_0016_scene_0003_cfg_0004_104422070011#108", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the object from the scale'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003565/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 149 + ], + [ + 306, + 122 + ], + [ + 323, + 151 + ], + [ + 304, + 147 + ], + [ + 330, + 177 + ], + [ + 313, + 177 + ] + ], + [ + [ + 343, + 153 + ], + [ + 321, + 145 + ], + [ + 330, + 147 + ], + [ + 359, + 159 + ], + [ + 350, + 154 + ], + [ + 381, + 166 + ] + ], + [ + [ + 300, + 158 + ], + [ + 303, + 132 + ], + [ + 295, + 152 + ], + [ + 304, + 164 + ], + [ + 292, + 180 + ], + [ + 302, + 189 + ] + ], + [ + [ + 318, + 161 + ], + [ + 328, + 149 + ], + [ + 292, + 162 + ], + [ + 343, + 159 + ], + [ + 284, + 173 + ], + [ + 333, + 170 + ] + ] + ] + }, + { + "task_id": "0003567", + "id": "RH20T_cfg4_task_0035_user_0017_scene_0005_cfg_0004_038522062288#110", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up one small block'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003567/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 278, + 195 + ], + [ + 276, + 181 + ], + [ + 249, + 192 + ], + [ + 308, + 198 + ], + [ + 252, + 205 + ], + [ + 308, + 211 + ] + ], + [ + [ + 306, + 194 + ], + [ + 314, + 165 + ], + [ + 297, + 199 + ], + [ + 313, + 189 + ], + [ + 289, + 228 + ], + [ + 306, + 215 + ] + ], + [ + [ + 259, + 182 + ], + [ + 234, + 165 + ], + [ + 268, + 165 + ], + [ + 249, + 200 + ], + [ + 290, + 181 + ], + [ + 273, + 217 + ] + ], + [ + [ + 260, + 207 + ], + [ + 232, + 222 + ], + [ + 260, + 207 + ], + [ + 260, + 208 + ], + [ + 289, + 192 + ], + [ + 285, + 194 + ] + ] + ] + }, + { + "task_id": "0003568", + "id": "RH20T_cfg5_task_0001_user_0014_scene_0002_cfg_0005_036422060215#216", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003568/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 230, + 141 + ], + [ + 244, + 149 + ], + [ + 218, + 159 + ], + [ + 241, + 123 + ], + [ + 202, + 151 + ], + [ + 226, + 114 + ] + ], + [ + [ + 235, + 130 + ], + [ + 220, + 114 + ], + [ + 252, + 113 + ], + [ + 218, + 146 + ], + [ + 267, + 129 + ], + [ + 234, + 162 + ] + ], + [ + [ + 256, + 143 + ], + [ + 245, + 130 + ], + [ + 267, + 152 + ], + [ + 244, + 133 + ], + [ + 277, + 165 + ], + [ + 255, + 147 + ] + ], + [ + [ + 243, + 140 + ], + [ + 246, + 122 + ], + [ + 263, + 135 + ], + [ + 223, + 145 + ], + [ + 259, + 151 + ], + [ + 221, + 161 + ] + ] + ] + }, + { + "task_id": "0003570", + "id": "RH20T_cfg5_task_0096_user_0007_scene_0008_cfg_0005_104122063678#474", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003570/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 339, + 276 + ], + [ + 330, + 293 + ], + [ + 320, + 252 + ], + [ + 358, + 299 + ], + [ + 329, + 238 + ], + [ + 364, + 282 + ] + ], + [ + [ + 366, + 227 + ], + [ + 358, + 253 + ], + [ + 392, + 243 + ], + [ + 339, + 211 + ], + [ + 398, + 218 + ], + [ + 347, + 188 + ] + ], + [ + [ + 348, + 290 + ], + [ + 375, + 303 + ], + [ + 344, + 319 + ], + [ + 352, + 260 + ], + [ + 320, + 305 + ], + [ + 327, + 249 + ] + ], + [ + [ + 371, + 260 + ], + [ + 379, + 242 + ], + [ + 341, + 261 + ], + [ + 401, + 258 + ], + [ + 335, + 277 + ], + [ + 393, + 273 + ] + ] + ] + }, + { + "task_id": "0003573", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0010_cfg_0005_104122061850#334", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003573/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 408, + 114 + ], + [ + 430, + 131 + ], + [ + 416, + 94 + ], + [ + 401, + 133 + ], + [ + 395, + 78 + ], + [ + 382, + 117 + ] + ], + [ + [ + 379, + 106 + ], + [ + 362, + 104 + ], + [ + 385, + 133 + ], + [ + 373, + 78 + ], + [ + 404, + 136 + ], + [ + 392, + 79 + ] + ], + [ + [ + 407, + 152 + ], + [ + 421, + 177 + ], + [ + 432, + 141 + ], + [ + 385, + 162 + ], + [ + 418, + 115 + ], + [ + 371, + 138 + ] + ], + [ + [ + 412, + 122 + ], + [ + 415, + 107 + ], + [ + 393, + 140 + ], + [ + 430, + 105 + ], + [ + 391, + 153 + ], + [ + 427, + 119 + ] + ] + ] + }, + { + "task_id": "0003578", + "id": "RH20T_cfg1_task_0053_user_0014_scene_0003_cfg_0001_750612070853#159", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a clamp'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003578/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 582, + 317 + ], + [ + 540, + 295 + ], + [ + 630, + 303 + ], + [ + 532, + 333 + ], + [ + 640, + 329 + ], + [ + 580, + 360 + ] + ], + [ + [ + 576, + 324 + ], + [ + 640, + 338 + ], + [ + 572, + 360 + ], + [ + 579, + 258 + ], + [ + 521, + 360 + ], + [ + 529, + 255 + ] + ], + [ + [ + 576, + 331 + ], + [ + 618, + 312 + ], + [ + 542, + 360 + ], + [ + 605, + 300 + ], + [ + 509, + 360 + ], + [ + 570, + 317 + ] + ], + [ + [ + 566, + 268 + ], + [ + 569, + 279 + ], + [ + 606, + 226 + ], + [ + 527, + 309 + ], + [ + 606, + 209 + ], + [ + 519, + 301 + ] + ] + ] + }, + { + "task_id": "0003585", + "id": "RH20T_cfg5_task_0085_user_0007_scene_0007_cfg_0005_036422060215#228", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the bubble ring from the assembled bubble ring and ball'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003585/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 262, + 151 + ], + [ + 246, + 158 + ], + [ + 274, + 147 + ], + [ + 252, + 155 + ], + [ + 290, + 140 + ], + [ + 268, + 148 + ] + ], + [ + [ + 281, + 174 + ], + [ + 269, + 181 + ], + [ + 293, + 189 + ], + [ + 268, + 158 + ], + [ + 306, + 182 + ], + [ + 280, + 150 + ] + ], + [ + [ + 269, + 185 + ], + [ + 286, + 178 + ], + [ + 280, + 184 + ], + [ + 259, + 186 + ], + [ + 262, + 192 + ], + [ + 242, + 194 + ] + ], + [ + [ + 265, + 166 + ], + [ + 268, + 151 + ], + [ + 278, + 179 + ], + [ + 253, + 154 + ], + [ + 274, + 194 + ], + [ + 250, + 169 + ] + ] + ] + }, + { + "task_id": "0003590", + "id": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_037522061512#276", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003590/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 176, + 197 + ], + [ + 154, + 191 + ], + [ + 184, + 212 + ], + [ + 168, + 183 + ], + [ + 204, + 216 + ], + [ + 188, + 189 + ] + ], + [ + [ + 203, + 198 + ], + [ + 198, + 220 + ], + [ + 228, + 203 + ], + [ + 178, + 193 + ], + [ + 232, + 182 + ], + [ + 183, + 172 + ] + ], + [ + [ + 178, + 213 + ], + [ + 179, + 191 + ], + [ + 194, + 202 + ], + [ + 161, + 224 + ], + [ + 193, + 222 + ], + [ + 161, + 245 + ] + ], + [ + [ + 181, + 202 + ], + [ + 166, + 203 + ], + [ + 169, + 225 + ], + [ + 191, + 181 + ], + [ + 185, + 224 + ], + [ + 207, + 179 + ] + ] + ] + }, + { + "task_id": "0003591", + "id": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_037522061512#410", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003591/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 182, + 150 + ], + [ + 159, + 142 + ], + [ + 194, + 143 + ], + [ + 169, + 159 + ], + [ + 216, + 152 + ], + [ + 194, + 168 + ] + ], + [ + [ + 178, + 159 + ], + [ + 185, + 136 + ], + [ + 192, + 160 + ], + [ + 163, + 157 + ], + [ + 185, + 182 + ], + [ + 155, + 181 + ] + ], + [ + [ + 196, + 154 + ], + [ + 193, + 131 + ], + [ + 200, + 156 + ], + [ + 192, + 153 + ], + [ + 202, + 178 + ], + [ + 194, + 177 + ] + ], + [ + [ + 172, + 177 + ], + [ + 183, + 190 + ], + [ + 149, + 188 + ], + [ + 195, + 166 + ], + [ + 139, + 175 + ], + [ + 184, + 154 + ] + ] + ] + }, + { + "task_id": "0003593", + "id": "RH20T_cfg1_task_0110_user_0015_scene_0003_cfg_0001_038522063145#97", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the pencil out from the pencil sharpener'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003593/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 392, + 113 + ], + [ + 398, + 95 + ], + [ + 409, + 126 + ], + [ + 376, + 102 + ], + [ + 403, + 142 + ], + [ + 372, + 118 + ] + ], + [ + [ + 395, + 120 + ], + [ + 394, + 132 + ], + [ + 414, + 112 + ], + [ + 376, + 128 + ], + [ + 417, + 99 + ], + [ + 377, + 115 + ] + ], + [ + [ + 393, + 107 + ], + [ + 407, + 101 + ], + [ + 377, + 105 + ], + [ + 410, + 109 + ], + [ + 363, + 111 + ], + [ + 396, + 116 + ] + ], + [ + [ + 398, + 121 + ], + [ + 385, + 138 + ], + [ + 410, + 136 + ], + [ + 386, + 105 + ], + [ + 422, + 120 + ], + [ + 400, + 88 + ] + ] + ] + }, + { + "task_id": "0003594", + "id": "RH20T_cfg1_task_0046_user_0001_scene_0003_cfg_0001_035622060973#209", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003594/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 151, + 133 + ], + [ + 147, + 123 + ], + [ + 131, + 142 + ], + [ + 170, + 124 + ], + [ + 134, + 151 + ], + [ + 172, + 133 + ] + ], + [ + [ + 148, + 143 + ], + [ + 138, + 149 + ], + [ + 130, + 130 + ], + [ + 164, + 155 + ], + [ + 140, + 123 + ], + [ + 174, + 149 + ] + ], + [ + [ + 134, + 126 + ], + [ + 120, + 126 + ], + [ + 151, + 133 + ], + [ + 118, + 118 + ], + [ + 164, + 132 + ], + [ + 132, + 118 + ] + ], + [ + [ + 135, + 132 + ], + [ + 119, + 141 + ], + [ + 125, + 113 + ], + [ + 145, + 151 + ], + [ + 141, + 104 + ], + [ + 162, + 142 + ] + ] + ] + }, + { + "task_id": "0003597", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0002_cfg_0006_104122064161#337", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003597/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 383, + 217 + ], + [ + 391, + 193 + ], + [ + 361, + 210 + ], + [ + 406, + 225 + ], + [ + 353, + 233 + ], + [ + 398, + 251 + ] + ], + [ + [ + 349, + 221 + ], + [ + 354, + 198 + ], + [ + 331, + 227 + ], + [ + 366, + 216 + ], + [ + 328, + 250 + ], + [ + 362, + 237 + ] + ], + [ + [ + 377, + 241 + ], + [ + 404, + 240 + ], + [ + 378, + 247 + ], + [ + 376, + 234 + ], + [ + 353, + 248 + ], + [ + 349, + 235 + ] + ], + [ + [ + 359, + 236 + ], + [ + 363, + 229 + ], + [ + 358, + 262 + ], + [ + 359, + 212 + ], + [ + 355, + 268 + ], + [ + 355, + 219 + ] + ] + ] + }, + { + "task_id": "0003598", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_104422070044#94", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003598/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 370, + 171 + ], + [ + 383, + 157 + ], + [ + 360, + 167 + ], + [ + 382, + 175 + ], + [ + 347, + 181 + ], + [ + 369, + 190 + ] + ], + [ + [ + 391, + 178 + ], + [ + 401, + 164 + ], + [ + 408, + 186 + ], + [ + 376, + 170 + ], + [ + 398, + 200 + ], + [ + 366, + 183 + ] + ], + [ + [ + 390, + 177 + ], + [ + 388, + 165 + ], + [ + 404, + 187 + ], + [ + 376, + 168 + ], + [ + 405, + 199 + ], + [ + 378, + 180 + ] + ], + [ + [ + 404, + 179 + ], + [ + 388, + 189 + ], + [ + 407, + 193 + ], + [ + 400, + 166 + ], + [ + 424, + 182 + ], + [ + 416, + 156 + ] + ] + ] + }, + { + "task_id": "0003599", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_104422070044#226", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003599/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 405, + 171 + ], + [ + 394, + 164 + ], + [ + 417, + 178 + ], + [ + 393, + 163 + ], + [ + 429, + 186 + ], + [ + 405, + 170 + ] + ], + [ + [ + 396, + 193 + ], + [ + 382, + 200 + ], + [ + 408, + 192 + ], + [ + 385, + 194 + ], + [ + 421, + 185 + ], + [ + 398, + 187 + ] + ], + [ + [ + 384, + 183 + ], + [ + 382, + 171 + ], + [ + 395, + 194 + ], + [ + 373, + 172 + ], + [ + 397, + 205 + ], + [ + 375, + 183 + ] + ], + [ + [ + 392, + 205 + ], + [ + 397, + 214 + ], + [ + 382, + 220 + ], + [ + 401, + 190 + ], + [ + 376, + 211 + ], + [ + 396, + 180 + ] + ] + ] + }, + { + "task_id": "0003600", + "id": "RH20T_cfg1_task_0033_user_0014_scene_0006_cfg_0001_038522062547#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003600/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 407, + 249 + ], + [ + 406, + 259 + ], + [ + 379, + 241 + ], + [ + 436, + 256 + ], + [ + 382, + 233 + ], + [ + 435, + 246 + ] + ], + [ + [ + 379, + 238 + ], + [ + 384, + 233 + ], + [ + 384, + 266 + ], + [ + 373, + 211 + ], + [ + 379, + 270 + ], + [ + 369, + 218 + ] + ], + [ + [ + 395, + 210 + ], + [ + 367, + 212 + ], + [ + 400, + 199 + ], + [ + 391, + 219 + ], + [ + 429, + 196 + ], + [ + 417, + 217 + ] + ], + [ + [ + 381, + 224 + ], + [ + 377, + 203 + ], + [ + 385, + 208 + ], + [ + 377, + 242 + ], + [ + 389, + 227 + ], + [ + 382, + 262 + ] + ] + ] + }, + { + "task_id": "0003602", + "id": "RH20T_cfg3_task_0023_user_0010_scene_0001_cfg_0003_104122062823#228", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003602/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 292, + 100 + ], + [ + 293, + 72 + ], + [ + 323, + 104 + ], + [ + 261, + 97 + ], + [ + 321, + 130 + ], + [ + 262, + 124 + ] + ], + [ + [ + 307, + 96 + ], + [ + 298, + 81 + ], + [ + 277, + 99 + ], + [ + 336, + 93 + ], + [ + 284, + 115 + ], + [ + 346, + 109 + ] + ], + [ + [ + 313, + 92 + ], + [ + 329, + 94 + ], + [ + 294, + 75 + ], + [ + 334, + 109 + ], + [ + 276, + 72 + ], + [ + 316, + 108 + ] + ], + [ + [ + 275, + 123 + ], + [ + 305, + 124 + ], + [ + 276, + 92 + ], + [ + 273, + 153 + ], + [ + 245, + 90 + ], + [ + 243, + 152 + ] + ] + ] + }, + { + "task_id": "0003604", + "id": "RH20T_cfg5_task_0045_user_0010_scene_0004_cfg_0005_105422061350#392", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the brush on the pen rack'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003604/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 208, + 136 + ], + [ + 188, + 156 + ], + [ + 205, + 124 + ], + [ + 212, + 151 + ], + [ + 225, + 105 + ], + [ + 233, + 130 + ] + ], + [ + [ + 209, + 144 + ], + [ + 234, + 134 + ], + [ + 204, + 129 + ], + [ + 214, + 158 + ], + [ + 177, + 141 + ], + [ + 190, + 169 + ] + ], + [ + [ + 187, + 122 + ], + [ + 164, + 136 + ], + [ + 199, + 124 + ], + [ + 176, + 120 + ], + [ + 222, + 110 + ], + [ + 198, + 108 + ] + ], + [ + [ + 211, + 120 + ], + [ + 208, + 103 + ], + [ + 222, + 100 + ], + [ + 200, + 142 + ], + [ + 225, + 117 + ], + [ + 204, + 157 + ] + ] + ] + }, + { + "task_id": "0003605", + "id": "RH20T_cfg2_task_0016_user_0003_scene_0009_cfg_0002_036422060215#156", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003605/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 323, + 254 + ], + [ + 329, + 239 + ], + [ + 334, + 270 + ], + [ + 313, + 238 + ], + [ + 328, + 283 + ], + [ + 308, + 252 + ] + ], + [ + [ + 312, + 258 + ], + [ + 331, + 251 + ], + [ + 311, + 242 + ], + [ + 314, + 273 + ], + [ + 291, + 249 + ], + [ + 295, + 280 + ] + ], + [ + [ + 321, + 249 + ], + [ + 339, + 248 + ], + [ + 321, + 270 + ], + [ + 322, + 229 + ], + [ + 303, + 271 + ], + [ + 303, + 230 + ] + ], + [ + [ + 323, + 263 + ], + [ + 305, + 268 + ], + [ + 332, + 259 + ], + [ + 315, + 267 + ], + [ + 350, + 255 + ], + [ + 332, + 263 + ] + ] + ] + }, + { + "task_id": "0003606", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122061330#266", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003606/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 222, + 263 + ], + [ + 248, + 272 + ], + [ + 224, + 284 + ], + [ + 219, + 240 + ], + [ + 200, + 275 + ], + [ + 193, + 231 + ] + ], + [ + [ + 237, + 249 + ], + [ + 226, + 238 + ], + [ + 226, + 277 + ], + [ + 248, + 223 + ], + [ + 237, + 286 + ], + [ + 258, + 233 + ] + ], + [ + [ + 228, + 219 + ], + [ + 258, + 218 + ], + [ + 224, + 205 + ], + [ + 232, + 232 + ], + [ + 193, + 206 + ], + [ + 204, + 233 + ] + ], + [ + [ + 248, + 255 + ], + [ + 275, + 250 + ], + [ + 247, + 280 + ], + [ + 248, + 228 + ], + [ + 219, + 286 + ], + [ + 218, + 233 + ] + ] + ] + }, + { + "task_id": "0003607", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122061330#704", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003607/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 262, + 227 + ], + [ + 257, + 216 + ], + [ + 247, + 248 + ], + [ + 277, + 207 + ], + [ + 252, + 258 + ], + [ + 281, + 218 + ] + ], + [ + [ + 238, + 237 + ], + [ + 253, + 254 + ], + [ + 220, + 234 + ], + [ + 255, + 240 + ], + [ + 206, + 217 + ], + [ + 241, + 225 + ] + ], + [ + [ + 226, + 240 + ], + [ + 222, + 212 + ], + [ + 251, + 238 + ], + [ + 202, + 241 + ], + [ + 255, + 266 + ], + [ + 206, + 268 + ] + ], + [ + [ + 260, + 229 + ], + [ + 246, + 253 + ], + [ + 267, + 235 + ], + [ + 251, + 223 + ], + [ + 280, + 213 + ], + [ + 266, + 198 + ] + ] + ] + }, + { + "task_id": "0003608", + "id": "RH20T_cfg3_task_0077_user_0010_scene_0009_cfg_0003_104122062823#222", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003608/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 246, + 66 + ], + [ + 237, + 39 + ], + [ + 228, + 68 + ], + [ + 266, + 65 + ], + [ + 237, + 93 + ], + [ + 274, + 91 + ] + ], + [ + [ + 223, + 53 + ], + [ + 194, + 53 + ], + [ + 219, + 45 + ], + [ + 229, + 62 + ], + [ + 246, + 46 + ], + [ + 258, + 62 + ] + ], + [ + [ + 210, + 87 + ], + [ + 192, + 78 + ], + [ + 192, + 77 + ], + [ + 226, + 97 + ], + [ + 211, + 87 + ], + [ + 244, + 107 + ] + ], + [ + [ + 229, + 82 + ], + [ + 228, + 57 + ], + [ + 250, + 72 + ], + [ + 208, + 93 + ], + [ + 250, + 95 + ], + [ + 210, + 117 + ] + ] + ] + }, + { + "task_id": "0003609", + "id": "RH20T_cfg3_task_0077_user_0010_scene_0009_cfg_0003_104122062823#344", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) randomly one at a time'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003609/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 260, + 128 + ], + [ + 242, + 126 + ], + [ + 227, + 138 + ], + [ + 291, + 118 + ], + [ + 246, + 141 + ], + [ + 312, + 120 + ] + ], + [ + [ + 254, + 149 + ], + [ + 227, + 129 + ], + [ + 271, + 126 + ], + [ + 237, + 170 + ], + [ + 298, + 146 + ], + [ + 263, + 188 + ] + ], + [ + [ + 237, + 149 + ], + [ + 240, + 133 + ], + [ + 205, + 155 + ], + [ + 270, + 142 + ], + [ + 199, + 173 + ], + [ + 269, + 160 + ] + ], + [ + [ + 240, + 138 + ], + [ + 242, + 105 + ], + [ + 260, + 127 + ], + [ + 217, + 150 + ], + [ + 257, + 156 + ], + [ + 216, + 182 + ] + ] + ] + }, + { + "task_id": "0003611", + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_037522062165#109", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003611/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 260, + 224 + ], + [ + 239, + 220 + ], + [ + 250, + 244 + ], + [ + 271, + 204 + ], + [ + 270, + 247 + ], + [ + 291, + 208 + ] + ], + [ + [ + 252, + 187 + ], + [ + 265, + 204 + ], + [ + 274, + 185 + ], + [ + 229, + 189 + ], + [ + 262, + 169 + ], + [ + 218, + 172 + ] + ], + [ + [ + 244, + 216 + ], + [ + 244, + 203 + ], + [ + 222, + 204 + ], + [ + 267, + 228 + ], + [ + 224, + 216 + ], + [ + 266, + 238 + ] + ], + [ + [ + 244, + 215 + ], + [ + 221, + 221 + ], + [ + 235, + 230 + ], + [ + 252, + 203 + ], + [ + 260, + 224 + ], + [ + 274, + 197 + ] + ] + ] + }, + { + "task_id": "0003612", + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_037522062165#328", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003612/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 202, + 205 + ], + [ + 197, + 190 + ], + [ + 196, + 185 + ], + [ + 207, + 226 + ], + [ + 201, + 199 + ], + [ + 211, + 239 + ] + ], + [ + [ + 220, + 203 + ], + [ + 244, + 205 + ], + [ + 217, + 215 + ], + [ + 222, + 190 + ], + [ + 192, + 213 + ], + [ + 196, + 188 + ] + ], + [ + [ + 229, + 221 + ], + [ + 243, + 202 + ], + [ + 211, + 205 + ], + [ + 247, + 237 + ], + [ + 195, + 226 + ], + [ + 233, + 257 + ] + ], + [ + [ + 186, + 199 + ], + [ + 178, + 178 + ], + [ + 162, + 205 + ], + [ + 211, + 194 + ], + [ + 171, + 225 + ], + [ + 218, + 214 + ] + ] + ] + }, + { + "task_id": "0003613", + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_037522062165#496", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003613/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 167 + ], + [ + 299, + 161 + ], + [ + 301, + 153 + ], + [ + 267, + 182 + ], + [ + 287, + 159 + ], + [ + 254, + 187 + ] + ], + [ + [ + 259, + 179 + ], + [ + 284, + 178 + ], + [ + 258, + 153 + ], + [ + 260, + 205 + ], + [ + 233, + 154 + ], + [ + 235, + 205 + ] + ], + [ + [ + 303, + 207 + ], + [ + 287, + 217 + ], + [ + 313, + 230 + ], + [ + 293, + 183 + ], + [ + 327, + 220 + ], + [ + 309, + 174 + ] + ], + [ + [ + 275, + 185 + ], + [ + 274, + 169 + ], + [ + 273, + 164 + ], + [ + 278, + 207 + ], + [ + 274, + 179 + ], + [ + 279, + 221 + ] + ] + ] + }, + { + "task_id": "0003614", + "id": "RH20T_cfg2_task_0034_user_0003_scene_0006_cfg_0002_037522062165#740", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003614/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 126, + 235 + ], + [ + 131, + 255 + ], + [ + 140, + 218 + ], + [ + 110, + 254 + ], + [ + 136, + 197 + ], + [ + 103, + 232 + ] + ], + [ + [ + 132, + 245 + ], + [ + 130, + 245 + ], + [ + 155, + 225 + ], + [ + 107, + 266 + ], + [ + 155, + 226 + ], + [ + 110, + 265 + ] + ], + [ + [ + 149, + 226 + ], + [ + 128, + 208 + ], + [ + 159, + 215 + ], + [ + 140, + 235 + ], + [ + 180, + 233 + ], + [ + 160, + 252 + ] + ], + [ + [ + 149, + 231 + ], + [ + 142, + 218 + ], + [ + 164, + 209 + ], + [ + 133, + 254 + ], + [ + 169, + 222 + ], + [ + 140, + 265 + ] + ] + ] + }, + { + "task_id": "0003617", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0009_cfg_0004_038522062288#273", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003617/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 195, + 109 + ], + [ + 227, + 113 + ], + [ + 171, + 108 + ], + [ + 216, + 110 + ], + [ + 140, + 105 + ], + [ + 187, + 106 + ] + ], + [ + [ + 204, + 137 + ], + [ + 195, + 113 + ], + [ + 168, + 145 + ], + [ + 239, + 128 + ], + [ + 178, + 166 + ], + [ + 245, + 149 + ] + ], + [ + [ + 236, + 150 + ], + [ + 259, + 124 + ], + [ + 261, + 173 + ], + [ + 210, + 126 + ], + [ + 239, + 199 + ], + [ + 185, + 152 + ] + ], + [ + [ + 202, + 119 + ], + [ + 195, + 146 + ], + [ + 172, + 102 + ], + [ + 229, + 136 + ], + [ + 180, + 77 + ], + [ + 235, + 111 + ] + ] + ] + }, + { + "task_id": "0003618", + "id": "RH20T_cfg2_task_0051_user_0016_scene_0003_cfg_0002_037522062165#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the dish off the dish rack'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003618/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 341, + 282 + ], + [ + 364, + 264 + ], + [ + 325, + 253 + ], + [ + 356, + 310 + ], + [ + 299, + 271 + ], + [ + 332, + 330 + ] + ], + [ + [ + 408, + 285 + ], + [ + 431, + 285 + ], + [ + 389, + 307 + ], + [ + 429, + 261 + ], + [ + 364, + 308 + ], + [ + 405, + 260 + ] + ], + [ + [ + 378, + 280 + ], + [ + 387, + 262 + ], + [ + 358, + 255 + ], + [ + 400, + 307 + ], + [ + 351, + 273 + ], + [ + 390, + 323 + ] + ], + [ + [ + 387, + 264 + ], + [ + 408, + 239 + ], + [ + 412, + 287 + ], + [ + 361, + 242 + ], + [ + 391, + 314 + ], + [ + 339, + 268 + ] + ] + ] + }, + { + "task_id": "0003619", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112071037#125", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003619/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 158, + 19 + ], + [ + 40, + 0 + ], + [ + 116, + 120 + ], + [ + 200, + 0 + ], + [ + 206, + 153 + ], + [ + 278, + 0 + ] + ], + [ + [ + 57, + 148 + ], + [ + 29, + 12 + ], + [ + 0, + 155 + ], + [ + 191, + 142 + ], + [ + 0, + 268 + ], + [ + 199, + 243 + ] + ], + [ + [ + 0, + 189 + ], + [ + 112, + 146 + ], + [ + 0, + 105 + ], + [ + 31, + 293 + ], + [ + 0, + 143 + ], + [ + 0, + 360 + ] + ], + [ + [ + 166, + 143 + ], + [ + 199, + 203 + ], + [ + 109, + 250 + ], + [ + 210, + 62 + ], + [ + 29, + 175 + ], + [ + 177, + 0 + ] + ] + ] + }, + { + "task_id": "0003621", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112071037#295", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003621/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 0, + 110 + ], + [ + 0, + 121 + ], + [ + 0, + 284 + ], + [ + 38, + 0 + ], + [ + 0, + 332 + ], + [ + 95, + 0 + ] + ], + [ + [ + 105, + 167 + ], + [ + 0, + 122 + ], + [ + 81, + 182 + ], + [ + 140, + 145 + ], + [ + 170, + 210 + ], + [ + 255, + 189 + ] + ], + [ + [ + 48, + 172 + ], + [ + 19, + 36 + ], + [ + 0, + 181 + ], + [ + 185, + 163 + ], + [ + 0, + 295 + ], + [ + 192, + 263 + ] + ], + [ + [ + 49, + 267 + ], + [ + 132, + 339 + ], + [ + 143, + 167 + ], + [ + 0, + 360 + ], + [ + 61, + 61 + ], + [ + 0, + 319 + ] + ] + ] + }, + { + "task_id": "0003623", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112071037#476", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003623/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 75, + 12 + ], + [ + 11, + 25 + ], + [ + 163, + 101 + ], + [ + 0, + 0 + ], + [ + 185, + 78 + ], + [ + 34, + 0 + ] + ], + [ + [ + 55, + 160 + ], + [ + 12, + 19 + ], + [ + 0, + 176 + ], + [ + 194, + 147 + ], + [ + 0, + 288 + ], + [ + 209, + 251 + ] + ], + [ + [ + 0, + 139 + ], + [ + 0, + 14 + ], + [ + 111, + 55 + ], + [ + 0, + 231 + ], + [ + 200, + 168 + ], + [ + 0, + 348 + ] + ], + [ + [ + 0, + 141 + ], + [ + 0, + 0 + ], + [ + 95, + 133 + ], + [ + 0, + 152 + ], + [ + 96, + 212 + ], + [ + 0, + 255 + ] + ] + ] + }, + { + "task_id": "0003624", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_819112071037#578", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003624/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 128, + 259 + ], + [ + 191, + 187 + ], + [ + 34, + 225 + ], + [ + 232, + 297 + ], + [ + 0, + 316 + ], + [ + 159, + 360 + ] + ], + [ + [ + 73, + 170 + ], + [ + 48, + 30 + ], + [ + 0, + 178 + ], + [ + 212, + 164 + ], + [ + 0, + 294 + ], + [ + 215, + 264 + ] + ], + [ + [ + 112, + 41 + ], + [ + 46, + 0 + ], + [ + 0, + 88 + ], + [ + 259, + 2 + ], + [ + 16, + 238 + ], + [ + 302, + 138 + ] + ], + [ + [ + 204, + 234 + ], + [ + 333, + 222 + ], + [ + 201, + 360 + ], + [ + 208, + 80 + ], + [ + 30, + 360 + ], + [ + 18, + 64 + ] + ] + ] + }, + { + "task_id": "0003625", + "id": "RH20T_cfg1_task_0041_user_0001_scene_0006_cfg_0001_750612070851#145", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the tabletop with a sponge'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003625/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 352, + 262 + ], + [ + 326, + 283 + ], + [ + 334, + 235 + ], + [ + 370, + 288 + ], + [ + 360, + 215 + ], + [ + 394, + 267 + ] + ], + [ + [ + 305, + 229 + ], + [ + 306, + 267 + ], + [ + 318, + 224 + ], + [ + 294, + 234 + ], + [ + 316, + 185 + ], + [ + 294, + 200 + ] + ], + [ + [ + 340, + 248 + ], + [ + 343, + 228 + ], + [ + 305, + 248 + ], + [ + 376, + 249 + ], + [ + 304, + 267 + ], + [ + 372, + 267 + ] + ], + [ + [ + 361, + 271 + ], + [ + 374, + 247 + ], + [ + 351, + 291 + ], + [ + 371, + 252 + ], + [ + 339, + 314 + ], + [ + 360, + 273 + ] + ] + ] + }, + { + "task_id": "0003626", + "id": "RH20T_cfg7_task_0050_user_0020_scene_0008_cfg_0007_037522061512#135", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the dish on the dish rack'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003626/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 245, + 149 + ], + [ + 242, + 171 + ], + [ + 232, + 149 + ], + [ + 260, + 149 + ], + [ + 235, + 128 + ], + [ + 264, + 127 + ] + ], + [ + [ + 247, + 111 + ], + [ + 226, + 102 + ], + [ + 248, + 120 + ], + [ + 247, + 101 + ], + [ + 268, + 128 + ], + [ + 270, + 110 + ] + ], + [ + [ + 237, + 140 + ], + [ + 232, + 128 + ], + [ + 240, + 121 + ], + [ + 233, + 160 + ], + [ + 245, + 132 + ], + [ + 238, + 170 + ] + ], + [ + [ + 240, + 125 + ], + [ + 224, + 142 + ], + [ + 223, + 109 + ], + [ + 257, + 141 + ], + [ + 240, + 93 + ], + [ + 273, + 124 + ] + ] + ] + }, + { + "task_id": "0003628", + "id": "RH20T_cfg1_task_0096_user_0015_scene_0005_cfg_0001_039422060546#186", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003628/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 101 + ], + [ + 428, + 104 + ], + [ + 389, + 136 + ], + [ + 397, + 66 + ], + [ + 355, + 133 + ], + [ + 362, + 63 + ] + ], + [ + [ + 340, + 114 + ], + [ + 321, + 82 + ], + [ + 372, + 94 + ], + [ + 309, + 133 + ], + [ + 390, + 126 + ], + [ + 328, + 163 + ] + ], + [ + [ + 361, + 76 + ], + [ + 363, + 46 + ], + [ + 360, + 102 + ], + [ + 362, + 52 + ], + [ + 358, + 128 + ], + [ + 361, + 78 + ] + ], + [ + [ + 326, + 110 + ], + [ + 296, + 88 + ], + [ + 337, + 90 + ], + [ + 313, + 132 + ], + [ + 364, + 110 + ], + [ + 343, + 154 + ] + ] + ] + }, + { + "task_id": "0003632", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_104422070011#312", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003632/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 323, + 224 + ], + [ + 306, + 202 + ], + [ + 344, + 206 + ], + [ + 303, + 242 + ], + [ + 362, + 228 + ], + [ + 320, + 265 + ] + ], + [ + [ + 345, + 228 + ], + [ + 337, + 241 + ], + [ + 318, + 223 + ], + [ + 371, + 232 + ], + [ + 326, + 208 + ], + [ + 381, + 218 + ] + ], + [ + [ + 332, + 248 + ], + [ + 307, + 231 + ], + [ + 328, + 265 + ], + [ + 337, + 229 + ], + [ + 353, + 281 + ], + [ + 365, + 247 + ] + ], + [ + [ + 333, + 226 + ], + [ + 339, + 211 + ], + [ + 310, + 239 + ], + [ + 355, + 215 + ], + [ + 306, + 252 + ], + [ + 349, + 228 + ] + ] + ] + }, + { + "task_id": "0003633", + "id": "RH20T_cfg4_task_0103_user_0007_scene_0003_cfg_0004_104122063550#50", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003633/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 270, + 241 + ], + [ + 268, + 232 + ], + [ + 277, + 268 + ], + [ + 262, + 215 + ], + [ + 278, + 274 + ], + [ + 264, + 224 + ] + ], + [ + [ + 295, + 232 + ], + [ + 285, + 256 + ], + [ + 271, + 219 + ], + [ + 319, + 246 + ], + [ + 280, + 195 + ], + [ + 329, + 221 + ] + ], + [ + [ + 285, + 239 + ], + [ + 259, + 247 + ], + [ + 288, + 239 + ], + [ + 282, + 239 + ], + [ + 315, + 230 + ], + [ + 307, + 232 + ] + ], + [ + [ + 242, + 237 + ], + [ + 216, + 228 + ], + [ + 247, + 235 + ], + [ + 237, + 238 + ], + [ + 271, + 244 + ], + [ + 264, + 248 + ] + ] + ] + }, + { + "task_id": "0003636", + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122061018#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003636/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 273, + 101 + ], + [ + 291, + 125 + ], + [ + 265, + 106 + ], + [ + 283, + 95 + ], + [ + 248, + 83 + ], + [ + 264, + 69 + ] + ], + [ + [ + 278, + 79 + ], + [ + 270, + 48 + ], + [ + 266, + 75 + ], + [ + 291, + 83 + ], + [ + 273, + 102 + ], + [ + 298, + 113 + ] + ], + [ + [ + 269, + 88 + ], + [ + 284, + 60 + ], + [ + 259, + 84 + ], + [ + 280, + 93 + ], + [ + 244, + 109 + ], + [ + 264, + 122 + ] + ], + [ + [ + 302, + 91 + ], + [ + 284, + 116 + ], + [ + 323, + 106 + ], + [ + 283, + 77 + ], + [ + 343, + 80 + ], + [ + 301, + 53 + ] + ] + ] + }, + { + "task_id": "0003638", + "id": "RH20T_cfg4_task_0009_user_0010_scene_0004_cfg_0004_104422070011#98", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003638/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 346, + 176 + ], + [ + 376, + 172 + ], + [ + 339, + 154 + ], + [ + 353, + 195 + ], + [ + 309, + 158 + ], + [ + 325, + 199 + ] + ], + [ + [ + 312, + 207 + ], + [ + 295, + 220 + ], + [ + 323, + 193 + ], + [ + 301, + 223 + ], + [ + 341, + 180 + ], + [ + 321, + 209 + ] + ], + [ + [ + 327, + 168 + ], + [ + 309, + 153 + ], + [ + 343, + 145 + ], + [ + 311, + 191 + ], + [ + 359, + 160 + ], + [ + 328, + 203 + ] + ], + [ + [ + 315, + 170 + ], + [ + 325, + 157 + ], + [ + 292, + 182 + ], + [ + 336, + 158 + ], + [ + 284, + 193 + ], + [ + 326, + 169 + ] + ] + ] + }, + { + "task_id": "0003640", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0009_cfg_0006_104122063633#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003640/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 225, + 169 + ], + [ + 246, + 173 + ], + [ + 222, + 162 + ], + [ + 228, + 175 + ], + [ + 199, + 158 + ], + [ + 208, + 171 + ] + ], + [ + [ + 257, + 178 + ], + [ + 279, + 174 + ], + [ + 262, + 200 + ], + [ + 252, + 155 + ], + [ + 239, + 204 + ], + [ + 230, + 158 + ] + ], + [ + [ + 242, + 182 + ], + [ + 247, + 161 + ], + [ + 243, + 186 + ], + [ + 241, + 179 + ], + [ + 239, + 207 + ], + [ + 236, + 202 + ] + ], + [ + [ + 234, + 195 + ], + [ + 240, + 173 + ], + [ + 214, + 188 + ], + [ + 255, + 202 + ], + [ + 208, + 210 + ], + [ + 249, + 224 + ] + ] + ] + }, + { + "task_id": "0003642", + "id": "RH20T_cfg3_task_0067_user_0010_scene_0010_cfg_0003_104122062823#180", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button from top to bottom with obstacles'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003642/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 0, + 98 + ], + [ + 0, + 116 + ], + [ + 0, + 56 + ], + [ + 23, + 137 + ], + [ + 18, + 40 + ], + [ + 66, + 119 + ] + ], + [ + [ + 43, + 145 + ], + [ + 79, + 152 + ], + [ + 61, + 114 + ], + [ + 22, + 179 + ], + [ + 27, + 107 + ], + [ + 0, + 171 + ] + ], + [ + [ + 15, + 124 + ], + [ + 0, + 88 + ], + [ + 27, + 144 + ], + [ + 4, + 106 + ], + [ + 45, + 176 + ], + [ + 21, + 137 + ] + ], + [ + [ + 49, + 89 + ], + [ + 24, + 51 + ], + [ + 25, + 81 + ], + [ + 68, + 95 + ], + [ + 53, + 122 + ], + [ + 90, + 129 + ] + ] + ] + }, + { + "task_id": "0003644", + "id": "RH20T_cfg2_task_0066_user_0014_scene_0006_cfg_0002_104422070044#336", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003644/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 463, + 233 + ], + [ + 445, + 236 + ], + [ + 468, + 253 + ], + [ + 458, + 213 + ], + [ + 487, + 251 + ], + [ + 476, + 209 + ] + ], + [ + [ + 435, + 227 + ], + [ + 420, + 215 + ], + [ + 448, + 213 + ], + [ + 422, + 242 + ], + [ + 463, + 226 + ], + [ + 437, + 255 + ] + ], + [ + [ + 446, + 223 + ], + [ + 447, + 210 + ], + [ + 466, + 227 + ], + [ + 426, + 218 + ], + [ + 464, + 239 + ], + [ + 426, + 230 + ] + ], + [ + [ + 431, + 216 + ], + [ + 429, + 230 + ], + [ + 442, + 229 + ], + [ + 420, + 203 + ], + [ + 444, + 216 + ], + [ + 423, + 191 + ] + ] + ] + }, + { + "task_id": "0003645", + "id": "RH20T_cfg2_task_0066_user_0014_scene_0006_cfg_0002_104422070044#634", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003645/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 419, + 248 + ], + [ + 426, + 262 + ], + [ + 412, + 236 + ], + [ + 426, + 261 + ], + [ + 405, + 222 + ], + [ + 419, + 246 + ] + ], + [ + [ + 450, + 258 + ], + [ + 458, + 244 + ], + [ + 440, + 268 + ], + [ + 460, + 249 + ], + [ + 433, + 281 + ], + [ + 453, + 262 + ] + ], + [ + [ + 434, + 249 + ], + [ + 435, + 237 + ], + [ + 454, + 254 + ], + [ + 414, + 244 + ], + [ + 453, + 265 + ], + [ + 414, + 256 + ] + ], + [ + [ + 433, + 250 + ], + [ + 448, + 261 + ], + [ + 440, + 234 + ], + [ + 427, + 266 + ], + [ + 427, + 224 + ], + [ + 414, + 256 + ] + ] + ] + }, + { + "task_id": "0003646", + "id": "RH20T_cfg2_task_0066_user_0014_scene_0006_cfg_0002_104422070044#799", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003646/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 412, + 263 + ], + [ + 415, + 249 + ], + [ + 429, + 275 + ], + [ + 396, + 251 + ], + [ + 425, + 288 + ], + [ + 394, + 264 + ] + ], + [ + [ + 415, + 265 + ], + [ + 432, + 274 + ], + [ + 411, + 255 + ], + [ + 420, + 277 + ], + [ + 395, + 247 + ], + [ + 403, + 268 + ] + ], + [ + [ + 410, + 265 + ], + [ + 392, + 256 + ], + [ + 411, + 271 + ], + [ + 409, + 259 + ], + [ + 429, + 279 + ], + [ + 429, + 269 + ] + ], + [ + [ + 413, + 260 + ], + [ + 413, + 249 + ], + [ + 434, + 265 + ], + [ + 393, + 256 + ], + [ + 434, + 276 + ], + [ + 395, + 267 + ] + ] + ] + }, + { + "task_id": "0003649", + "id": "RH20T_cfg4_task_0002_user_0010_scene_0007_cfg_0004_104122062295#134", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003649/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 436, + 254 + ], + [ + 400, + 286 + ], + [ + 402, + 216 + ], + [ + 472, + 293 + ], + [ + 436, + 187 + ], + [ + 505, + 260 + ] + ], + [ + [ + 389, + 315 + ], + [ + 382, + 284 + ], + [ + 352, + 344 + ], + [ + 427, + 285 + ], + [ + 357, + 360 + ], + [ + 441, + 320 + ] + ], + [ + [ + 346, + 256 + ], + [ + 394, + 233 + ], + [ + 366, + 277 + ], + [ + 323, + 231 + ], + [ + 324, + 297 + ], + [ + 275, + 256 + ] + ], + [ + [ + 411, + 263 + ], + [ + 394, + 226 + ], + [ + 368, + 262 + ], + [ + 460, + 264 + ], + [ + 385, + 293 + ], + [ + 475, + 300 + ] + ] + ] + }, + { + "task_id": "0003650", + "id": "RH20T_cfg1_task_0076_user_0010_scene_0001_cfg_0001_039422060546#152", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) one on top of the other every time'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003650/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 331, + 297 + ], + [ + 301, + 315 + ], + [ + 340, + 319 + ], + [ + 324, + 278 + ], + [ + 372, + 300 + ], + [ + 352, + 260 + ] + ], + [ + [ + 339, + 286 + ], + [ + 337, + 268 + ], + [ + 329, + 320 + ], + [ + 348, + 255 + ], + [ + 332, + 335 + ], + [ + 350, + 272 + ] + ], + [ + [ + 333, + 312 + ], + [ + 352, + 303 + ], + [ + 312, + 338 + ], + [ + 352, + 289 + ], + [ + 296, + 345 + ], + [ + 336, + 298 + ] + ], + [ + [ + 344, + 290 + ], + [ + 316, + 269 + ], + [ + 331, + 307 + ], + [ + 358, + 272 + ], + [ + 358, + 329 + ], + [ + 388, + 295 + ] + ] + ] + }, + { + "task_id": "0003653", + "id": "RH20T_cfg3_task_0007_user_0016_scene_0006_cfg_0003_104122063550#244", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003653/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 364, + 243 + ], + [ + 365, + 235 + ], + [ + 376, + 268 + ], + [ + 354, + 220 + ], + [ + 375, + 274 + ], + [ + 354, + 229 + ] + ], + [ + [ + 354, + 234 + ], + [ + 359, + 259 + ], + [ + 331, + 241 + ], + [ + 379, + 226 + ], + [ + 328, + 218 + ], + [ + 374, + 203 + ] + ], + [ + [ + 364, + 260 + ], + [ + 341, + 272 + ], + [ + 370, + 282 + ], + [ + 357, + 237 + ], + [ + 391, + 270 + ], + [ + 380, + 225 + ] + ], + [ + [ + 345, + 254 + ], + [ + 327, + 259 + ], + [ + 323, + 244 + ], + [ + 368, + 264 + ], + [ + 339, + 240 + ], + [ + 383, + 259 + ] + ] + ] + }, + { + "task_id": "0003654", + "id": "RH20T_cfg2_task_0062_user_0007_scene_0001_cfg_0002_036422060215#247", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the toilet paper on its holder'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003654/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 238, + 133 + ], + [ + 252, + 143 + ], + [ + 233, + 149 + ], + [ + 244, + 115 + ], + [ + 220, + 139 + ], + [ + 231, + 105 + ] + ], + [ + [ + 227, + 141 + ], + [ + 234, + 130 + ], + [ + 237, + 127 + ], + [ + 217, + 154 + ], + [ + 230, + 138 + ], + [ + 210, + 166 + ] + ], + [ + [ + 256, + 150 + ], + [ + 239, + 146 + ], + [ + 262, + 132 + ], + [ + 249, + 168 + ], + [ + 279, + 136 + ], + [ + 266, + 172 + ] + ], + [ + [ + 246, + 142 + ], + [ + 249, + 127 + ], + [ + 266, + 148 + ], + [ + 227, + 137 + ], + [ + 263, + 162 + ], + [ + 225, + 152 + ] + ] + ] + }, + { + "task_id": "0003660", + "id": "RH20T_cfg5_task_0044_user_0010_scene_0007_cfg_0005_036422060215#192", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003660/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 275, + 154 + ], + [ + 258, + 140 + ], + [ + 278, + 156 + ], + [ + 272, + 151 + ], + [ + 294, + 169 + ], + [ + 290, + 165 + ] + ], + [ + [ + 280, + 130 + ], + [ + 286, + 148 + ], + [ + 259, + 134 + ], + [ + 299, + 126 + ], + [ + 253, + 117 + ], + [ + 293, + 109 + ] + ], + [ + [ + 290, + 140 + ], + [ + 293, + 127 + ], + [ + 311, + 139 + ], + [ + 269, + 141 + ], + [ + 307, + 151 + ], + [ + 266, + 152 + ] + ], + [ + [ + 266, + 135 + ], + [ + 274, + 125 + ], + [ + 253, + 119 + ], + [ + 279, + 151 + ], + [ + 244, + 131 + ], + [ + 271, + 163 + ] + ] + ] + }, + { + "task_id": "0003664", + "id": "RH20T_cfg6_task_0022_user_0014_scene_0007_cfg_0006_104122061018#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Wave the flag'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003664/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 173 + ], + [ + 289, + 180 + ], + [ + 330, + 170 + ], + [ + 309, + 176 + ], + [ + 360, + 163 + ], + [ + 336, + 169 + ] + ], + [ + [ + 337, + 216 + ], + [ + 322, + 187 + ], + [ + 365, + 200 + ], + [ + 308, + 232 + ], + [ + 379, + 227 + ], + [ + 322, + 259 + ] + ], + [ + [ + 331, + 194 + ], + [ + 336, + 176 + ], + [ + 321, + 168 + ], + [ + 342, + 221 + ], + [ + 318, + 185 + ], + [ + 337, + 236 + ] + ], + [ + [ + 307, + 188 + ], + [ + 280, + 195 + ], + [ + 299, + 158 + ], + [ + 315, + 217 + ], + [ + 325, + 152 + ], + [ + 340, + 210 + ] + ] + ] + }, + { + "task_id": "0003665", + "id": "RH20T_cfg6_task_0022_user_0014_scene_0007_cfg_0006_104122061018#176", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Wave the flag'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003665/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 330, + 197 + ], + [ + 335, + 180 + ], + [ + 318, + 171 + ], + [ + 342, + 224 + ], + [ + 314, + 187 + ], + [ + 336, + 238 + ] + ], + [ + [ + 331, + 232 + ], + [ + 301, + 229 + ], + [ + 340, + 228 + ], + [ + 324, + 235 + ], + [ + 372, + 232 + ], + [ + 352, + 238 + ] + ], + [ + [ + 344, + 179 + ], + [ + 363, + 180 + ], + [ + 331, + 203 + ], + [ + 358, + 153 + ], + [ + 310, + 203 + ], + [ + 338, + 151 + ] + ], + [ + [ + 325, + 206 + ], + [ + 339, + 206 + ], + [ + 302, + 185 + ], + [ + 349, + 229 + ], + [ + 284, + 184 + ], + [ + 333, + 231 + ] + ] + ] + }, + { + "task_id": "0003673", + "id": "RH20T_cfg3_task_0006_user_0016_scene_0003_cfg_0003_104122062295#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003673/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 87, + 193 + ], + [ + 83, + 231 + ], + [ + 54, + 167 + ], + [ + 115, + 215 + ], + [ + 60, + 133 + ], + [ + 117, + 183 + ] + ], + [ + [ + 112, + 211 + ], + [ + 65, + 207 + ], + [ + 111, + 225 + ], + [ + 114, + 195 + ], + [ + 152, + 229 + ], + [ + 161, + 200 + ] + ], + [ + [ + 114, + 146 + ], + [ + 155, + 155 + ], + [ + 108, + 162 + ], + [ + 122, + 127 + ], + [ + 66, + 153 + ], + [ + 74, + 116 + ] + ], + [ + [ + 104, + 189 + ], + [ + 105, + 153 + ], + [ + 75, + 171 + ], + [ + 135, + 208 + ], + [ + 76, + 204 + ], + [ + 133, + 241 + ] + ] + ] + }, + { + "task_id": "0003676", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122063633#374", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003676/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 193, + 226 + ], + [ + 196, + 208 + ], + [ + 194, + 208 + ], + [ + 191, + 246 + ], + [ + 191, + 225 + ], + [ + 188, + 263 + ] + ], + [ + [ + 211, + 231 + ], + [ + 230, + 234 + ], + [ + 210, + 256 + ], + [ + 212, + 206 + ], + [ + 192, + 253 + ], + [ + 194, + 204 + ] + ], + [ + [ + 194, + 207 + ], + [ + 198, + 183 + ], + [ + 197, + 206 + ], + [ + 190, + 208 + ], + [ + 193, + 230 + ], + [ + 186, + 234 + ] + ], + [ + [ + 177, + 190 + ], + [ + 177, + 174 + ], + [ + 187, + 173 + ], + [ + 166, + 208 + ], + [ + 186, + 187 + ], + [ + 166, + 222 + ] + ] + ] + }, + { + "task_id": "0003677", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122063633#508", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003677/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 210, + 217 + ], + [ + 202, + 193 + ], + [ + 221, + 208 + ], + [ + 198, + 226 + ], + [ + 228, + 231 + ], + [ + 206, + 251 + ] + ], + [ + [ + 197, + 225 + ], + [ + 198, + 217 + ], + [ + 204, + 202 + ], + [ + 191, + 250 + ], + [ + 203, + 210 + ], + [ + 191, + 257 + ] + ], + [ + [ + 189, + 207 + ], + [ + 163, + 207 + ], + [ + 190, + 186 + ], + [ + 188, + 230 + ], + [ + 215, + 186 + ], + [ + 214, + 230 + ] + ], + [ + [ + 196, + 243 + ], + [ + 174, + 239 + ], + [ + 184, + 233 + ], + [ + 209, + 253 + ], + [ + 204, + 237 + ], + [ + 230, + 257 + ] + ] + ] + }, + { + "task_id": "0003680", + "id": "RH20T_cfg4_task_0016_user_0016_scene_0005_cfg_0004_104122062823#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003680/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 260, + 17 + ], + [ + 233, + 15 + ], + [ + 262, + 46 + ], + [ + 257, + 0 + ], + [ + 288, + 49 + ], + [ + 285, + 0 + ] + ], + [ + [ + 272, + 28 + ], + [ + 291, + 35 + ], + [ + 253, + 51 + ], + [ + 291, + 5 + ], + [ + 236, + 44 + ], + [ + 273, + 0 + ] + ], + [ + [ + 258, + 8 + ], + [ + 256, + 0 + ], + [ + 287, + 13 + ], + [ + 230, + 3 + ], + [ + 288, + 40 + ], + [ + 234, + 30 + ] + ], + [ + [ + 234, + 10 + ], + [ + 246, + 0 + ], + [ + 261, + 23 + ], + [ + 207, + 0 + ], + [ + 249, + 45 + ], + [ + 193, + 19 + ] + ] + ] + }, + { + "task_id": "0003682", + "id": "RH20T_cfg4_task_0046_user_0014_scene_0015_cfg_0004_104122063550#93", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003682/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 287, + 262 + ], + [ + 306, + 235 + ], + [ + 315, + 281 + ], + [ + 259, + 243 + ], + [ + 295, + 310 + ], + [ + 239, + 270 + ] + ], + [ + [ + 324, + 221 + ], + [ + 299, + 222 + ], + [ + 345, + 233 + ], + [ + 304, + 210 + ], + [ + 367, + 232 + ], + [ + 326, + 211 + ] + ], + [ + [ + 311, + 250 + ], + [ + 310, + 233 + ], + [ + 343, + 248 + ], + [ + 280, + 252 + ], + [ + 343, + 263 + ], + [ + 283, + 266 + ] + ], + [ + [ + 310, + 240 + ], + [ + 281, + 234 + ], + [ + 304, + 270 + ], + [ + 316, + 211 + ], + [ + 334, + 276 + ], + [ + 345, + 217 + ] + ] + ] + }, + { + "task_id": "0003685", + "id": "RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_104122061330#232", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003685/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 346, + 192 + ], + [ + 342, + 177 + ], + [ + 334, + 218 + ], + [ + 356, + 168 + ], + [ + 338, + 231 + ], + [ + 359, + 183 + ] + ], + [ + [ + 328, + 185 + ], + [ + 356, + 187 + ], + [ + 326, + 214 + ], + [ + 330, + 157 + ], + [ + 297, + 211 + ], + [ + 302, + 154 + ] + ], + [ + [ + 351, + 222 + ], + [ + 345, + 192 + ], + [ + 366, + 220 + ], + [ + 335, + 224 + ], + [ + 373, + 249 + ], + [ + 342, + 257 + ] + ], + [ + [ + 317, + 188 + ], + [ + 343, + 175 + ], + [ + 317, + 176 + ], + [ + 316, + 200 + ], + [ + 289, + 190 + ], + [ + 292, + 212 + ] + ] + ] + }, + { + "task_id": "0003686", + "id": "RH20T_cfg5_task_0063_user_0007_scene_0007_cfg_0005_104122063678#243", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a shovel to scoop up an object'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003686/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 385, + 246 + ], + [ + 395, + 233 + ], + [ + 396, + 268 + ], + [ + 374, + 226 + ], + [ + 386, + 280 + ], + [ + 365, + 238 + ] + ], + [ + [ + 381, + 224 + ], + [ + 371, + 240 + ], + [ + 360, + 223 + ], + [ + 401, + 225 + ], + [ + 370, + 206 + ], + [ + 411, + 208 + ] + ], + [ + [ + 396, + 255 + ], + [ + 419, + 252 + ], + [ + 401, + 271 + ], + [ + 390, + 240 + ], + [ + 377, + 275 + ], + [ + 368, + 244 + ] + ], + [ + [ + 348, + 242 + ], + [ + 346, + 224 + ], + [ + 371, + 246 + ], + [ + 326, + 238 + ], + [ + 372, + 263 + ], + [ + 328, + 254 + ] + ] + ] + }, + { + "task_id": "0003687", + "id": "RH20T_cfg6_task_0106_user_0020_scene_0004_cfg_0006_104122063633#173", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the object down from the shelf'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003687/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 230, + 49 + ], + [ + 238, + 26 + ], + [ + 231, + 55 + ], + [ + 229, + 43 + ], + [ + 224, + 77 + ], + [ + 223, + 63 + ] + ], + [ + [ + 223, + 37 + ], + [ + 224, + 13 + ], + [ + 227, + 29 + ], + [ + 218, + 45 + ], + [ + 226, + 51 + ], + [ + 217, + 69 + ] + ], + [ + [ + 239, + 16 + ], + [ + 263, + 23 + ], + [ + 242, + 8 + ], + [ + 235, + 25 + ], + [ + 219, + 1 + ], + [ + 210, + 18 + ] + ], + [ + [ + 204, + 64 + ], + [ + 212, + 68 + ], + [ + 186, + 80 + ], + [ + 222, + 48 + ], + [ + 179, + 76 + ], + [ + 213, + 45 + ] + ] + ] + }, + { + "task_id": "0003689", + "id": "RH20T_cfg4_task_0127_user_0014_scene_0010_cfg_0004_104122062823#89", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003689/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 451, + 0 + ], + [ + 423, + 0 + ], + [ + 482, + 0 + ], + [ + 417, + 0 + ], + [ + 516, + 0 + ], + [ + 447, + 0 + ] + ], + [ + [ + 451, + 0 + ], + [ + 478, + 0 + ], + [ + 417, + 0 + ], + [ + 486, + 0 + ], + [ + 392, + 0 + ], + [ + 458, + 0 + ] + ], + [ + [ + 395, + 0 + ], + [ + 432, + 0 + ], + [ + 378, + 0 + ], + [ + 414, + 0 + ], + [ + 340, + 0 + ], + [ + 375, + 0 + ] + ], + [ + [ + 421, + 0 + ], + [ + 415, + 0 + ], + [ + 450, + 0 + ], + [ + 391, + 0 + ], + [ + 453, + 0 + ], + [ + 398, + 1 + ] + ] + ] + }, + { + "task_id": "0003691", + "id": "RH20T_cfg2_task_0010_user_0003_scene_0002_cfg_0002_104122063678#403", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003691/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 196 + ], + [ + 329, + 189 + ], + [ + 303, + 198 + ], + [ + 316, + 194 + ], + [ + 284, + 205 + ], + [ + 295, + 202 + ] + ], + [ + [ + 338, + 171 + ], + [ + 324, + 172 + ], + [ + 341, + 194 + ], + [ + 335, + 148 + ], + [ + 353, + 192 + ], + [ + 347, + 148 + ] + ], + [ + [ + 342, + 202 + ], + [ + 321, + 205 + ], + [ + 335, + 192 + ], + [ + 348, + 212 + ], + [ + 357, + 189 + ], + [ + 369, + 210 + ] + ], + [ + [ + 327, + 195 + ], + [ + 327, + 180 + ], + [ + 338, + 211 + ], + [ + 318, + 180 + ], + [ + 338, + 225 + ], + [ + 319, + 194 + ] + ] + ] + }, + { + "task_id": "0003692", + "id": "RH20T_cfg6_task_0038_user_0014_scene_0010_cfg_0006_104122064161#245", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003692/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 328, + 119 + ], + [ + 328, + 101 + ], + [ + 325, + 135 + ], + [ + 330, + 104 + ], + [ + 325, + 153 + ], + [ + 330, + 122 + ] + ], + [ + [ + 345, + 122 + ], + [ + 356, + 102 + ], + [ + 363, + 133 + ], + [ + 327, + 110 + ], + [ + 352, + 152 + ], + [ + 315, + 130 + ] + ], + [ + [ + 331, + 122 + ], + [ + 342, + 133 + ], + [ + 314, + 136 + ], + [ + 349, + 108 + ], + [ + 302, + 124 + ], + [ + 338, + 95 + ] + ], + [ + [ + 331, + 130 + ], + [ + 344, + 150 + ], + [ + 330, + 133 + ], + [ + 332, + 127 + ], + [ + 317, + 114 + ], + [ + 317, + 106 + ] + ] + ] + }, + { + "task_id": "0003694", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0002_cfg_0006_104122060811#274", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003694/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 321, + 142 + ], + [ + 325, + 142 + ], + [ + 332, + 132 + ], + [ + 311, + 152 + ], + [ + 329, + 132 + ], + [ + 307, + 152 + ] + ], + [ + [ + 316, + 139 + ], + [ + 324, + 126 + ], + [ + 315, + 141 + ], + [ + 318, + 136 + ], + [ + 308, + 154 + ], + [ + 310, + 149 + ] + ], + [ + [ + 309, + 142 + ], + [ + 323, + 138 + ], + [ + 310, + 153 + ], + [ + 307, + 130 + ], + [ + 296, + 157 + ], + [ + 293, + 134 + ] + ], + [ + [ + 322, + 136 + ], + [ + 328, + 123 + ], + [ + 336, + 141 + ], + [ + 308, + 130 + ], + [ + 330, + 155 + ], + [ + 302, + 143 + ] + ] + ] + }, + { + "task_id": "0003695", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0002_cfg_0006_104122060811#542", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003695/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 121 + ], + [ + 361, + 119 + ], + [ + 345, + 129 + ], + [ + 349, + 113 + ], + [ + 331, + 132 + ], + [ + 334, + 116 + ] + ], + [ + [ + 349, + 128 + ], + [ + 352, + 127 + ], + [ + 362, + 136 + ], + [ + 336, + 121 + ], + [ + 359, + 138 + ], + [ + 333, + 123 + ] + ], + [ + [ + 353, + 115 + ], + [ + 357, + 127 + ], + [ + 356, + 106 + ], + [ + 351, + 123 + ], + [ + 352, + 95 + ], + [ + 347, + 112 + ] + ], + [ + [ + 350, + 109 + ], + [ + 363, + 105 + ], + [ + 349, + 120 + ], + [ + 352, + 99 + ], + [ + 336, + 124 + ], + [ + 339, + 103 + ] + ] + ] + }, + { + "task_id": "0003696", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0002_cfg_0006_104122060811#610", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003696/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 362, + 104 + ], + [ + 377, + 104 + ], + [ + 361, + 119 + ], + [ + 363, + 89 + ], + [ + 346, + 118 + ], + [ + 349, + 89 + ] + ], + [ + [ + 357, + 94 + ], + [ + 343, + 89 + ], + [ + 362, + 84 + ], + [ + 353, + 103 + ], + [ + 376, + 89 + ], + [ + 367, + 108 + ] + ], + [ + [ + 354, + 110 + ], + [ + 369, + 107 + ], + [ + 353, + 119 + ], + [ + 356, + 101 + ], + [ + 339, + 121 + ], + [ + 341, + 104 + ] + ], + [ + [ + 350, + 103 + ], + [ + 346, + 92 + ], + [ + 363, + 95 + ], + [ + 338, + 111 + ], + [ + 368, + 106 + ], + [ + 342, + 123 + ] + ] + ] + }, + { + "task_id": "0003697", + "id": "RH20T_cfg4_task_0067_user_0015_scene_0006_cfg_0004_104122062295#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button from top to bottom with obstacles'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003697/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 174, + 126 + ], + [ + 138, + 141 + ], + [ + 168, + 99 + ], + [ + 181, + 155 + ], + [ + 203, + 85 + ], + [ + 220, + 139 + ] + ], + [ + [ + 153, + 85 + ], + [ + 160, + 47 + ], + [ + 120, + 71 + ], + [ + 187, + 100 + ], + [ + 115, + 106 + ], + [ + 180, + 135 + ] + ], + [ + [ + 118, + 113 + ], + [ + 142, + 103 + ], + [ + 90, + 80 + ], + [ + 145, + 146 + ], + [ + 59, + 89 + ], + [ + 120, + 161 + ] + ], + [ + [ + 136, + 93 + ], + [ + 150, + 62 + ], + [ + 95, + 82 + ], + [ + 173, + 104 + ], + [ + 83, + 113 + ], + [ + 159, + 132 + ] + ] + ] + }, + { + "task_id": "0003698", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0004_cfg_0006_104122060811#250", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003698/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 381, + 101 + ], + [ + 394, + 96 + ], + [ + 384, + 104 + ], + [ + 378, + 98 + ], + [ + 370, + 109 + ], + [ + 365, + 103 + ] + ], + [ + [ + 400, + 102 + ], + [ + 399, + 89 + ], + [ + 400, + 106 + ], + [ + 400, + 99 + ], + [ + 402, + 119 + ], + [ + 401, + 113 + ] + ], + [ + [ + 381, + 105 + ], + [ + 391, + 107 + ], + [ + 373, + 114 + ], + [ + 390, + 95 + ], + [ + 363, + 112 + ], + [ + 379, + 92 + ] + ], + [ + [ + 394, + 102 + ], + [ + 393, + 115 + ], + [ + 404, + 106 + ], + [ + 383, + 98 + ], + [ + 405, + 93 + ], + [ + 384, + 85 + ] + ] + ] + }, + { + "task_id": "0003699", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0004_cfg_0006_104122060811#588", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003699/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 416, + 101 + ], + [ + 420, + 88 + ], + [ + 428, + 106 + ], + [ + 403, + 96 + ], + [ + 423, + 119 + ], + [ + 399, + 110 + ] + ], + [ + [ + 400, + 103 + ], + [ + 413, + 101 + ], + [ + 397, + 112 + ], + [ + 404, + 94 + ], + [ + 384, + 114 + ], + [ + 391, + 96 + ] + ], + [ + [ + 395, + 96 + ], + [ + 392, + 109 + ], + [ + 383, + 97 + ], + [ + 408, + 96 + ], + [ + 386, + 84 + ], + [ + 412, + 83 + ] + ], + [ + [ + 401, + 92 + ], + [ + 415, + 87 + ], + [ + 398, + 80 + ], + [ + 404, + 102 + ], + [ + 385, + 85 + ], + [ + 391, + 106 + ] + ] + ] + }, + { + "task_id": "0003700", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0004_cfg_0006_104122060811#928", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003700/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 388, + 141 + ], + [ + 377, + 139 + ], + [ + 390, + 154 + ], + [ + 385, + 128 + ], + [ + 401, + 156 + ], + [ + 396, + 130 + ] + ], + [ + [ + 387, + 129 + ], + [ + 402, + 128 + ], + [ + 386, + 120 + ], + [ + 387, + 138 + ], + [ + 370, + 122 + ], + [ + 372, + 139 + ] + ], + [ + [ + 384, + 129 + ], + [ + 395, + 131 + ], + [ + 376, + 138 + ], + [ + 393, + 120 + ], + [ + 365, + 137 + ], + [ + 382, + 118 + ] + ], + [ + [ + 381, + 137 + ], + [ + 378, + 124 + ], + [ + 381, + 131 + ], + [ + 380, + 142 + ], + [ + 384, + 144 + ], + [ + 383, + 155 + ] + ] + ] + }, + { + "task_id": "0003701", + "id": "RH20T_cfg1_task_0119_user_0014_scene_0004_cfg_0001_819112070044#105", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn off the water tap'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003701/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 342, + 171 + ], + [ + 344, + 155 + ], + [ + 354, + 162 + ], + [ + 330, + 181 + ], + [ + 351, + 178 + ], + [ + 328, + 197 + ] + ], + [ + [ + 342, + 191 + ], + [ + 328, + 179 + ], + [ + 340, + 203 + ], + [ + 344, + 178 + ], + [ + 354, + 215 + ], + [ + 359, + 191 + ] + ], + [ + [ + 346, + 155 + ], + [ + 337, + 144 + ], + [ + 350, + 171 + ], + [ + 341, + 139 + ], + [ + 358, + 181 + ], + [ + 350, + 149 + ] + ], + [ + [ + 364, + 157 + ], + [ + 384, + 157 + ], + [ + 365, + 177 + ], + [ + 363, + 136 + ], + [ + 346, + 177 + ], + [ + 343, + 136 + ] + ] + ] + }, + { + "task_id": "0003705", + "id": "RH20T_cfg3_task_0038_user_0010_scene_0003_cfg_0003_104122062823#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003705/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 156, + 134 + ], + [ + 124, + 163 + ], + [ + 121, + 101 + ], + [ + 189, + 165 + ], + [ + 152, + 76 + ], + [ + 216, + 138 + ] + ], + [ + [ + 212, + 161 + ], + [ + 203, + 141 + ], + [ + 224, + 130 + ], + [ + 201, + 190 + ], + [ + 236, + 152 + ], + [ + 211, + 214 + ] + ], + [ + [ + 173, + 131 + ], + [ + 174, + 94 + ], + [ + 204, + 148 + ], + [ + 144, + 115 + ], + [ + 202, + 180 + ], + [ + 145, + 147 + ] + ], + [ + [ + 137, + 114 + ], + [ + 107, + 80 + ], + [ + 150, + 115 + ], + [ + 125, + 113 + ], + [ + 178, + 148 + ], + [ + 151, + 142 + ] + ] + ] + }, + { + "task_id": "0003709", + "id": "RH20T_cfg6_task_0060_user_0014_scene_0006_cfg_0006_104122060811#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to decrease the volume of the speaker'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003709/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 158 + ], + [ + 376, + 159 + ], + [ + 359, + 171 + ], + [ + 368, + 144 + ], + [ + 346, + 169 + ], + [ + 355, + 143 + ] + ], + [ + [ + 369, + 126 + ], + [ + 367, + 117 + ], + [ + 384, + 126 + ], + [ + 354, + 126 + ], + [ + 386, + 136 + ], + [ + 355, + 135 + ] + ], + [ + [ + 367, + 143 + ], + [ + 361, + 129 + ], + [ + 378, + 140 + ], + [ + 355, + 145 + ], + [ + 384, + 154 + ], + [ + 361, + 159 + ] + ], + [ + [ + 371, + 142 + ], + [ + 384, + 136 + ], + [ + 375, + 148 + ], + [ + 366, + 136 + ], + [ + 362, + 154 + ], + [ + 352, + 143 + ] + ] + ] + }, + { + "task_id": "0003714", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0003_cfg_0006_104122061018#715", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003714/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 312, + 74 + ], + [ + 319, + 52 + ], + [ + 300, + 45 + ], + [ + 324, + 103 + ], + [ + 294, + 66 + ], + [ + 317, + 120 + ] + ], + [ + [ + 347, + 63 + ], + [ + 382, + 55 + ], + [ + 333, + 43 + ], + [ + 359, + 81 + ], + [ + 298, + 51 + ], + [ + 328, + 87 + ] + ], + [ + [ + 341, + 97 + ], + [ + 309, + 101 + ], + [ + 333, + 120 + ], + [ + 350, + 71 + ], + [ + 361, + 116 + ], + [ + 381, + 67 + ] + ], + [ + [ + 358, + 77 + ], + [ + 334, + 84 + ], + [ + 357, + 115 + ], + [ + 360, + 41 + ], + [ + 384, + 110 + ], + [ + 387, + 31 + ] + ] + ] + }, + { + "task_id": "0003715", + "id": "RH20T_cfg5_task_0082_user_0007_scene_0004_cfg_0005_037522062165#231", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003715/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 206 + ], + [ + 291, + 226 + ], + [ + 280, + 195 + ], + [ + 332, + 217 + ], + [ + 294, + 177 + ], + [ + 344, + 198 + ] + ], + [ + [ + 283, + 248 + ], + [ + 254, + 241 + ], + [ + 274, + 261 + ], + [ + 292, + 233 + ], + [ + 301, + 267 + ], + [ + 321, + 240 + ] + ], + [ + [ + 282, + 199 + ], + [ + 278, + 205 + ], + [ + 314, + 205 + ], + [ + 252, + 193 + ], + [ + 315, + 199 + ], + [ + 257, + 187 + ] + ], + [ + [ + 270, + 209 + ], + [ + 271, + 191 + ], + [ + 249, + 192 + ], + [ + 292, + 226 + ], + [ + 249, + 208 + ], + [ + 291, + 242 + ] + ] + ] + }, + { + "task_id": "0003720", + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_038522062547#253", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003720/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 380, + 265 + ], + [ + 384, + 258 + ], + [ + 407, + 288 + ], + [ + 354, + 242 + ], + [ + 402, + 292 + ], + [ + 353, + 249 + ] + ], + [ + [ + 396, + 237 + ], + [ + 370, + 253 + ], + [ + 380, + 207 + ], + [ + 412, + 266 + ], + [ + 405, + 192 + ], + [ + 437, + 250 + ] + ], + [ + [ + 346, + 255 + ], + [ + 368, + 244 + ], + [ + 326, + 233 + ], + [ + 367, + 279 + ], + [ + 301, + 244 + ], + [ + 343, + 293 + ] + ], + [ + [ + 374, + 300 + ], + [ + 385, + 299 + ], + [ + 382, + 268 + ], + [ + 365, + 332 + ], + [ + 371, + 267 + ], + [ + 352, + 336 + ] + ] + ] + }, + { + "task_id": "0003721", + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_038522062547#374", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003721/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 383, + 248 + ], + [ + 409, + 246 + ], + [ + 384, + 280 + ], + [ + 383, + 218 + ], + [ + 360, + 281 + ], + [ + 359, + 221 + ] + ], + [ + [ + 371, + 274 + ], + [ + 386, + 263 + ], + [ + 350, + 294 + ], + [ + 390, + 256 + ], + [ + 337, + 303 + ], + [ + 376, + 266 + ] + ], + [ + [ + 392, + 267 + ], + [ + 396, + 259 + ], + [ + 417, + 292 + ], + [ + 368, + 243 + ], + [ + 412, + 298 + ], + [ + 366, + 251 + ] + ], + [ + [ + 425, + 270 + ], + [ + 404, + 291 + ], + [ + 409, + 241 + ], + [ + 440, + 297 + ], + [ + 430, + 222 + ], + [ + 458, + 276 + ] + ] + ] + }, + { + "task_id": "0003722", + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_038522062547#615", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003722/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 383, + 254 + ], + [ + 376, + 252 + ], + [ + 383, + 286 + ], + [ + 383, + 224 + ], + [ + 389, + 287 + ], + [ + 389, + 228 + ] + ], + [ + [ + 399, + 270 + ], + [ + 387, + 253 + ], + [ + 405, + 239 + ], + [ + 393, + 298 + ], + [ + 419, + 257 + ], + [ + 405, + 319 + ] + ], + [ + [ + 365, + 264 + ], + [ + 367, + 263 + ], + [ + 383, + 292 + ], + [ + 347, + 236 + ], + [ + 380, + 291 + ], + [ + 346, + 239 + ] + ], + [ + [ + 344, + 267 + ], + [ + 364, + 240 + ], + [ + 357, + 274 + ], + [ + 332, + 261 + ], + [ + 336, + 305 + ], + [ + 313, + 287 + ] + ] + ] + }, + { + "task_id": "0003724", + "id": "RH20T_cfg4_task_0009_user_0010_scene_0004_cfg_0004_104122063550#75", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003724/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 335, + 204 + ], + [ + 363, + 209 + ], + [ + 323, + 219 + ], + [ + 345, + 189 + ], + [ + 296, + 213 + ], + [ + 320, + 185 + ] + ], + [ + [ + 289, + 198 + ], + [ + 314, + 184 + ], + [ + 299, + 218 + ], + [ + 278, + 177 + ], + [ + 275, + 232 + ], + [ + 252, + 192 + ] + ], + [ + [ + 321, + 220 + ], + [ + 294, + 223 + ], + [ + 330, + 241 + ], + [ + 312, + 201 + ], + [ + 356, + 238 + ], + [ + 337, + 199 + ] + ], + [ + [ + 311, + 214 + ], + [ + 312, + 203 + ], + [ + 323, + 240 + ], + [ + 299, + 189 + ], + [ + 322, + 248 + ], + [ + 299, + 199 + ] + ] + ] + }, + { + "task_id": "0003725", + "id": "RH20T_cfg1_task_0015_user_0001_scene_0006_cfg_0001_750612070851#285", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003725/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 386, + 161 + ], + [ + 415, + 157 + ], + [ + 401, + 137 + ], + [ + 369, + 186 + ], + [ + 375, + 142 + ], + [ + 343, + 188 + ] + ], + [ + [ + 340, + 150 + ], + [ + 344, + 131 + ], + [ + 358, + 179 + ], + [ + 323, + 124 + ], + [ + 353, + 194 + ], + [ + 321, + 142 + ] + ], + [ + [ + 355, + 133 + ], + [ + 361, + 123 + ], + [ + 381, + 157 + ], + [ + 330, + 109 + ], + [ + 376, + 169 + ], + [ + 322, + 119 + ] + ], + [ + [ + 344, + 178 + ], + [ + 343, + 172 + ], + [ + 316, + 155 + ], + [ + 371, + 202 + ], + [ + 314, + 160 + ], + [ + 375, + 211 + ] + ] + ] + }, + { + "task_id": "0003726", + "id": "RH20T_cfg1_task_0001_user_0009_scene_0003_cfg_0001_750612070851#72", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003726/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 310, + 202 + ], + [ + 325, + 228 + ], + [ + 309, + 173 + ], + [ + 310, + 228 + ], + [ + 295, + 150 + ], + [ + 297, + 205 + ] + ], + [ + [ + 368, + 183 + ], + [ + 331, + 166 + ], + [ + 353, + 218 + ], + [ + 382, + 149 + ], + [ + 390, + 236 + ], + [ + 417, + 166 + ] + ], + [ + [ + 330, + 186 + ], + [ + 326, + 220 + ], + [ + 339, + 197 + ], + [ + 318, + 173 + ], + [ + 343, + 167 + ], + [ + 323, + 139 + ] + ], + [ + [ + 335, + 193 + ], + [ + 336, + 173 + ], + [ + 297, + 195 + ], + [ + 373, + 192 + ], + [ + 298, + 213 + ], + [ + 370, + 210 + ] + ] + ] + }, + { + "task_id": "0003727", + "id": "RH20T_cfg6_task_0013_user_0014_scene_0007_cfg_0006_104122064161#84", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003727/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 290, + 159 + ], + [ + 288, + 136 + ], + [ + 304, + 166 + ], + [ + 277, + 152 + ], + [ + 306, + 189 + ], + [ + 280, + 174 + ] + ], + [ + [ + 320, + 150 + ], + [ + 339, + 165 + ], + [ + 304, + 170 + ], + [ + 336, + 130 + ], + [ + 285, + 155 + ], + [ + 316, + 115 + ] + ], + [ + [ + 284, + 140 + ], + [ + 307, + 145 + ], + [ + 289, + 119 + ], + [ + 280, + 163 + ], + [ + 266, + 114 + ], + [ + 256, + 158 + ] + ], + [ + [ + 298, + 159 + ], + [ + 300, + 147 + ], + [ + 294, + 180 + ], + [ + 302, + 138 + ], + [ + 292, + 190 + ], + [ + 300, + 150 + ] + ] + ] + }, + { + "task_id": "0003732", + "id": "RH20T_cfg3_task_0091_user_0010_scene_0006_cfg_0003_104122062823#237", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003732/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 310, + 340 + ], + [ + 263, + 320 + ], + [ + 295, + 355 + ], + [ + 327, + 323 + ], + [ + 338, + 360 + ], + [ + 377, + 346 + ] + ], + [ + [ + 341, + 274 + ], + [ + 333, + 292 + ], + [ + 318, + 236 + ], + [ + 363, + 311 + ], + [ + 326, + 224 + ], + [ + 367, + 293 + ] + ], + [ + [ + 310, + 269 + ], + [ + 315, + 228 + ], + [ + 358, + 276 + ], + [ + 263, + 263 + ], + [ + 352, + 314 + ], + [ + 261, + 301 + ] + ], + [ + [ + 272, + 280 + ], + [ + 290, + 274 + ], + [ + 244, + 308 + ], + [ + 301, + 251 + ], + [ + 221, + 317 + ], + [ + 282, + 255 + ] + ] + ] + }, + { + "task_id": "0003735", + "id": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_104122063550#70", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003735/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 130 + ], + [ + 324, + 133 + ], + [ + 343, + 122 + ], + [ + 352, + 138 + ], + [ + 366, + 118 + ], + [ + 377, + 135 + ] + ], + [ + [ + 358, + 126 + ], + [ + 380, + 119 + ], + [ + 351, + 104 + ], + [ + 365, + 149 + ], + [ + 328, + 111 + ], + [ + 342, + 155 + ] + ], + [ + [ + 362, + 138 + ], + [ + 365, + 145 + ], + [ + 338, + 140 + ], + [ + 387, + 136 + ], + [ + 334, + 133 + ], + [ + 385, + 129 + ] + ], + [ + [ + 339, + 138 + ], + [ + 342, + 126 + ], + [ + 343, + 160 + ], + [ + 335, + 116 + ], + [ + 340, + 170 + ], + [ + 332, + 128 + ] + ] + ] + }, + { + "task_id": "0003736", + "id": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_104122063550#83", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003736/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 386, + 134 + ], + [ + 400, + 154 + ], + [ + 377, + 142 + ], + [ + 396, + 125 + ], + [ + 364, + 122 + ], + [ + 381, + 103 + ] + ], + [ + [ + 381, + 136 + ], + [ + 370, + 140 + ], + [ + 360, + 134 + ], + [ + 402, + 138 + ], + [ + 372, + 129 + ], + [ + 416, + 133 + ] + ], + [ + [ + 358, + 137 + ], + [ + 363, + 125 + ], + [ + 363, + 159 + ], + [ + 354, + 116 + ], + [ + 359, + 169 + ], + [ + 350, + 128 + ] + ], + [ + [ + 366, + 157 + ], + [ + 346, + 148 + ], + [ + 355, + 166 + ], + [ + 380, + 148 + ], + [ + 374, + 175 + ], + [ + 400, + 158 + ] + ] + ] + }, + { + "task_id": "0003738", + "id": "RH20T_cfg3_task_0058_user_0010_scene_0010_cfg_0003_104122063550#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003738/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 413, + 180 + ], + [ + 417, + 171 + ], + [ + 420, + 203 + ], + [ + 407, + 158 + ], + [ + 417, + 211 + ], + [ + 404, + 167 + ] + ], + [ + [ + 432, + 184 + ], + [ + 436, + 208 + ], + [ + 409, + 186 + ], + [ + 454, + 181 + ], + [ + 404, + 163 + ], + [ + 449, + 158 + ] + ], + [ + [ + 432, + 148 + ], + [ + 415, + 165 + ], + [ + 424, + 154 + ], + [ + 439, + 142 + ], + [ + 443, + 134 + ], + [ + 456, + 125 + ] + ], + [ + [ + 427, + 181 + ], + [ + 415, + 171 + ], + [ + 405, + 187 + ], + [ + 448, + 176 + ], + [ + 417, + 198 + ], + [ + 462, + 187 + ] + ] + ] + }, + { + "task_id": "0003739", + "id": "RH20T_cfg6_task_0014_user_0014_scene_0008_cfg_0006_037522061512#110", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the object from the scale'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003739/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 269, + 142 + ], + [ + 252, + 149 + ], + [ + 271, + 165 + ], + [ + 267, + 118 + ], + [ + 287, + 157 + ], + [ + 284, + 112 + ] + ], + [ + [ + 248, + 141 + ], + [ + 258, + 159 + ], + [ + 272, + 133 + ], + [ + 224, + 150 + ], + [ + 262, + 117 + ], + [ + 215, + 133 + ] + ], + [ + [ + 251, + 148 + ], + [ + 246, + 127 + ], + [ + 264, + 135 + ], + [ + 238, + 163 + ], + [ + 269, + 154 + ], + [ + 244, + 183 + ] + ], + [ + [ + 239, + 139 + ], + [ + 223, + 123 + ], + [ + 236, + 129 + ], + [ + 241, + 149 + ], + [ + 253, + 146 + ], + [ + 256, + 164 + ] + ] + ] + }, + { + "task_id": "0003740", + "id": "RH20T_cfg5_task_0042_user_0007_scene_0003_cfg_0005_037522062165#359", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003740/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 201, + 292 + ], + [ + 199, + 278 + ], + [ + 180, + 275 + ], + [ + 224, + 310 + ], + [ + 183, + 288 + ], + [ + 225, + 322 + ] + ], + [ + [ + 205, + 299 + ], + [ + 228, + 279 + ], + [ + 200, + 288 + ], + [ + 210, + 309 + ], + [ + 174, + 310 + ], + [ + 187, + 329 + ] + ], + [ + [ + 233, + 294 + ], + [ + 205, + 301 + ], + [ + 242, + 280 + ], + [ + 222, + 310 + ], + [ + 268, + 274 + ], + [ + 253, + 303 + ] + ], + [ + [ + 167, + 294 + ], + [ + 164, + 297 + ], + [ + 175, + 324 + ], + [ + 159, + 264 + ], + [ + 177, + 320 + ], + [ + 162, + 263 + ] + ] + ] + }, + { + "task_id": "0003742", + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122061602#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003742/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 519, + 189 + ], + [ + 497, + 194 + ], + [ + 537, + 211 + ], + [ + 502, + 168 + ], + [ + 558, + 205 + ], + [ + 523, + 164 + ] + ], + [ + [ + 495, + 184 + ], + [ + 516, + 176 + ], + [ + 508, + 210 + ], + [ + 483, + 159 + ], + [ + 488, + 217 + ], + [ + 465, + 168 + ] + ], + [ + [ + 463, + 189 + ], + [ + 456, + 185 + ], + [ + 437, + 198 + ], + [ + 491, + 180 + ], + [ + 444, + 201 + ], + [ + 496, + 183 + ] + ], + [ + [ + 497, + 201 + ], + [ + 501, + 227 + ], + [ + 492, + 206 + ], + [ + 502, + 194 + ], + [ + 488, + 182 + ], + [ + 497, + 168 + ] + ] + ] + }, + { + "task_id": "0003749", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0004_cfg_0004_104422070011#405", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003749/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 318, + 135 + ], + [ + 317, + 150 + ], + [ + 288, + 138 + ], + [ + 347, + 132 + ], + [ + 286, + 121 + ], + [ + 350, + 115 + ] + ], + [ + [ + 309, + 100 + ], + [ + 324, + 126 + ], + [ + 318, + 101 + ], + [ + 301, + 100 + ], + [ + 301, + 72 + ], + [ + 286, + 74 + ] + ], + [ + [ + 299, + 101 + ], + [ + 309, + 88 + ], + [ + 294, + 130 + ], + [ + 304, + 74 + ], + [ + 286, + 141 + ], + [ + 295, + 87 + ] + ], + [ + [ + 315, + 123 + ], + [ + 347, + 116 + ], + [ + 312, + 141 + ], + [ + 318, + 107 + ], + [ + 279, + 147 + ], + [ + 289, + 114 + ] + ] + ] + }, + { + "task_id": "0003750", + "id": "RH20T_cfg2_task_0010_user_0003_scene_0002_cfg_0002_037522062165#255", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003750/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 282, + 292 + ], + [ + 304, + 270 + ], + [ + 262, + 271 + ], + [ + 301, + 312 + ], + [ + 238, + 294 + ], + [ + 279, + 334 + ] + ], + [ + [ + 286, + 278 + ], + [ + 288, + 265 + ], + [ + 262, + 264 + ], + [ + 311, + 292 + ], + [ + 262, + 276 + ], + [ + 308, + 303 + ] + ], + [ + [ + 259, + 285 + ], + [ + 257, + 281 + ], + [ + 243, + 309 + ], + [ + 275, + 261 + ], + [ + 244, + 315 + ], + [ + 278, + 264 + ] + ], + [ + [ + 259, + 283 + ], + [ + 279, + 263 + ], + [ + 276, + 306 + ], + [ + 241, + 260 + ], + [ + 255, + 328 + ], + [ + 218, + 281 + ] + ] + ] + }, + { + "task_id": "0003751", + "id": "RH20T_cfg4_task_0024_user_0014_scene_0007_cfg_0004_104122063550#48", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn off the power strip by pressing the button'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003751/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 270, + 243 + ], + [ + 265, + 226 + ], + [ + 294, + 254 + ], + [ + 247, + 232 + ], + [ + 298, + 270 + ], + [ + 253, + 247 + ] + ], + [ + [ + 260, + 206 + ], + [ + 238, + 209 + ], + [ + 241, + 198 + ], + [ + 281, + 214 + ], + [ + 261, + 196 + ], + [ + 300, + 211 + ] + ], + [ + [ + 260, + 221 + ], + [ + 262, + 221 + ], + [ + 256, + 249 + ], + [ + 264, + 193 + ], + [ + 253, + 251 + ], + [ + 262, + 192 + ] + ], + [ + [ + 232, + 243 + ], + [ + 232, + 252 + ], + [ + 201, + 240 + ], + [ + 262, + 247 + ], + [ + 199, + 230 + ], + [ + 264, + 237 + ] + ] + ] + }, + { + "task_id": "0003756", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0010_cfg_0003_104122062823#368", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003756/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 384, + 200 + ], + [ + 399, + 181 + ], + [ + 422, + 210 + ], + [ + 346, + 190 + ], + [ + 407, + 225 + ], + [ + 336, + 207 + ] + ], + [ + [ + 342, + 187 + ], + [ + 309, + 202 + ], + [ + 345, + 169 + ], + [ + 339, + 203 + ], + [ + 379, + 154 + ], + [ + 369, + 189 + ] + ], + [ + [ + 359, + 162 + ], + [ + 325, + 181 + ], + [ + 378, + 196 + ], + [ + 341, + 127 + ], + [ + 412, + 177 + ], + [ + 375, + 109 + ] + ], + [ + [ + 409, + 213 + ], + [ + 426, + 182 + ], + [ + 445, + 213 + ], + [ + 371, + 213 + ], + [ + 429, + 241 + ], + [ + 357, + 242 + ] + ] + ] + }, + { + "task_id": "0003757", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0010_cfg_0003_104122062823#404", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003757/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 418, + 201 + ], + [ + 437, + 181 + ], + [ + 455, + 209 + ], + [ + 380, + 192 + ], + [ + 436, + 225 + ], + [ + 366, + 210 + ] + ], + [ + [ + 448, + 191 + ], + [ + 440, + 224 + ], + [ + 408, + 192 + ], + [ + 487, + 189 + ], + [ + 414, + 155 + ], + [ + 497, + 152 + ] + ], + [ + [ + 414, + 239 + ], + [ + 422, + 281 + ], + [ + 454, + 234 + ], + [ + 374, + 244 + ], + [ + 445, + 194 + ], + [ + 367, + 204 + ] + ], + [ + [ + 443, + 179 + ], + [ + 443, + 148 + ], + [ + 403, + 176 + ], + [ + 485, + 182 + ], + [ + 401, + 209 + ], + [ + 487, + 217 + ] + ] + ] + }, + { + "task_id": "0003758", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0010_cfg_0003_104122062823#481", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003758/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 434, + 209 + ], + [ + 429, + 189 + ], + [ + 418, + 244 + ], + [ + 449, + 176 + ], + [ + 423, + 259 + ], + [ + 452, + 195 + ] + ], + [ + [ + 459, + 198 + ], + [ + 461, + 165 + ], + [ + 476, + 214 + ], + [ + 439, + 181 + ], + [ + 474, + 248 + ], + [ + 436, + 217 + ] + ], + [ + [ + 449, + 203 + ], + [ + 468, + 183 + ], + [ + 486, + 211 + ], + [ + 411, + 196 + ], + [ + 468, + 228 + ], + [ + 397, + 214 + ] + ], + [ + [ + 449, + 226 + ], + [ + 467, + 263 + ], + [ + 448, + 199 + ], + [ + 451, + 250 + ], + [ + 431, + 167 + ], + [ + 436, + 219 + ] + ] + ] + }, + { + "task_id": "0003760", + "id": "RH20T_cfg4_task_0020_user_0016_scene_0002_cfg_0004_104422070011#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003760/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 352, + 115 + ], + [ + 355, + 104 + ], + [ + 344, + 137 + ], + [ + 360, + 94 + ], + [ + 342, + 146 + ], + [ + 357, + 105 + ] + ], + [ + [ + 336, + 106 + ], + [ + 318, + 123 + ], + [ + 321, + 91 + ], + [ + 352, + 121 + ], + [ + 339, + 75 + ], + [ + 370, + 105 + ] + ], + [ + [ + 359, + 103 + ], + [ + 337, + 97 + ], + [ + 355, + 91 + ], + [ + 363, + 116 + ], + [ + 376, + 97 + ], + [ + 386, + 123 + ] + ], + [ + [ + 351, + 125 + ], + [ + 372, + 113 + ], + [ + 355, + 123 + ], + [ + 347, + 128 + ], + [ + 332, + 136 + ], + [ + 327, + 139 + ] + ] + ] + }, + { + "task_id": "0003761", + "id": "RH20T_cfg6_task_0006_user_0014_scene_0005_cfg_0006_104122060811#184", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003761/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 396, + 132 + ], + [ + 401, + 123 + ], + [ + 408, + 132 + ], + [ + 383, + 133 + ], + [ + 403, + 140 + ], + [ + 378, + 141 + ] + ], + [ + [ + 395, + 135 + ], + [ + 409, + 139 + ], + [ + 398, + 128 + ], + [ + 392, + 142 + ], + [ + 384, + 124 + ], + [ + 377, + 138 + ] + ], + [ + [ + 403, + 132 + ], + [ + 402, + 146 + ], + [ + 406, + 133 + ], + [ + 400, + 131 + ], + [ + 407, + 118 + ], + [ + 401, + 117 + ] + ], + [ + [ + 385, + 135 + ], + [ + 399, + 132 + ], + [ + 385, + 143 + ], + [ + 384, + 128 + ], + [ + 371, + 146 + ], + [ + 371, + 131 + ] + ] + ] + }, + { + "task_id": "0003764", + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_036422060215#169", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003764/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 396, + 257 + ], + [ + 402, + 242 + ], + [ + 413, + 256 + ], + [ + 378, + 259 + ], + [ + 407, + 270 + ], + [ + 372, + 274 + ] + ], + [ + [ + 379, + 251 + ], + [ + 382, + 231 + ], + [ + 359, + 248 + ], + [ + 400, + 253 + ], + [ + 356, + 268 + ], + [ + 397, + 274 + ] + ], + [ + [ + 400, + 252 + ], + [ + 400, + 272 + ], + [ + 397, + 258 + ], + [ + 403, + 246 + ], + [ + 397, + 237 + ], + [ + 402, + 226 + ] + ], + [ + [ + 376, + 277 + ], + [ + 385, + 290 + ], + [ + 375, + 292 + ], + [ + 376, + 260 + ], + [ + 366, + 280 + ], + [ + 367, + 248 + ] + ] + ] + }, + { + "task_id": "0003765", + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_036422060215#235", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003765/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 285 + ], + [ + 369, + 268 + ], + [ + 385, + 286 + ], + [ + 341, + 284 + ], + [ + 378, + 302 + ], + [ + 336, + 300 + ] + ], + [ + [ + 379, + 283 + ], + [ + 367, + 264 + ], + [ + 388, + 283 + ], + [ + 370, + 284 + ], + [ + 400, + 302 + ], + [ + 383, + 305 + ] + ], + [ + [ + 364, + 258 + ], + [ + 376, + 274 + ], + [ + 378, + 241 + ], + [ + 349, + 275 + ], + [ + 366, + 227 + ], + [ + 338, + 260 + ] + ], + [ + [ + 369, + 299 + ], + [ + 360, + 318 + ], + [ + 362, + 284 + ], + [ + 375, + 313 + ], + [ + 371, + 265 + ], + [ + 383, + 295 + ] + ] + ] + }, + { + "task_id": "0003769", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104422070044#524", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003769/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 420, + 138 + ], + [ + 435, + 127 + ], + [ + 426, + 140 + ], + [ + 413, + 136 + ], + [ + 412, + 151 + ], + [ + 398, + 147 + ] + ], + [ + [ + 424, + 153 + ], + [ + 409, + 144 + ], + [ + 416, + 170 + ], + [ + 433, + 136 + ], + [ + 432, + 178 + ], + [ + 448, + 145 + ] + ], + [ + [ + 416, + 141 + ], + [ + 425, + 134 + ], + [ + 434, + 147 + ], + [ + 399, + 134 + ], + [ + 425, + 154 + ], + [ + 389, + 141 + ] + ], + [ + [ + 429, + 138 + ], + [ + 428, + 125 + ], + [ + 435, + 151 + ], + [ + 423, + 126 + ], + [ + 436, + 164 + ], + [ + 424, + 138 + ] + ] + ] + }, + { + "task_id": "0003770", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104422070044#767", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003770/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 333, + 137 + ], + [ + 322, + 132 + ], + [ + 345, + 129 + ], + [ + 319, + 146 + ], + [ + 357, + 135 + ], + [ + 331, + 152 + ] + ], + [ + [ + 328, + 148 + ], + [ + 324, + 136 + ], + [ + 320, + 160 + ], + [ + 336, + 136 + ], + [ + 324, + 170 + ], + [ + 339, + 147 + ] + ], + [ + [ + 327, + 142 + ], + [ + 330, + 127 + ], + [ + 340, + 150 + ], + [ + 314, + 134 + ], + [ + 338, + 165 + ], + [ + 310, + 149 + ] + ], + [ + [ + 314, + 155 + ], + [ + 300, + 152 + ], + [ + 314, + 170 + ], + [ + 313, + 140 + ], + [ + 329, + 174 + ], + [ + 327, + 144 + ] + ] + ] + }, + { + "task_id": "0003772", + "id": "RH20T_cfg1_task_0105_user_0010_scene_0005_cfg_0001_039422060546#79", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003772/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 340, + 215 + ], + [ + 322, + 210 + ], + [ + 310, + 221 + ], + [ + 367, + 209 + ], + [ + 330, + 227 + ], + [ + 389, + 214 + ] + ], + [ + [ + 337, + 233 + ], + [ + 315, + 206 + ], + [ + 345, + 219 + ], + [ + 330, + 244 + ], + [ + 370, + 250 + ], + [ + 352, + 271 + ] + ], + [ + [ + 317, + 222 + ], + [ + 347, + 218 + ], + [ + 339, + 232 + ], + [ + 292, + 211 + ], + [ + 313, + 235 + ], + [ + 265, + 215 + ] + ], + [ + [ + 335, + 232 + ], + [ + 336, + 212 + ], + [ + 336, + 265 + ], + [ + 334, + 203 + ], + [ + 335, + 282 + ], + [ + 333, + 222 + ] + ] + ] + }, + { + "task_id": "0003773", + "id": "RH20T_cfg5_task_0016_user_0010_scene_0001_cfg_0005_104122063678#298", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003773/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 222 + ], + [ + 270, + 243 + ], + [ + 279, + 213 + ], + [ + 292, + 231 + ], + [ + 293, + 193 + ], + [ + 308, + 209 + ] + ], + [ + [ + 303, + 193 + ], + [ + 328, + 196 + ], + [ + 295, + 212 + ], + [ + 312, + 175 + ], + [ + 270, + 209 + ], + [ + 288, + 173 + ] + ], + [ + [ + 288, + 213 + ], + [ + 293, + 195 + ], + [ + 293, + 235 + ], + [ + 283, + 192 + ], + [ + 288, + 250 + ], + [ + 278, + 208 + ] + ], + [ + [ + 288, + 220 + ], + [ + 262, + 217 + ], + [ + 287, + 246 + ], + [ + 289, + 195 + ], + [ + 313, + 248 + ], + [ + 314, + 198 + ] + ] + ] + }, + { + "task_id": "0003774", + "id": "RH20T_cfg2_task_0041_user_0007_scene_0009_cfg_0002_036422060215#496", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the tabletop with a sponge'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003774/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 418, + 252 + ], + [ + 424, + 238 + ], + [ + 432, + 249 + ], + [ + 404, + 255 + ], + [ + 427, + 262 + ], + [ + 399, + 268 + ] + ], + [ + [ + 420, + 244 + ], + [ + 436, + 245 + ], + [ + 430, + 245 + ], + [ + 411, + 244 + ], + [ + 413, + 244 + ], + [ + 395, + 243 + ] + ], + [ + [ + 423, + 242 + ], + [ + 432, + 253 + ], + [ + 427, + 253 + ], + [ + 418, + 230 + ], + [ + 419, + 242 + ], + [ + 409, + 220 + ] + ], + [ + [ + 406, + 254 + ], + [ + 420, + 261 + ], + [ + 417, + 250 + ], + [ + 393, + 258 + ], + [ + 404, + 243 + ], + [ + 380, + 250 + ] + ] + ] + }, + { + "task_id": "0003775", + "id": "RH20T_cfg6_task_0038_user_0020_scene_0001_cfg_0006_104122061602#573", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003775/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 443, + 139 + ], + [ + 424, + 140 + ], + [ + 458, + 128 + ], + [ + 429, + 149 + ], + [ + 477, + 128 + ], + [ + 447, + 148 + ] + ], + [ + [ + 440, + 159 + ], + [ + 453, + 162 + ], + [ + 444, + 133 + ], + [ + 437, + 184 + ], + [ + 432, + 132 + ], + [ + 426, + 180 + ] + ], + [ + [ + 449, + 141 + ], + [ + 466, + 153 + ], + [ + 446, + 161 + ], + [ + 452, + 119 + ], + [ + 430, + 150 + ], + [ + 435, + 109 + ] + ], + [ + [ + 457, + 136 + ], + [ + 435, + 121 + ], + [ + 460, + 133 + ], + [ + 455, + 139 + ], + [ + 484, + 150 + ], + [ + 476, + 154 + ] + ] + ] + }, + { + "task_id": "0003779", + "id": "RH20T_cfg4_task_0032_user_0016_scene_0002_cfg_0004_104122062295#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003779/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 181, + 200 + ], + [ + 173, + 232 + ], + [ + 214, + 160 + ], + [ + 150, + 239 + ], + [ + 217, + 138 + ], + [ + 160, + 210 + ] + ], + [ + [ + 210, + 213 + ], + [ + 189, + 156 + ], + [ + 168, + 226 + ], + [ + 257, + 198 + ], + [ + 188, + 278 + ], + [ + 279, + 257 + ] + ], + [ + [ + 206, + 229 + ], + [ + 233, + 189 + ], + [ + 156, + 205 + ], + [ + 253, + 251 + ], + [ + 132, + 245 + ], + [ + 228, + 286 + ] + ], + [ + [ + 206, + 202 + ], + [ + 182, + 165 + ], + [ + 185, + 256 + ], + [ + 223, + 156 + ], + [ + 208, + 283 + ], + [ + 239, + 189 + ] + ] + ] + }, + { + "task_id": "0003780", + "id": "RH20T_cfg4_task_0032_user_0016_scene_0002_cfg_0004_104122062295#186", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003780/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 238, + 221 + ], + [ + 285, + 236 + ], + [ + 210, + 219 + ], + [ + 273, + 223 + ], + [ + 158, + 202 + ], + [ + 214, + 203 + ] + ], + [ + [ + 212, + 220 + ], + [ + 215, + 214 + ], + [ + 233, + 281 + ], + [ + 190, + 158 + ], + [ + 232, + 298 + ], + [ + 182, + 154 + ] + ], + [ + [ + 169, + 219 + ], + [ + 202, + 178 + ], + [ + 114, + 194 + ], + [ + 218, + 242 + ], + [ + 83, + 235 + ], + [ + 187, + 277 + ] + ], + [ + [ + 168, + 205 + ], + [ + 196, + 243 + ], + [ + 171, + 141 + ], + [ + 165, + 261 + ], + [ + 146, + 115 + ], + [ + 143, + 227 + ] + ] + ] + }, + { + "task_id": "0003785", + "id": "RH20T_cfg6_task_0011_user_0014_scene_0007_cfg_0006_104122061330#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003785/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 248, + 123 + ], + [ + 235, + 134 + ], + [ + 257, + 101 + ], + [ + 240, + 148 + ], + [ + 271, + 88 + ], + [ + 254, + 136 + ] + ], + [ + [ + 274, + 143 + ], + [ + 258, + 125 + ], + [ + 250, + 149 + ], + [ + 300, + 136 + ], + [ + 266, + 165 + ], + [ + 314, + 154 + ] + ], + [ + [ + 264, + 142 + ], + [ + 266, + 165 + ], + [ + 294, + 145 + ], + [ + 235, + 138 + ], + [ + 294, + 119 + ], + [ + 232, + 113 + ] + ], + [ + [ + 271, + 163 + ], + [ + 261, + 149 + ], + [ + 257, + 188 + ], + [ + 285, + 139 + ], + [ + 267, + 199 + ], + [ + 293, + 152 + ] + ] + ] + }, + { + "task_id": "0003786", + "id": "RH20T_cfg4_task_0006_user_0010_scene_0008_cfg_0004_104122062295#135", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003786/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 0, + 73 + ], + [ + 0, + 62 + ], + [ + 35, + 91 + ], + [ + 0, + 54 + ], + [ + 39, + 104 + ], + [ + 0, + 66 + ] + ], + [ + [ + 34, + 142 + ], + [ + 1, + 165 + ], + [ + 48, + 168 + ], + [ + 18, + 113 + ], + [ + 76, + 148 + ], + [ + 50, + 93 + ] + ], + [ + [ + 0, + 126 + ], + [ + 28, + 124 + ], + [ + 0, + 137 + ], + [ + 4, + 117 + ], + [ + 0, + 140 + ], + [ + 0, + 120 + ] + ], + [ + [ + 0, + 111 + ], + [ + 0, + 78 + ], + [ + 0, + 93 + ], + [ + 13, + 131 + ], + [ + 0, + 123 + ], + [ + 19, + 161 + ] + ] + ] + }, + { + "task_id": "0003789", + "id": "RH20T_cfg3_task_0071_user_0010_scene_0008_cfg_0003_104122063550#94", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003789/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 214, + 216 + ], + [ + 206, + 211 + ], + [ + 216, + 243 + ], + [ + 212, + 190 + ], + [ + 223, + 247 + ], + [ + 219, + 196 + ] + ], + [ + [ + 214, + 206 + ], + [ + 201, + 213 + ], + [ + 234, + 220 + ], + [ + 193, + 191 + ], + [ + 248, + 212 + ], + [ + 207, + 182 + ] + ], + [ + [ + 238, + 195 + ], + [ + 244, + 176 + ], + [ + 243, + 178 + ], + [ + 233, + 214 + ], + [ + 238, + 195 + ], + [ + 228, + 232 + ] + ], + [ + [ + 223, + 212 + ], + [ + 238, + 210 + ], + [ + 199, + 219 + ], + [ + 245, + 206 + ], + [ + 185, + 221 + ], + [ + 231, + 208 + ] + ] + ] + }, + { + "task_id": "0003790", + "id": "RH20T_cfg3_task_0071_user_0010_scene_0008_cfg_0003_104122063550#108", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003790/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 195, + 227 + ], + [ + 213, + 242 + ], + [ + 189, + 247 + ], + [ + 202, + 206 + ], + [ + 172, + 232 + ], + [ + 184, + 191 + ] + ], + [ + [ + 192, + 232 + ], + [ + 198, + 256 + ], + [ + 175, + 223 + ], + [ + 208, + 240 + ], + [ + 169, + 199 + ], + [ + 202, + 218 + ] + ], + [ + [ + 213, + 221 + ], + [ + 207, + 216 + ], + [ + 216, + 248 + ], + [ + 209, + 195 + ], + [ + 222, + 251 + ], + [ + 215, + 201 + ] + ], + [ + [ + 232, + 236 + ], + [ + 204, + 243 + ], + [ + 225, + 230 + ], + [ + 239, + 242 + ], + [ + 255, + 223 + ], + [ + 265, + 236 + ] + ] + ] + }, + { + "task_id": "0003792", + "id": "RH20T_cfg6_task_0047_user_0014_scene_0002_cfg_0006_104122064161#163", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003792/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 372, + 290 + ], + [ + 390, + 315 + ], + [ + 348, + 307 + ], + [ + 397, + 273 + ], + [ + 332, + 282 + ], + [ + 379, + 249 + ] + ], + [ + [ + 372, + 281 + ], + [ + 375, + 254 + ], + [ + 377, + 301 + ], + [ + 368, + 263 + ], + [ + 374, + 327 + ], + [ + 365, + 288 + ] + ], + [ + [ + 358, + 314 + ], + [ + 343, + 299 + ], + [ + 336, + 301 + ], + [ + 379, + 325 + ], + [ + 353, + 317 + ], + [ + 397, + 341 + ] + ], + [ + [ + 391, + 261 + ], + [ + 399, + 247 + ], + [ + 372, + 235 + ], + [ + 411, + 288 + ], + [ + 367, + 249 + ], + [ + 403, + 299 + ] + ] + ] + }, + { + "task_id": "0003794", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122063633#705", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003794/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 192, + 149 + ], + [ + 214, + 145 + ], + [ + 188, + 128 + ], + [ + 196, + 171 + ], + [ + 167, + 132 + ], + [ + 175, + 175 + ] + ], + [ + [ + 192, + 156 + ], + [ + 194, + 144 + ], + [ + 192, + 174 + ], + [ + 193, + 136 + ], + [ + 190, + 188 + ], + [ + 190, + 148 + ] + ], + [ + [ + 205, + 149 + ], + [ + 214, + 128 + ], + [ + 228, + 158 + ], + [ + 182, + 140 + ], + [ + 219, + 179 + ], + [ + 172, + 162 + ] + ], + [ + [ + 183, + 140 + ], + [ + 159, + 140 + ], + [ + 189, + 160 + ], + [ + 177, + 121 + ], + [ + 211, + 159 + ], + [ + 198, + 122 + ] + ] + ] + }, + { + "task_id": "0003795", + "id": "RH20T_cfg6_task_0034_user_0020_scene_0008_cfg_0006_104122063633#1228", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003795/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 131, + 245 + ], + [ + 115, + 224 + ], + [ + 154, + 227 + ], + [ + 106, + 263 + ], + [ + 170, + 248 + ], + [ + 122, + 285 + ] + ], + [ + [ + 152, + 259 + ], + [ + 132, + 270 + ], + [ + 140, + 234 + ], + [ + 164, + 284 + ], + [ + 160, + 225 + ], + [ + 182, + 272 + ] + ], + [ + [ + 129, + 286 + ], + [ + 121, + 301 + ], + [ + 135, + 259 + ], + [ + 122, + 312 + ], + [ + 141, + 248 + ], + [ + 128, + 298 + ] + ], + [ + [ + 141, + 261 + ], + [ + 147, + 233 + ], + [ + 149, + 264 + ], + [ + 132, + 257 + ], + [ + 143, + 291 + ], + [ + 125, + 287 + ] + ] + ] + }, + { + "task_id": "0003800", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0005_cfg_0006_104122063633#167", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003800/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 201, + 169 + ], + [ + 200, + 181 + ], + [ + 171, + 171 + ], + [ + 230, + 168 + ], + [ + 170, + 157 + ], + [ + 232, + 155 + ] + ], + [ + [ + 169, + 178 + ], + [ + 150, + 180 + ], + [ + 154, + 156 + ], + [ + 184, + 201 + ], + [ + 172, + 156 + ], + [ + 201, + 199 + ] + ], + [ + [ + 195, + 191 + ], + [ + 168, + 180 + ], + [ + 187, + 190 + ], + [ + 201, + 191 + ], + [ + 216, + 201 + ], + [ + 226, + 200 + ] + ], + [ + [ + 171, + 194 + ], + [ + 173, + 166 + ], + [ + 180, + 185 + ], + [ + 160, + 205 + ], + [ + 178, + 211 + ], + [ + 159, + 234 + ] + ] + ] + }, + { + "task_id": "0003804", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522062547#436", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003804/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 380, + 131 + ], + [ + 376, + 121 + ], + [ + 399, + 144 + ], + [ + 361, + 117 + ], + [ + 404, + 156 + ], + [ + 364, + 128 + ] + ], + [ + [ + 365, + 112 + ], + [ + 359, + 114 + ], + [ + 389, + 101 + ], + [ + 342, + 124 + ], + [ + 393, + 100 + ], + [ + 349, + 122 + ] + ], + [ + [ + 351, + 151 + ], + [ + 370, + 149 + ], + [ + 358, + 174 + ], + [ + 344, + 127 + ], + [ + 341, + 175 + ], + [ + 327, + 130 + ] + ], + [ + [ + 365, + 128 + ], + [ + 364, + 117 + ], + [ + 375, + 150 + ], + [ + 354, + 107 + ], + [ + 375, + 159 + ], + [ + 355, + 118 + ] + ] + ] + }, + { + "task_id": "0003805", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522062547#652", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003805/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 368, + 192 + ], + [ + 368, + 182 + ], + [ + 393, + 202 + ], + [ + 343, + 183 + ], + [ + 391, + 212 + ], + [ + 343, + 193 + ] + ], + [ + [ + 366, + 172 + ], + [ + 379, + 159 + ], + [ + 355, + 184 + ], + [ + 379, + 158 + ], + [ + 341, + 198 + ], + [ + 365, + 172 + ] + ], + [ + [ + 354, + 177 + ], + [ + 371, + 164 + ], + [ + 365, + 202 + ], + [ + 343, + 152 + ], + [ + 347, + 216 + ], + [ + 325, + 165 + ] + ], + [ + [ + 360, + 188 + ], + [ + 340, + 197 + ], + [ + 372, + 212 + ], + [ + 348, + 164 + ], + [ + 394, + 203 + ], + [ + 369, + 153 + ] + ] + ] + }, + { + "task_id": "0003806", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0010_cfg_0007_037522061512#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003806/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 205, + 158 + ], + [ + 195, + 145 + ], + [ + 225, + 161 + ], + [ + 186, + 155 + ], + [ + 234, + 173 + ], + [ + 197, + 167 + ] + ], + [ + [ + 224, + 130 + ], + [ + 206, + 130 + ], + [ + 207, + 137 + ], + [ + 240, + 124 + ], + [ + 227, + 137 + ], + [ + 259, + 124 + ] + ], + [ + [ + 205, + 162 + ], + [ + 185, + 150 + ], + [ + 201, + 174 + ], + [ + 209, + 149 + ], + [ + 221, + 185 + ], + [ + 230, + 161 + ] + ], + [ + [ + 213, + 148 + ], + [ + 227, + 151 + ], + [ + 206, + 170 + ], + [ + 220, + 125 + ], + [ + 191, + 167 + ], + [ + 206, + 120 + ] + ] + ] + }, + { + "task_id": "0003808", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0010_cfg_0007_037522061512#407", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003808/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 213, + 149 + ], + [ + 224, + 158 + ], + [ + 202, + 170 + ], + [ + 223, + 129 + ], + [ + 189, + 161 + ], + [ + 211, + 118 + ] + ], + [ + [ + 227, + 160 + ], + [ + 207, + 152 + ], + [ + 215, + 176 + ], + [ + 239, + 146 + ], + [ + 236, + 184 + ], + [ + 259, + 153 + ] + ], + [ + [ + 215, + 148 + ], + [ + 203, + 134 + ], + [ + 206, + 168 + ], + [ + 223, + 130 + ], + [ + 217, + 180 + ], + [ + 233, + 144 + ] + ], + [ + [ + 203, + 126 + ], + [ + 225, + 127 + ], + [ + 197, + 106 + ], + [ + 209, + 146 + ], + [ + 175, + 106 + ], + [ + 188, + 145 + ] + ] + ] + }, + { + "task_id": "0003809", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0010_cfg_0007_037522061512#553", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003809/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 204, + 145 + ], + [ + 189, + 134 + ], + [ + 208, + 159 + ], + [ + 199, + 129 + ], + [ + 223, + 170 + ], + [ + 215, + 140 + ] + ], + [ + [ + 202, + 157 + ], + [ + 182, + 146 + ], + [ + 190, + 177 + ], + [ + 214, + 137 + ], + [ + 210, + 188 + ], + [ + 234, + 147 + ] + ], + [ + [ + 211, + 166 + ], + [ + 206, + 184 + ], + [ + 226, + 156 + ], + [ + 194, + 177 + ], + [ + 231, + 138 + ], + [ + 198, + 158 + ] + ], + [ + [ + 207, + 147 + ], + [ + 217, + 126 + ], + [ + 220, + 152 + ], + [ + 194, + 142 + ], + [ + 211, + 172 + ], + [ + 183, + 165 + ] + ] + ] + }, + { + "task_id": "0003811", + "id": "RH20T_cfg3_task_0023_user_0010_scene_0003_cfg_0003_104122062823#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003811/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 295, + 65 + ], + [ + 310, + 39 + ], + [ + 303, + 57 + ], + [ + 286, + 75 + ], + [ + 290, + 81 + ], + [ + 272, + 101 + ] + ], + [ + [ + 297, + 96 + ], + [ + 318, + 80 + ], + [ + 319, + 94 + ], + [ + 273, + 99 + ], + [ + 299, + 108 + ], + [ + 254, + 114 + ] + ], + [ + [ + 309, + 121 + ], + [ + 288, + 101 + ], + [ + 286, + 143 + ], + [ + 331, + 100 + ], + [ + 307, + 164 + ], + [ + 353, + 120 + ] + ], + [ + [ + 290, + 97 + ], + [ + 292, + 69 + ], + [ + 321, + 101 + ], + [ + 260, + 92 + ], + [ + 318, + 126 + ], + [ + 259, + 117 + ] + ] + ] + }, + { + "task_id": "0003813", + "id": "RH20T_cfg7_task_0028_user_0020_scene_0003_cfg_0007_104122061018#187", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and close the drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003813/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 291, + 180 + ], + [ + 316, + 161 + ], + [ + 277, + 153 + ], + [ + 305, + 205 + ], + [ + 249, + 173 + ], + [ + 280, + 225 + ] + ], + [ + [ + 309, + 165 + ], + [ + 333, + 177 + ], + [ + 325, + 163 + ], + [ + 294, + 167 + ], + [ + 298, + 150 + ], + [ + 269, + 155 + ] + ], + [ + [ + 297, + 200 + ], + [ + 284, + 169 + ], + [ + 294, + 201 + ], + [ + 301, + 200 + ], + [ + 307, + 229 + ], + [ + 315, + 233 + ] + ], + [ + [ + 293, + 181 + ], + [ + 319, + 172 + ], + [ + 269, + 164 + ], + [ + 316, + 196 + ], + [ + 246, + 174 + ], + [ + 293, + 204 + ] + ] + ] + }, + { + "task_id": "0003815", + "id": "RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_037522061512#150", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003815/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 182, + 197 + ], + [ + 180, + 223 + ], + [ + 164, + 190 + ], + [ + 198, + 205 + ], + [ + 166, + 163 + ], + [ + 200, + 181 + ] + ], + [ + [ + 138, + 165 + ], + [ + 124, + 177 + ], + [ + 161, + 154 + ], + [ + 117, + 176 + ], + [ + 173, + 144 + ], + [ + 130, + 165 + ] + ], + [ + [ + 146, + 190 + ], + [ + 170, + 204 + ], + [ + 161, + 166 + ], + [ + 132, + 214 + ], + [ + 136, + 152 + ], + [ + 107, + 200 + ] + ], + [ + [ + 162, + 185 + ], + [ + 183, + 179 + ], + [ + 160, + 209 + ], + [ + 164, + 160 + ], + [ + 138, + 217 + ], + [ + 142, + 165 + ] + ] + ] + }, + { + "task_id": "0003816", + "id": "RH20T_cfg5_task_0044_user_0010_scene_0001_cfg_0005_104422070044#318", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003816/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 502, + 157 + ], + [ + 495, + 154 + ], + [ + 497, + 176 + ], + [ + 508, + 136 + ], + [ + 505, + 178 + ], + [ + 515, + 139 + ] + ], + [ + [ + 508, + 170 + ], + [ + 509, + 153 + ], + [ + 520, + 181 + ], + [ + 496, + 160 + ], + [ + 518, + 198 + ], + [ + 495, + 176 + ] + ], + [ + [ + 501, + 151 + ], + [ + 498, + 133 + ], + [ + 480, + 154 + ], + [ + 523, + 148 + ], + [ + 483, + 171 + ], + [ + 526, + 165 + ] + ], + [ + [ + 488, + 162 + ], + [ + 481, + 166 + ], + [ + 504, + 175 + ], + [ + 471, + 149 + ], + [ + 513, + 172 + ], + [ + 477, + 144 + ] + ] + ] + }, + { + "task_id": "0003817", + "id": "RH20T_cfg1_task_0108_user_0014_scene_0004_cfg_0001_038522062547#130", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Sharpen the pencil with a pencil sharpener'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003817/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 396, + 92 + ], + [ + 375, + 106 + ], + [ + 401, + 92 + ], + [ + 389, + 91 + ], + [ + 422, + 78 + ], + [ + 412, + 75 + ] + ], + [ + [ + 393, + 107 + ], + [ + 403, + 113 + ], + [ + 386, + 84 + ], + [ + 400, + 130 + ], + [ + 375, + 77 + ], + [ + 389, + 125 + ] + ], + [ + [ + 397, + 115 + ], + [ + 401, + 142 + ], + [ + 375, + 121 + ], + [ + 421, + 108 + ], + [ + 372, + 96 + ], + [ + 418, + 80 + ] + ], + [ + [ + 398, + 91 + ], + [ + 402, + 75 + ], + [ + 420, + 105 + ], + [ + 376, + 77 + ], + [ + 415, + 119 + ], + [ + 373, + 91 + ] + ] + ] + }, + { + "task_id": "0003818", + "id": "RH20T_cfg1_task_0015_user_0008_scene_0002_cfg_0001_819112071037#116", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003818/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 148 + ], + [ + 359, + 178 + ], + [ + 301, + 104 + ], + [ + 391, + 187 + ], + [ + 297, + 85 + ], + [ + 378, + 161 + ] + ], + [ + [ + 379, + 141 + ], + [ + 389, + 75 + ], + [ + 330, + 113 + ], + [ + 438, + 175 + ], + [ + 326, + 167 + ], + [ + 424, + 232 + ] + ], + [ + [ + 392, + 169 + ], + [ + 398, + 109 + ], + [ + 330, + 172 + ], + [ + 452, + 167 + ], + [ + 329, + 225 + ], + [ + 444, + 220 + ] + ], + [ + [ + 403, + 201 + ], + [ + 404, + 221 + ], + [ + 470, + 186 + ], + [ + 339, + 215 + ], + [ + 461, + 172 + ], + [ + 346, + 198 + ] + ] + ] + }, + { + "task_id": "0003823", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_037522062165#1013", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003823/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 351, + 335 + ], + [ + 364, + 313 + ], + [ + 358, + 315 + ], + [ + 344, + 354 + ], + [ + 344, + 340 + ], + [ + 330, + 360 + ] + ], + [ + [ + 329, + 342 + ], + [ + 335, + 360 + ], + [ + 313, + 360 + ], + [ + 344, + 323 + ], + [ + 307, + 334 + ], + [ + 339, + 298 + ] + ], + [ + [ + 364, + 314 + ], + [ + 352, + 315 + ], + [ + 378, + 286 + ], + [ + 349, + 343 + ], + [ + 387, + 287 + ], + [ + 360, + 341 + ] + ], + [ + [ + 343, + 319 + ], + [ + 347, + 306 + ], + [ + 322, + 297 + ], + [ + 365, + 343 + ], + [ + 319, + 310 + ], + [ + 360, + 354 + ] + ] + ] + }, + { + "task_id": "0003824", + "id": "RH20T_cfg5_task_0039_user_0007_scene_0007_cfg_0005_104122063678#299", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003824/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 322, + 147 + ], + [ + 326, + 130 + ], + [ + 320, + 163 + ], + [ + 323, + 132 + ], + [ + 316, + 178 + ], + [ + 318, + 147 + ] + ], + [ + [ + 344, + 149 + ], + [ + 353, + 169 + ], + [ + 359, + 148 + ], + [ + 328, + 150 + ], + [ + 351, + 130 + ], + [ + 319, + 130 + ] + ], + [ + [ + 340, + 147 + ], + [ + 318, + 144 + ], + [ + 343, + 125 + ], + [ + 338, + 169 + ], + [ + 365, + 128 + ], + [ + 360, + 172 + ] + ], + [ + [ + 301, + 130 + ], + [ + 284, + 142 + ], + [ + 304, + 142 + ], + [ + 299, + 118 + ], + [ + 322, + 129 + ], + [ + 315, + 106 + ] + ] + ] + }, + { + "task_id": "0003826", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0008_cfg_0001_038522063145#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003826/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 456, + 150 + ], + [ + 463, + 132 + ], + [ + 478, + 164 + ], + [ + 435, + 136 + ], + [ + 470, + 179 + ], + [ + 429, + 152 + ] + ], + [ + [ + 460, + 145 + ], + [ + 442, + 161 + ], + [ + 475, + 158 + ], + [ + 445, + 133 + ], + [ + 495, + 140 + ], + [ + 462, + 116 + ] + ], + [ + [ + 435, + 136 + ], + [ + 461, + 131 + ], + [ + 435, + 160 + ], + [ + 435, + 112 + ], + [ + 411, + 164 + ], + [ + 411, + 116 + ] + ], + [ + [ + 446, + 126 + ], + [ + 430, + 145 + ], + [ + 432, + 112 + ], + [ + 461, + 140 + ], + [ + 448, + 92 + ], + [ + 478, + 120 + ] + ] + ] + }, + { + "task_id": "0003828", + "id": "RH20T_cfg6_task_0130_user_0014_scene_0004_cfg_0006_104122064161#201", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003828/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 174 + ], + [ + 430, + 177 + ], + [ + 404, + 197 + ], + [ + 409, + 151 + ], + [ + 380, + 194 + ], + [ + 385, + 148 + ] + ], + [ + [ + 383, + 159 + ], + [ + 383, + 155 + ], + [ + 385, + 182 + ], + [ + 382, + 137 + ], + [ + 385, + 185 + ], + [ + 382, + 142 + ] + ], + [ + [ + 397, + 158 + ], + [ + 379, + 145 + ], + [ + 387, + 158 + ], + [ + 406, + 157 + ], + [ + 406, + 172 + ], + [ + 424, + 170 + ] + ], + [ + [ + 395, + 169 + ], + [ + 415, + 162 + ], + [ + 392, + 149 + ], + [ + 398, + 189 + ], + [ + 371, + 156 + ], + [ + 378, + 196 + ] + ] + ] + }, + { + "task_id": "0003829", + "id": "RH20T_cfg6_task_0130_user_0014_scene_0004_cfg_0006_104122064161#314", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003829/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 360, + 179 + ], + [ + 366, + 183 + ], + [ + 356, + 204 + ], + [ + 365, + 154 + ], + [ + 350, + 199 + ], + [ + 359, + 151 + ] + ], + [ + [ + 361, + 165 + ], + [ + 354, + 149 + ], + [ + 385, + 150 + ], + [ + 339, + 179 + ], + [ + 394, + 167 + ], + [ + 346, + 197 + ] + ], + [ + [ + 358, + 178 + ], + [ + 381, + 182 + ], + [ + 368, + 190 + ], + [ + 350, + 168 + ], + [ + 343, + 186 + ], + [ + 327, + 164 + ] + ], + [ + [ + 358, + 189 + ], + [ + 338, + 179 + ], + [ + 374, + 177 + ], + [ + 342, + 202 + ], + [ + 396, + 187 + ], + [ + 365, + 214 + ] + ] + ] + }, + { + "task_id": "0003834", + "id": "RH20T_cfg7_task_0037_user_0020_scene_0006_cfg_0007_104122061018#555", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003834/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 342, + 171 + ], + [ + 336, + 135 + ], + [ + 325, + 165 + ], + [ + 361, + 179 + ], + [ + 330, + 197 + ], + [ + 366, + 215 + ] + ], + [ + [ + 307, + 187 + ], + [ + 315, + 201 + ], + [ + 281, + 166 + ], + [ + 331, + 207 + ], + [ + 276, + 155 + ], + [ + 323, + 194 + ] + ], + [ + [ + 336, + 194 + ], + [ + 370, + 182 + ], + [ + 331, + 186 + ], + [ + 342, + 202 + ], + [ + 300, + 197 + ], + [ + 304, + 215 + ] + ], + [ + [ + 326, + 212 + ], + [ + 303, + 191 + ], + [ + 352, + 199 + ], + [ + 298, + 227 + ], + [ + 377, + 220 + ], + [ + 324, + 252 + ] + ] + ] + }, + { + "task_id": "0003835", + "id": "RH20T_cfg2_task_0066_user_0014_scene_0006_cfg_0002_036422060215#200", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003835/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 329, + 128 + ], + [ + 319, + 141 + ], + [ + 321, + 115 + ], + [ + 338, + 141 + ], + [ + 332, + 102 + ], + [ + 348, + 128 + ] + ], + [ + [ + 339, + 133 + ], + [ + 342, + 120 + ], + [ + 354, + 128 + ], + [ + 324, + 137 + ], + [ + 351, + 140 + ], + [ + 322, + 149 + ] + ], + [ + [ + 335, + 142 + ], + [ + 328, + 155 + ], + [ + 318, + 136 + ], + [ + 351, + 147 + ], + [ + 325, + 122 + ], + [ + 358, + 133 + ] + ], + [ + [ + 337, + 133 + ], + [ + 335, + 150 + ], + [ + 322, + 128 + ], + [ + 352, + 137 + ], + [ + 324, + 111 + ], + [ + 355, + 120 + ] + ] + ] + }, + { + "task_id": "0003837", + "id": "RH20T_cfg2_task_0066_user_0014_scene_0006_cfg_0002_036422060215#938", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003837/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 367, + 176 + ], + [ + 371, + 169 + ], + [ + 363, + 160 + ], + [ + 371, + 193 + ], + [ + 360, + 166 + ], + [ + 367, + 199 + ] + ], + [ + [ + 372, + 144 + ], + [ + 354, + 142 + ], + [ + 371, + 161 + ], + [ + 373, + 126 + ], + [ + 390, + 163 + ], + [ + 392, + 128 + ] + ], + [ + [ + 364, + 162 + ], + [ + 369, + 148 + ], + [ + 375, + 155 + ], + [ + 352, + 170 + ], + [ + 371, + 168 + ], + [ + 348, + 183 + ] + ], + [ + [ + 351, + 153 + ], + [ + 334, + 150 + ], + [ + 359, + 144 + ], + [ + 341, + 162 + ], + [ + 376, + 147 + ], + [ + 358, + 165 + ] + ] + ] + }, + { + "task_id": "0003841", + "id": "RH20T_cfg6_task_0091_user_0020_scene_0003_cfg_0006_104122061330#197", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003841/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 191 + ], + [ + 336, + 208 + ], + [ + 302, + 212 + ], + [ + 329, + 169 + ], + [ + 282, + 196 + ], + [ + 308, + 153 + ] + ], + [ + [ + 319, + 188 + ], + [ + 302, + 163 + ], + [ + 324, + 185 + ], + [ + 313, + 191 + ], + [ + 340, + 209 + ], + [ + 331, + 219 + ] + ], + [ + [ + 316, + 191 + ], + [ + 311, + 173 + ], + [ + 302, + 214 + ], + [ + 330, + 169 + ], + [ + 308, + 230 + ], + [ + 334, + 186 + ] + ], + [ + [ + 308, + 174 + ], + [ + 284, + 185 + ], + [ + 312, + 194 + ], + [ + 304, + 152 + ], + [ + 334, + 184 + ], + [ + 329, + 142 + ] + ] + ] + }, + { + "task_id": "0003844", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0003_cfg_0004_036422060909#93", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003844/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 294, + 134 + ], + [ + 294, + 115 + ], + [ + 313, + 123 + ], + [ + 274, + 145 + ], + [ + 312, + 140 + ], + [ + 274, + 162 + ] + ], + [ + [ + 291, + 154 + ], + [ + 304, + 172 + ], + [ + 313, + 145 + ], + [ + 269, + 163 + ], + [ + 300, + 129 + ], + [ + 257, + 146 + ] + ], + [ + [ + 271, + 135 + ], + [ + 268, + 110 + ], + [ + 247, + 138 + ], + [ + 295, + 132 + ], + [ + 250, + 162 + ], + [ + 298, + 156 + ] + ], + [ + [ + 306, + 105 + ], + [ + 289, + 95 + ], + [ + 321, + 84 + ], + [ + 292, + 126 + ], + [ + 339, + 95 + ], + [ + 310, + 138 + ] + ] + ] + }, + { + "task_id": "0003847", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0003_cfg_0004_036422060909#261", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003847/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 318, + 148 + ], + [ + 303, + 152 + ], + [ + 318, + 174 + ], + [ + 318, + 122 + ], + [ + 335, + 171 + ], + [ + 335, + 116 + ] + ], + [ + [ + 307, + 151 + ], + [ + 330, + 142 + ], + [ + 303, + 128 + ], + [ + 310, + 172 + ], + [ + 278, + 137 + ], + [ + 286, + 181 + ] + ], + [ + [ + 277, + 142 + ], + [ + 265, + 139 + ], + [ + 284, + 115 + ], + [ + 271, + 169 + ], + [ + 298, + 117 + ], + [ + 284, + 174 + ] + ], + [ + [ + 305, + 165 + ], + [ + 306, + 147 + ], + [ + 325, + 152 + ], + [ + 284, + 178 + ], + [ + 323, + 169 + ], + [ + 284, + 194 + ] + ] + ] + }, + { + "task_id": "0003848", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0003_cfg_0004_036422060909#298", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003848/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 234, + 128 + ], + [ + 229, + 105 + ], + [ + 214, + 139 + ], + [ + 255, + 117 + ], + [ + 219, + 163 + ], + [ + 261, + 142 + ] + ], + [ + [ + 217, + 149 + ], + [ + 212, + 137 + ], + [ + 223, + 171 + ], + [ + 211, + 126 + ], + [ + 229, + 185 + ], + [ + 217, + 138 + ] + ], + [ + [ + 243, + 147 + ], + [ + 245, + 131 + ], + [ + 266, + 141 + ], + [ + 218, + 153 + ], + [ + 264, + 155 + ], + [ + 217, + 167 + ] + ], + [ + [ + 249, + 162 + ], + [ + 250, + 142 + ], + [ + 268, + 173 + ], + [ + 228, + 150 + ], + [ + 268, + 193 + ], + [ + 226, + 171 + ] + ] + ] + }, + { + "task_id": "0003851", + "id": "RH20T_cfg7_task_0064_user_0020_scene_0009_cfg_0007_104122060811#163", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003851/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 322, + 134 + ], + [ + 319, + 147 + ], + [ + 329, + 141 + ], + [ + 316, + 127 + ], + [ + 333, + 127 + ], + [ + 320, + 114 + ] + ], + [ + [ + 307, + 136 + ], + [ + 318, + 138 + ], + [ + 298, + 144 + ], + [ + 317, + 128 + ], + [ + 287, + 142 + ], + [ + 305, + 125 + ] + ], + [ + [ + 287, + 141 + ], + [ + 287, + 126 + ], + [ + 272, + 142 + ], + [ + 303, + 140 + ], + [ + 273, + 156 + ], + [ + 303, + 155 + ] + ], + [ + [ + 311, + 144 + ], + [ + 313, + 138 + ], + [ + 297, + 151 + ], + [ + 324, + 138 + ], + [ + 295, + 157 + ], + [ + 321, + 144 + ] + ] + ] + }, + { + "task_id": "0003853", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061602#595", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003853/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 386, + 140 + ], + [ + 394, + 129 + ], + [ + 396, + 159 + ], + [ + 375, + 121 + ], + [ + 388, + 168 + ], + [ + 368, + 131 + ] + ], + [ + [ + 370, + 163 + ], + [ + 350, + 168 + ], + [ + 381, + 169 + ], + [ + 360, + 157 + ], + [ + 402, + 164 + ], + [ + 380, + 152 + ] + ], + [ + [ + 365, + 149 + ], + [ + 347, + 139 + ], + [ + 364, + 135 + ], + [ + 366, + 161 + ], + [ + 383, + 145 + ], + [ + 384, + 171 + ] + ], + [ + [ + 399, + 133 + ], + [ + 399, + 109 + ], + [ + 386, + 133 + ], + [ + 413, + 132 + ], + [ + 386, + 155 + ], + [ + 413, + 156 + ] + ] + ] + }, + { + "task_id": "0003854", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061602#868", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003854/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 444, + 171 + ], + [ + 444, + 166 + ], + [ + 454, + 192 + ], + [ + 434, + 149 + ], + [ + 454, + 198 + ], + [ + 434, + 153 + ] + ], + [ + [ + 430, + 152 + ], + [ + 445, + 154 + ], + [ + 434, + 174 + ], + [ + 425, + 129 + ], + [ + 421, + 171 + ], + [ + 412, + 128 + ] + ], + [ + [ + 422, + 157 + ], + [ + 446, + 157 + ], + [ + 422, + 136 + ], + [ + 423, + 179 + ], + [ + 399, + 137 + ], + [ + 399, + 179 + ] + ], + [ + [ + 433, + 159 + ], + [ + 445, + 152 + ], + [ + 418, + 141 + ], + [ + 449, + 178 + ], + [ + 405, + 148 + ], + [ + 436, + 186 + ] + ] + ] + }, + { + "task_id": "0003855", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061602#1134", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003855/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 497, + 183 + ], + [ + 500, + 164 + ], + [ + 474, + 171 + ], + [ + 519, + 195 + ], + [ + 470, + 192 + ], + [ + 517, + 216 + ] + ], + [ + [ + 459, + 182 + ], + [ + 447, + 175 + ], + [ + 482, + 168 + ], + [ + 436, + 196 + ], + [ + 496, + 174 + ], + [ + 447, + 205 + ] + ], + [ + [ + 476, + 193 + ], + [ + 484, + 175 + ], + [ + 494, + 211 + ], + [ + 459, + 176 + ], + [ + 485, + 227 + ], + [ + 453, + 192 + ] + ], + [ + [ + 453, + 206 + ], + [ + 468, + 191 + ], + [ + 469, + 227 + ], + [ + 438, + 187 + ], + [ + 455, + 241 + ], + [ + 426, + 201 + ] + ] + ] + }, + { + "task_id": "0003856", + "id": "RH20T_cfg5_task_0002_user_0014_scene_0003_cfg_0005_104122063678#219", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003856/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 354, + 157 + ], + [ + 353, + 167 + ], + [ + 369, + 174 + ], + [ + 338, + 139 + ], + [ + 369, + 163 + ], + [ + 339, + 130 + ] + ], + [ + [ + 346, + 182 + ], + [ + 349, + 166 + ], + [ + 348, + 203 + ], + [ + 343, + 164 + ], + [ + 345, + 218 + ], + [ + 341, + 179 + ] + ], + [ + [ + 326, + 185 + ], + [ + 301, + 194 + ], + [ + 319, + 193 + ], + [ + 333, + 176 + ], + [ + 342, + 186 + ], + [ + 358, + 168 + ] + ], + [ + [ + 325, + 191 + ], + [ + 351, + 189 + ], + [ + 324, + 183 + ], + [ + 327, + 198 + ], + [ + 296, + 184 + ], + [ + 302, + 199 + ] + ] + ] + }, + { + "task_id": "0003858", + "id": "RH20T_cfg4_task_0032_user_0016_scene_0006_cfg_0004_104122062295#165", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003858/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 235, + 257 + ], + [ + 266, + 235 + ], + [ + 266, + 265 + ], + [ + 207, + 249 + ], + [ + 230, + 292 + ], + [ + 171, + 272 + ] + ], + [ + [ + 239, + 218 + ], + [ + 230, + 172 + ], + [ + 202, + 219 + ], + [ + 271, + 217 + ], + [ + 212, + 273 + ], + [ + 280, + 261 + ] + ], + [ + [ + 253, + 222 + ], + [ + 283, + 236 + ], + [ + 249, + 268 + ], + [ + 257, + 181 + ], + [ + 212, + 254 + ], + [ + 224, + 163 + ] + ], + [ + [ + 272, + 248 + ], + [ + 275, + 207 + ], + [ + 243, + 233 + ], + [ + 304, + 265 + ], + [ + 241, + 270 + ], + [ + 300, + 306 + ] + ] + ] + }, + { + "task_id": "0003859", + "id": "RH20T_cfg4_task_0032_user_0016_scene_0006_cfg_0004_104122062295#268", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003859/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 354, + 292 + ], + [ + 354, + 247 + ], + [ + 326, + 275 + ], + [ + 387, + 312 + ], + [ + 327, + 315 + ], + [ + 386, + 358 + ] + ], + [ + [ + 321, + 284 + ], + [ + 294, + 327 + ], + [ + 355, + 304 + ], + [ + 284, + 262 + ], + [ + 380, + 264 + ], + [ + 312, + 219 + ] + ], + [ + [ + 365, + 285 + ], + [ + 399, + 263 + ], + [ + 397, + 319 + ], + [ + 337, + 255 + ], + [ + 359, + 346 + ], + [ + 302, + 276 + ] + ], + [ + [ + 366, + 304 + ], + [ + 410, + 306 + ], + [ + 362, + 282 + ], + [ + 371, + 330 + ], + [ + 322, + 279 + ], + [ + 323, + 327 + ] + ] + ] + }, + { + "task_id": "0003860", + "id": "RH20T_cfg5_task_0002_user_0010_scene_0001_cfg_0005_105422061350#145", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003860/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 241, + 178 + ], + [ + 234, + 169 + ], + [ + 268, + 164 + ], + [ + 214, + 193 + ], + [ + 278, + 174 + ], + [ + 221, + 205 + ] + ], + [ + [ + 222, + 135 + ], + [ + 230, + 138 + ], + [ + 253, + 133 + ], + [ + 192, + 137 + ], + [ + 247, + 130 + ], + [ + 181, + 133 + ] + ], + [ + [ + 237, + 141 + ], + [ + 227, + 144 + ], + [ + 249, + 114 + ], + [ + 227, + 168 + ], + [ + 257, + 113 + ], + [ + 236, + 164 + ] + ], + [ + [ + 253, + 159 + ], + [ + 256, + 157 + ], + [ + 242, + 185 + ], + [ + 263, + 133 + ], + [ + 239, + 190 + ], + [ + 261, + 133 + ] + ] + ] + }, + { + "task_id": "0003863", + "id": "RH20T_cfg1_task_0005_user_0004_scene_0007_cfg_0001_819112070044#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003863/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 125 + ], + [ + 332, + 109 + ], + [ + 343, + 123 + ], + [ + 308, + 127 + ], + [ + 337, + 138 + ], + [ + 302, + 143 + ] + ], + [ + [ + 325, + 132 + ], + [ + 316, + 118 + ], + [ + 343, + 125 + ], + [ + 307, + 139 + ], + [ + 353, + 139 + ], + [ + 316, + 154 + ] + ], + [ + [ + 328, + 114 + ], + [ + 326, + 99 + ], + [ + 330, + 99 + ], + [ + 326, + 127 + ], + [ + 332, + 114 + ], + [ + 328, + 142 + ] + ], + [ + [ + 349, + 119 + ], + [ + 351, + 112 + ], + [ + 334, + 133 + ], + [ + 363, + 105 + ], + [ + 332, + 140 + ], + [ + 362, + 111 + ] + ] + ] + }, + { + "task_id": "0003865", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_104122061850#410", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003865/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 370, + 183 + ], + [ + 347, + 174 + ], + [ + 368, + 202 + ], + [ + 372, + 162 + ], + [ + 391, + 212 + ], + [ + 398, + 172 + ] + ], + [ + [ + 346, + 208 + ], + [ + 330, + 197 + ], + [ + 347, + 183 + ], + [ + 345, + 231 + ], + [ + 364, + 193 + ], + [ + 361, + 244 + ] + ], + [ + [ + 337, + 197 + ], + [ + 352, + 222 + ], + [ + 349, + 196 + ], + [ + 326, + 199 + ], + [ + 333, + 170 + ], + [ + 311, + 176 + ] + ], + [ + [ + 373, + 197 + ], + [ + 368, + 183 + ], + [ + 359, + 220 + ], + [ + 386, + 174 + ], + [ + 364, + 231 + ], + [ + 389, + 188 + ] + ] + ] + }, + { + "task_id": "0003866", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_104122061850#640", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003866/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 397, + 163 + ], + [ + 372, + 161 + ], + [ + 389, + 169 + ], + [ + 405, + 157 + ], + [ + 412, + 171 + ], + [ + 430, + 160 + ] + ], + [ + [ + 408, + 169 + ], + [ + 410, + 171 + ], + [ + 382, + 168 + ], + [ + 434, + 170 + ], + [ + 381, + 166 + ], + [ + 431, + 168 + ] + ], + [ + [ + 393, + 151 + ], + [ + 392, + 136 + ], + [ + 382, + 172 + ], + [ + 404, + 131 + ], + [ + 383, + 186 + ], + [ + 404, + 146 + ] + ], + [ + [ + 396, + 171 + ], + [ + 422, + 174 + ], + [ + 392, + 197 + ], + [ + 400, + 145 + ], + [ + 368, + 193 + ], + [ + 375, + 143 + ] + ] + ] + }, + { + "task_id": "0003869", + "id": "RH20T_cfg6_task_0018_user_0014_scene_0007_cfg_0006_104122061018#170", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003869/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 341, + 155 + ], + [ + 329, + 141 + ], + [ + 367, + 150 + ], + [ + 314, + 159 + ], + [ + 382, + 165 + ], + [ + 327, + 175 + ] + ], + [ + [ + 352, + 126 + ], + [ + 374, + 123 + ], + [ + 341, + 139 + ], + [ + 364, + 111 + ], + [ + 319, + 142 + ], + [ + 340, + 114 + ] + ], + [ + [ + 341, + 129 + ], + [ + 343, + 108 + ], + [ + 327, + 111 + ], + [ + 356, + 148 + ], + [ + 326, + 130 + ], + [ + 354, + 166 + ] + ], + [ + [ + 345, + 114 + ], + [ + 322, + 124 + ], + [ + 329, + 111 + ], + [ + 360, + 116 + ], + [ + 354, + 99 + ], + [ + 384, + 106 + ] + ] + ] + }, + { + "task_id": "0003872", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_039422060546#76", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003872/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 275, + 297 + ], + [ + 258, + 262 + ], + [ + 282, + 310 + ], + [ + 268, + 286 + ], + [ + 300, + 348 + ], + [ + 284, + 318 + ] + ], + [ + [ + 254, + 345 + ], + [ + 212, + 347 + ], + [ + 260, + 345 + ], + [ + 245, + 344 + ], + [ + 299, + 343 + ], + [ + 291, + 342 + ] + ], + [ + [ + 273, + 360 + ], + [ + 231, + 360 + ], + [ + 275, + 360 + ], + [ + 271, + 332 + ], + [ + 321, + 360 + ], + [ + 310, + 319 + ] + ], + [ + [ + 280, + 334 + ], + [ + 276, + 317 + ], + [ + 279, + 360 + ], + [ + 282, + 296 + ], + [ + 283, + 360 + ], + [ + 286, + 313 + ] + ] + ] + }, + { + "task_id": "0003873", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_039422060546#189", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003873/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 289, + 336 + ], + [ + 249, + 337 + ], + [ + 294, + 321 + ], + [ + 283, + 354 + ], + [ + 332, + 321 + ], + [ + 328, + 353 + ] + ], + [ + [ + 308, + 305 + ], + [ + 305, + 288 + ], + [ + 295, + 344 + ], + [ + 319, + 271 + ], + [ + 298, + 358 + ], + [ + 321, + 288 + ] + ], + [ + [ + 269, + 312 + ], + [ + 237, + 286 + ], + [ + 279, + 285 + ], + [ + 261, + 336 + ], + [ + 315, + 312 + ], + [ + 293, + 360 + ] + ], + [ + [ + 345, + 285 + ], + [ + 366, + 275 + ], + [ + 348, + 246 + ], + [ + 343, + 328 + ], + [ + 331, + 258 + ], + [ + 325, + 334 + ] + ] + ] + }, + { + "task_id": "0003876", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_039422060546#556", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003876/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 383, + 199 + ], + [ + 415, + 207 + ], + [ + 378, + 227 + ], + [ + 388, + 170 + ], + [ + 348, + 219 + ], + [ + 356, + 163 + ] + ], + [ + [ + 349, + 202 + ], + [ + 348, + 180 + ], + [ + 340, + 231 + ], + [ + 358, + 176 + ], + [ + 342, + 250 + ], + [ + 359, + 197 + ] + ], + [ + [ + 315, + 218 + ], + [ + 289, + 237 + ], + [ + 304, + 225 + ], + [ + 324, + 212 + ], + [ + 334, + 205 + ], + [ + 350, + 194 + ] + ], + [ + [ + 375, + 241 + ], + [ + 409, + 254 + ], + [ + 383, + 240 + ], + [ + 365, + 244 + ], + [ + 353, + 229 + ], + [ + 331, + 231 + ] + ] + ] + }, + { + "task_id": "0003878", + "id": "RH20T_cfg6_task_0106_user_0020_scene_0004_cfg_0006_037522061512#191", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the object down from the shelf'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003878/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 287, + 0 + ], + [ + 286, + 0 + ], + [ + 301, + 0 + ], + [ + 273, + 8 + ], + [ + 301, + 13 + ], + [ + 275, + 32 + ] + ], + [ + [ + 295, + 19 + ], + [ + 277, + 23 + ], + [ + 284, + 0 + ], + [ + 306, + 41 + ], + [ + 302, + 0 + ], + [ + 323, + 37 + ] + ], + [ + [ + 311, + 11 + ], + [ + 334, + 18 + ], + [ + 307, + 6 + ], + [ + 315, + 15 + ], + [ + 283, + 0 + ], + [ + 293, + 9 + ] + ], + [ + [ + 289, + 29 + ], + [ + 300, + 17 + ], + [ + 296, + 52 + ], + [ + 282, + 4 + ], + [ + 285, + 65 + ], + [ + 269, + 16 + ] + ] + ] + }, + { + "task_id": "0003880", + "id": "RH20T_cfg7_task_0001_user_0014_scene_0007_cfg_0007_104122061602#128", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003880/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 503, + 185 + ], + [ + 508, + 161 + ], + [ + 525, + 199 + ], + [ + 483, + 173 + ], + [ + 519, + 222 + ], + [ + 479, + 195 + ] + ], + [ + [ + 491, + 187 + ], + [ + 469, + 196 + ], + [ + 504, + 212 + ], + [ + 477, + 162 + ], + [ + 529, + 204 + ], + [ + 500, + 151 + ] + ], + [ + [ + 493, + 177 + ], + [ + 492, + 204 + ], + [ + 464, + 176 + ], + [ + 522, + 178 + ], + [ + 465, + 150 + ], + [ + 522, + 152 + ] + ], + [ + [ + 504, + 152 + ], + [ + 527, + 158 + ], + [ + 517, + 163 + ], + [ + 490, + 140 + ], + [ + 496, + 156 + ], + [ + 469, + 134 + ] + ] + ] + }, + { + "task_id": "0003882", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061018#143", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003882/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 221, + 98 + ], + [ + 221, + 77 + ], + [ + 210, + 78 + ], + [ + 234, + 118 + ], + [ + 210, + 97 + ], + [ + 233, + 136 + ] + ], + [ + [ + 247, + 119 + ], + [ + 249, + 144 + ], + [ + 229, + 128 + ], + [ + 263, + 111 + ], + [ + 225, + 100 + ], + [ + 261, + 86 + ] + ], + [ + [ + 218, + 137 + ], + [ + 243, + 125 + ], + [ + 231, + 155 + ], + [ + 205, + 119 + ], + [ + 205, + 167 + ], + [ + 179, + 130 + ] + ], + [ + [ + 225, + 92 + ], + [ + 195, + 98 + ], + [ + 219, + 73 + ], + [ + 232, + 113 + ], + [ + 248, + 68 + ], + [ + 263, + 107 + ] + ] + ] + }, + { + "task_id": "0003883", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061018#250", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003883/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 243, + 130 + ], + [ + 220, + 122 + ], + [ + 262, + 116 + ], + [ + 222, + 146 + ], + [ + 285, + 124 + ], + [ + 247, + 155 + ] + ], + [ + [ + 218, + 126 + ], + [ + 224, + 117 + ], + [ + 222, + 101 + ], + [ + 215, + 153 + ], + [ + 217, + 110 + ], + [ + 210, + 160 + ] + ], + [ + [ + 243, + 132 + ], + [ + 214, + 140 + ], + [ + 245, + 112 + ], + [ + 241, + 151 + ], + [ + 273, + 105 + ], + [ + 267, + 143 + ] + ], + [ + [ + 219, + 126 + ], + [ + 221, + 109 + ], + [ + 208, + 103 + ], + [ + 231, + 151 + ], + [ + 207, + 120 + ], + [ + 229, + 165 + ] + ] + ] + }, + { + "task_id": "0003884", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061018#373", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003884/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 257, + 158 + ], + [ + 269, + 157 + ], + [ + 230, + 147 + ], + [ + 283, + 168 + ], + [ + 220, + 149 + ], + [ + 271, + 168 + ] + ], + [ + [ + 236, + 146 + ], + [ + 236, + 125 + ], + [ + 225, + 125 + ], + [ + 248, + 169 + ], + [ + 226, + 144 + ], + [ + 248, + 187 + ] + ], + [ + [ + 249, + 110 + ], + [ + 221, + 124 + ], + [ + 257, + 133 + ], + [ + 241, + 85 + ], + [ + 283, + 120 + ], + [ + 269, + 71 + ] + ], + [ + [ + 248, + 149 + ], + [ + 261, + 143 + ], + [ + 244, + 180 + ], + [ + 252, + 117 + ], + [ + 229, + 189 + ], + [ + 236, + 122 + ] + ] + ] + }, + { + "task_id": "0003885", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061018#489", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003885/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 244, + 180 + ], + [ + 274, + 182 + ], + [ + 255, + 164 + ], + [ + 231, + 198 + ], + [ + 227, + 163 + ], + [ + 200, + 196 + ] + ], + [ + [ + 250, + 170 + ], + [ + 254, + 199 + ], + [ + 219, + 175 + ], + [ + 280, + 165 + ], + [ + 215, + 146 + ], + [ + 276, + 136 + ] + ], + [ + [ + 246, + 164 + ], + [ + 247, + 141 + ], + [ + 236, + 144 + ], + [ + 257, + 187 + ], + [ + 236, + 165 + ], + [ + 256, + 207 + ] + ], + [ + [ + 256, + 123 + ], + [ + 239, + 138 + ], + [ + 229, + 115 + ], + [ + 281, + 129 + ], + [ + 248, + 97 + ], + [ + 299, + 113 + ] + ] + ] + }, + { + "task_id": "0003886", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061018#650", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003886/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 217, + 171 + ], + [ + 189, + 164 + ], + [ + 199, + 191 + ], + [ + 236, + 150 + ], + [ + 224, + 197 + ], + [ + 261, + 158 + ] + ], + [ + [ + 266, + 195 + ], + [ + 264, + 227 + ], + [ + 254, + 199 + ], + [ + 277, + 191 + ], + [ + 256, + 164 + ], + [ + 279, + 160 + ] + ], + [ + [ + 248, + 191 + ], + [ + 247, + 168 + ], + [ + 236, + 169 + ], + [ + 260, + 214 + ], + [ + 237, + 189 + ], + [ + 260, + 234 + ] + ], + [ + [ + 270, + 170 + ], + [ + 245, + 151 + ], + [ + 260, + 181 + ], + [ + 279, + 161 + ], + [ + 287, + 201 + ], + [ + 303, + 178 + ] + ] + ] + }, + { + "task_id": "0003887", + "id": "RH20T_cfg6_task_0060_user_0014_scene_0006_cfg_0006_104122064161#213", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to decrease the volume of the speaker'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003887/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 341, + 188 + ], + [ + 336, + 205 + ], + [ + 319, + 189 + ], + [ + 363, + 186 + ], + [ + 324, + 173 + ], + [ + 367, + 169 + ] + ], + [ + [ + 368, + 163 + ], + [ + 379, + 152 + ], + [ + 391, + 168 + ], + [ + 345, + 159 + ], + [ + 380, + 178 + ], + [ + 336, + 170 + ] + ], + [ + [ + 357, + 174 + ], + [ + 358, + 173 + ], + [ + 353, + 198 + ], + [ + 360, + 151 + ], + [ + 352, + 198 + ], + [ + 358, + 153 + ] + ], + [ + [ + 380, + 183 + ], + [ + 363, + 199 + ], + [ + 370, + 177 + ], + [ + 390, + 188 + ], + [ + 389, + 159 + ], + [ + 408, + 171 + ] + ] + ] + }, + { + "task_id": "0003888", + "id": "RH20T_cfg5_task_0101_user_0007_scene_0010_cfg_0005_036422060215#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the table with a cloth'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003888/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 316, + 176 + ], + [ + 302, + 162 + ], + [ + 302, + 190 + ], + [ + 330, + 162 + ], + [ + 316, + 205 + ], + [ + 343, + 177 + ] + ], + [ + [ + 306, + 173 + ], + [ + 314, + 156 + ], + [ + 287, + 165 + ], + [ + 325, + 180 + ], + [ + 279, + 182 + ], + [ + 318, + 197 + ] + ], + [ + [ + 299, + 156 + ], + [ + 305, + 143 + ], + [ + 318, + 157 + ], + [ + 280, + 156 + ], + [ + 312, + 170 + ], + [ + 275, + 169 + ] + ], + [ + [ + 296, + 160 + ], + [ + 304, + 177 + ], + [ + 308, + 149 + ], + [ + 285, + 170 + ], + [ + 300, + 132 + ], + [ + 278, + 153 + ] + ] + ] + }, + { + "task_id": "0003892", + "id": "RH20T_cfg4_task_0013_user_0010_scene_0002_cfg_0004_104422070011#111", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003892/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 428, + 195 + ], + [ + 425, + 172 + ], + [ + 424, + 195 + ], + [ + 431, + 195 + ], + [ + 426, + 219 + ], + [ + 433, + 217 + ] + ], + [ + [ + 423, + 170 + ], + [ + 432, + 158 + ], + [ + 400, + 169 + ], + [ + 444, + 170 + ], + [ + 393, + 180 + ], + [ + 435, + 181 + ] + ], + [ + [ + 425, + 168 + ], + [ + 442, + 185 + ], + [ + 416, + 182 + ], + [ + 435, + 154 + ], + [ + 400, + 166 + ], + [ + 417, + 137 + ] + ], + [ + [ + 422, + 175 + ], + [ + 419, + 195 + ], + [ + 397, + 171 + ], + [ + 448, + 179 + ], + [ + 399, + 151 + ], + [ + 452, + 159 + ] + ] + ] + }, + { + "task_id": "0003894", + "id": "RH20T_cfg5_task_0044_user_0010_scene_0008_cfg_0005_037522062165#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003894/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 222, + 242 + ], + [ + 250, + 226 + ], + [ + 211, + 217 + ], + [ + 232, + 266 + ], + [ + 179, + 234 + ], + [ + 203, + 283 + ] + ], + [ + [ + 215, + 234 + ], + [ + 224, + 235 + ], + [ + 244, + 236 + ], + [ + 184, + 231 + ], + [ + 235, + 235 + ], + [ + 177, + 231 + ] + ], + [ + [ + 227, + 212 + ], + [ + 237, + 188 + ], + [ + 199, + 195 + ], + [ + 253, + 228 + ], + [ + 186, + 221 + ], + [ + 243, + 254 + ] + ], + [ + [ + 231, + 216 + ], + [ + 227, + 196 + ], + [ + 211, + 200 + ], + [ + 251, + 233 + ], + [ + 216, + 218 + ], + [ + 254, + 251 + ] + ] + ] + }, + { + "task_id": "0003899", + "id": "RH20T_cfg1_task_0061_user_0015_scene_0001_cfg_0001_038522062547#730", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003899/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 47 + ], + [ + 369, + 34 + ], + [ + 359, + 68 + ], + [ + 335, + 25 + ], + [ + 338, + 80 + ], + [ + 314, + 39 + ] + ], + [ + [ + 349, + 57 + ], + [ + 346, + 40 + ], + [ + 325, + 66 + ], + [ + 373, + 47 + ], + [ + 327, + 84 + ], + [ + 376, + 65 + ] + ], + [ + [ + 353, + 88 + ], + [ + 357, + 108 + ], + [ + 370, + 77 + ], + [ + 335, + 98 + ], + [ + 367, + 55 + ], + [ + 331, + 76 + ] + ], + [ + [ + 351, + 68 + ], + [ + 353, + 55 + ], + [ + 366, + 86 + ], + [ + 337, + 50 + ], + [ + 364, + 97 + ], + [ + 336, + 63 + ] + ] + ] + }, + { + "task_id": "0003902", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0001_cfg_0004_104122062823#217", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003902/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 432, + 103 + ], + [ + 440, + 70 + ], + [ + 470, + 115 + ], + [ + 397, + 91 + ], + [ + 461, + 145 + ], + [ + 392, + 120 + ] + ], + [ + [ + 386, + 134 + ], + [ + 369, + 150 + ], + [ + 406, + 165 + ], + [ + 366, + 104 + ], + [ + 420, + 149 + ], + [ + 382, + 91 + ] + ], + [ + [ + 463, + 111 + ], + [ + 470, + 127 + ], + [ + 436, + 90 + ], + [ + 493, + 133 + ], + [ + 427, + 70 + ], + [ + 487, + 115 + ] + ], + [ + [ + 416, + 89 + ], + [ + 448, + 102 + ], + [ + 422, + 60 + ], + [ + 410, + 113 + ], + [ + 389, + 48 + ], + [ + 381, + 101 + ] + ] + ] + }, + { + "task_id": "0003907", + "id": "RH20T_cfg4_task_0057_user_0014_scene_0009_cfg_0004_104422070011#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003907/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 317, + 140 + ], + [ + 320, + 121 + ], + [ + 337, + 155 + ], + [ + 297, + 124 + ], + [ + 335, + 175 + ], + [ + 293, + 145 + ] + ], + [ + [ + 254, + 142 + ], + [ + 259, + 131 + ], + [ + 232, + 123 + ], + [ + 275, + 160 + ], + [ + 225, + 133 + ], + [ + 270, + 172 + ] + ], + [ + [ + 272, + 179 + ], + [ + 294, + 169 + ], + [ + 287, + 197 + ], + [ + 255, + 161 + ], + [ + 266, + 206 + ], + [ + 234, + 171 + ] + ], + [ + [ + 280, + 143 + ], + [ + 282, + 125 + ], + [ + 256, + 154 + ], + [ + 302, + 132 + ], + [ + 255, + 171 + ], + [ + 298, + 149 + ] + ] + ] + }, + { + "task_id": "0003908", + "id": "RH20T_cfg4_task_0057_user_0014_scene_0009_cfg_0004_104422070011#201", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003908/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 283, + 188 + ], + [ + 285, + 191 + ], + [ + 280, + 159 + ], + [ + 286, + 218 + ], + [ + 278, + 154 + ], + [ + 285, + 217 + ] + ], + [ + [ + 293, + 185 + ], + [ + 296, + 170 + ], + [ + 270, + 198 + ], + [ + 316, + 172 + ], + [ + 269, + 212 + ], + [ + 313, + 186 + ] + ], + [ + [ + 298, + 164 + ], + [ + 321, + 148 + ], + [ + 301, + 154 + ], + [ + 294, + 174 + ], + [ + 279, + 169 + ], + [ + 270, + 190 + ] + ], + [ + [ + 261, + 171 + ], + [ + 273, + 175 + ], + [ + 288, + 164 + ], + [ + 234, + 177 + ], + [ + 275, + 160 + ], + [ + 219, + 173 + ] + ] + ] + }, + { + "task_id": "0003909", + "id": "RH20T_cfg4_task_0057_user_0014_scene_0009_cfg_0004_104422070011#229", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003909/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 267, + 178 + ], + [ + 270, + 154 + ], + [ + 236, + 175 + ], + [ + 297, + 180 + ], + [ + 232, + 201 + ], + [ + 295, + 206 + ] + ], + [ + [ + 290, + 180 + ], + [ + 294, + 165 + ], + [ + 266, + 194 + ], + [ + 313, + 167 + ], + [ + 264, + 207 + ], + [ + 308, + 181 + ] + ], + [ + [ + 297, + 169 + ], + [ + 299, + 189 + ], + [ + 273, + 160 + ], + [ + 320, + 178 + ], + [ + 272, + 142 + ], + [ + 317, + 161 + ] + ], + [ + [ + 315, + 198 + ], + [ + 340, + 209 + ], + [ + 328, + 174 + ], + [ + 300, + 223 + ], + [ + 305, + 165 + ], + [ + 276, + 212 + ] + ] + ] + }, + { + "task_id": "0003910", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_037522062165#265", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003910/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 224, + 223 + ], + [ + 248, + 226 + ], + [ + 204, + 220 + ], + [ + 243, + 226 + ], + [ + 182, + 218 + ], + [ + 221, + 224 + ] + ], + [ + [ + 224, + 231 + ], + [ + 250, + 222 + ], + [ + 223, + 252 + ], + [ + 225, + 213 + ], + [ + 197, + 261 + ], + [ + 202, + 222 + ] + ], + [ + [ + 228, + 218 + ], + [ + 226, + 202 + ], + [ + 206, + 205 + ], + [ + 251, + 231 + ], + [ + 209, + 220 + ], + [ + 251, + 245 + ] + ], + [ + [ + 198, + 230 + ], + [ + 212, + 240 + ], + [ + 217, + 238 + ], + [ + 177, + 221 + ], + [ + 204, + 229 + ], + [ + 164, + 212 + ] + ] + ] + }, + { + "task_id": "0003912", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_037522062165#777", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003912/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 178, + 287 + ], + [ + 174, + 272 + ], + [ + 184, + 260 + ], + [ + 172, + 316 + ], + [ + 187, + 274 + ], + [ + 176, + 329 + ] + ], + [ + [ + 158, + 275 + ], + [ + 127, + 281 + ], + [ + 154, + 246 + ], + [ + 163, + 304 + ], + [ + 183, + 241 + ], + [ + 193, + 297 + ] + ], + [ + [ + 208, + 262 + ], + [ + 177, + 260 + ], + [ + 212, + 261 + ], + [ + 204, + 263 + ], + [ + 240, + 264 + ], + [ + 236, + 265 + ] + ], + [ + [ + 139, + 287 + ], + [ + 143, + 320 + ], + [ + 105, + 294 + ], + [ + 173, + 279 + ], + [ + 102, + 262 + ], + [ + 169, + 248 + ] + ] + ] + }, + { + "task_id": "0003914", + "id": "RH20T_cfg3_task_0018_user_0011_scene_0004_cfg_0003_036422060909#143", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003914/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 324, + 170 + ], + [ + 327, + 154 + ], + [ + 343, + 158 + ], + [ + 305, + 182 + ], + [ + 340, + 173 + ], + [ + 303, + 197 + ] + ], + [ + [ + 326, + 175 + ], + [ + 309, + 156 + ], + [ + 308, + 192 + ], + [ + 344, + 157 + ], + [ + 325, + 212 + ], + [ + 362, + 177 + ] + ], + [ + [ + 324, + 179 + ], + [ + 332, + 188 + ], + [ + 335, + 200 + ], + [ + 313, + 159 + ], + [ + 326, + 190 + ], + [ + 303, + 148 + ] + ], + [ + [ + 313, + 178 + ], + [ + 313, + 171 + ], + [ + 325, + 158 + ], + [ + 301, + 198 + ], + [ + 325, + 165 + ], + [ + 300, + 207 + ] + ] + ] + }, + { + "task_id": "0003916", + "id": "RH20T_cfg1_task_0015_user_0001_scene_0006_cfg_0001_750612070853#159", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003916/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 267 + ], + [ + 382, + 255 + ], + [ + 388, + 287 + ], + [ + 315, + 250 + ], + [ + 348, + 303 + ], + [ + 276, + 261 + ] + ], + [ + [ + 383, + 237 + ], + [ + 337, + 230 + ], + [ + 397, + 228 + ], + [ + 367, + 249 + ], + [ + 443, + 235 + ], + [ + 423, + 259 + ] + ], + [ + [ + 354, + 291 + ], + [ + 351, + 255 + ], + [ + 336, + 331 + ], + [ + 369, + 258 + ], + [ + 340, + 360 + ], + [ + 371, + 290 + ] + ], + [ + [ + 327, + 277 + ], + [ + 371, + 288 + ], + [ + 338, + 234 + ], + [ + 314, + 323 + ], + [ + 297, + 224 + ], + [ + 272, + 311 + ] + ] + ] + }, + { + "task_id": "0003918", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0004_cfg_0006_104122064161#179", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003918/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 420, + 134 + ], + [ + 419, + 118 + ], + [ + 422, + 150 + ], + [ + 419, + 120 + ], + [ + 423, + 165 + ], + [ + 420, + 135 + ] + ], + [ + [ + 442, + 144 + ], + [ + 420, + 149 + ], + [ + 439, + 126 + ], + [ + 445, + 161 + ], + [ + 462, + 120 + ], + [ + 467, + 156 + ] + ], + [ + [ + 419, + 149 + ], + [ + 442, + 144 + ], + [ + 417, + 144 + ], + [ + 422, + 153 + ], + [ + 394, + 149 + ], + [ + 401, + 157 + ] + ], + [ + [ + 422, + 115 + ], + [ + 441, + 119 + ], + [ + 425, + 94 + ], + [ + 419, + 136 + ], + [ + 406, + 91 + ], + [ + 401, + 132 + ] + ] + ] + }, + { + "task_id": "0003919", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0004_cfg_0006_104122064161#401", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003919/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 385, + 172 + ], + [ + 408, + 178 + ], + [ + 391, + 150 + ], + [ + 380, + 195 + ], + [ + 368, + 144 + ], + [ + 357, + 189 + ] + ], + [ + [ + 380, + 192 + ], + [ + 378, + 210 + ], + [ + 376, + 207 + ], + [ + 384, + 175 + ], + [ + 377, + 190 + ], + [ + 386, + 158 + ] + ], + [ + [ + 368, + 183 + ], + [ + 351, + 201 + ], + [ + 365, + 179 + ], + [ + 370, + 187 + ], + [ + 382, + 160 + ], + [ + 385, + 170 + ] + ], + [ + [ + 393, + 191 + ], + [ + 414, + 180 + ], + [ + 397, + 207 + ], + [ + 388, + 177 + ], + [ + 375, + 218 + ], + [ + 368, + 188 + ] + ] + ] + }, + { + "task_id": "0003920", + "id": "RH20T_cfg5_task_0023_user_0010_scene_0007_cfg_0005_104422070044#227", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003920/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 343, + 200 + ], + [ + 342, + 185 + ], + [ + 359, + 206 + ], + [ + 327, + 194 + ], + [ + 359, + 220 + ], + [ + 328, + 209 + ] + ], + [ + [ + 321, + 195 + ], + [ + 321, + 186 + ], + [ + 302, + 193 + ], + [ + 340, + 197 + ], + [ + 300, + 203 + ], + [ + 339, + 207 + ] + ], + [ + [ + 348, + 208 + ], + [ + 336, + 224 + ], + [ + 334, + 198 + ], + [ + 360, + 218 + ], + [ + 346, + 182 + ], + [ + 372, + 203 + ] + ], + [ + [ + 333, + 203 + ], + [ + 322, + 218 + ], + [ + 330, + 198 + ], + [ + 336, + 209 + ], + [ + 340, + 184 + ], + [ + 347, + 195 + ] + ] + ] + }, + { + "task_id": "0003923", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0002_cfg_0005_104422070044#82", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003923/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 344, + 127 + ], + [ + 350, + 145 + ], + [ + 340, + 128 + ], + [ + 347, + 125 + ], + [ + 333, + 109 + ], + [ + 341, + 108 + ] + ], + [ + [ + 356, + 150 + ], + [ + 374, + 149 + ], + [ + 354, + 131 + ], + [ + 359, + 169 + ], + [ + 335, + 132 + ], + [ + 340, + 170 + ] + ], + [ + [ + 342, + 127 + ], + [ + 361, + 126 + ], + [ + 344, + 114 + ], + [ + 340, + 139 + ], + [ + 324, + 115 + ], + [ + 321, + 140 + ] + ], + [ + [ + 343, + 140 + ], + [ + 342, + 125 + ], + [ + 350, + 151 + ], + [ + 338, + 130 + ], + [ + 351, + 165 + ], + [ + 339, + 144 + ] + ] + ] + }, + { + "task_id": "0003926", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0002_cfg_0005_104422070044#237", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003926/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 329, + 161 + ], + [ + 346, + 157 + ], + [ + 331, + 175 + ], + [ + 326, + 147 + ], + [ + 315, + 180 + ], + [ + 309, + 151 + ] + ], + [ + [ + 357, + 148 + ], + [ + 363, + 164 + ], + [ + 360, + 145 + ], + [ + 353, + 150 + ], + [ + 355, + 130 + ], + [ + 347, + 134 + ] + ], + [ + [ + 341, + 143 + ], + [ + 339, + 129 + ], + [ + 345, + 154 + ], + [ + 337, + 134 + ], + [ + 347, + 168 + ], + [ + 339, + 147 + ] + ], + [ + [ + 336, + 145 + ], + [ + 320, + 147 + ], + [ + 346, + 139 + ], + [ + 326, + 151 + ], + [ + 361, + 138 + ], + [ + 341, + 150 + ] + ] + ] + }, + { + "task_id": "0003927", + "id": "RH20T_cfg2_task_0042_user_0007_scene_0002_cfg_0002_037522062165#238", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003927/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 270, + 219 + ], + [ + 242, + 231 + ], + [ + 264, + 192 + ], + [ + 275, + 244 + ], + [ + 291, + 181 + ], + [ + 300, + 233 + ] + ], + [ + [ + 253, + 241 + ], + [ + 265, + 236 + ], + [ + 235, + 216 + ], + [ + 270, + 267 + ], + [ + 225, + 222 + ], + [ + 259, + 270 + ] + ], + [ + [ + 252, + 231 + ], + [ + 253, + 217 + ], + [ + 228, + 213 + ], + [ + 277, + 249 + ], + [ + 228, + 226 + ], + [ + 274, + 260 + ] + ], + [ + [ + 224, + 238 + ], + [ + 200, + 259 + ], + [ + 245, + 256 + ], + [ + 203, + 221 + ], + [ + 267, + 235 + ], + [ + 226, + 203 + ] + ] + ] + }, + { + "task_id": "0003932", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0003_cfg_0003_036422060909#283", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003932/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 127 + ], + [ + 331, + 105 + ], + [ + 331, + 111 + ], + [ + 321, + 141 + ], + [ + 324, + 135 + ], + [ + 315, + 164 + ] + ], + [ + [ + 363, + 110 + ], + [ + 351, + 121 + ], + [ + 385, + 124 + ], + [ + 341, + 96 + ], + [ + 400, + 112 + ], + [ + 353, + 83 + ] + ], + [ + [ + 317, + 148 + ], + [ + 303, + 132 + ], + [ + 295, + 160 + ], + [ + 338, + 136 + ], + [ + 309, + 178 + ], + [ + 353, + 152 + ] + ], + [ + [ + 336, + 131 + ], + [ + 341, + 111 + ], + [ + 357, + 123 + ], + [ + 315, + 140 + ], + [ + 352, + 141 + ], + [ + 312, + 158 + ] + ] + ] + }, + { + "task_id": "0003934", + "id": "RH20T_cfg1_task_0015_user_0001_scene_0007_cfg_0001_819112070044#266", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003934/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 322, + 120 + ], + [ + 328, + 105 + ], + [ + 316, + 106 + ], + [ + 328, + 136 + ], + [ + 310, + 121 + ], + [ + 323, + 150 + ] + ], + [ + [ + 311, + 120 + ], + [ + 291, + 125 + ], + [ + 317, + 132 + ], + [ + 303, + 107 + ], + [ + 336, + 127 + ], + [ + 324, + 102 + ] + ], + [ + [ + 337, + 98 + ], + [ + 320, + 100 + ], + [ + 326, + 86 + ], + [ + 349, + 109 + ], + [ + 342, + 84 + ], + [ + 365, + 106 + ] + ], + [ + [ + 309, + 121 + ], + [ + 297, + 135 + ], + [ + 323, + 123 + ], + [ + 294, + 119 + ], + [ + 336, + 109 + ], + [ + 308, + 103 + ] + ] + ] + }, + { + "task_id": "0003937", + "id": "RH20T_cfg4_task_0064_user_0016_scene_0003_cfg_0004_104122062823#159", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003937/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 423, + 18 + ], + [ + 452, + 18 + ], + [ + 437, + 0 + ], + [ + 409, + 54 + ], + [ + 410, + 0 + ], + [ + 384, + 52 + ] + ], + [ + [ + 390, + 48 + ], + [ + 393, + 18 + ], + [ + 411, + 30 + ], + [ + 368, + 67 + ], + [ + 407, + 58 + ], + [ + 367, + 93 + ] + ], + [ + [ + 422, + 8 + ], + [ + 445, + 6 + ], + [ + 416, + 0 + ], + [ + 428, + 44 + ], + [ + 396, + 0 + ], + [ + 408, + 43 + ] + ], + [ + [ + 376, + 36 + ], + [ + 347, + 30 + ], + [ + 358, + 65 + ], + [ + 395, + 4 + ], + [ + 384, + 69 + ], + [ + 423, + 11 + ] + ] + ] + }, + { + "task_id": "0003941", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0008_cfg_0004_104422070011#108", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003941/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 286, + 143 + ], + [ + 299, + 139 + ], + [ + 281, + 110 + ], + [ + 292, + 176 + ], + [ + 270, + 116 + ], + [ + 281, + 178 + ] + ], + [ + [ + 319, + 143 + ], + [ + 316, + 144 + ], + [ + 284, + 143 + ], + [ + 353, + 144 + ], + [ + 284, + 142 + ], + [ + 360, + 143 + ] + ], + [ + [ + 273, + 121 + ], + [ + 257, + 137 + ], + [ + 276, + 93 + ], + [ + 270, + 151 + ], + [ + 294, + 74 + ], + [ + 289, + 135 + ] + ], + [ + [ + 298, + 144 + ], + [ + 311, + 129 + ], + [ + 275, + 161 + ], + [ + 319, + 129 + ], + [ + 264, + 174 + ], + [ + 307, + 142 + ] + ] + ] + }, + { + "task_id": "0003943", + "id": "RH20T_cfg5_task_0018_user_0010_scene_0009_cfg_0005_104122061850#289", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003943/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 388, + 128 + ], + [ + 400, + 129 + ], + [ + 411, + 120 + ], + [ + 366, + 135 + ], + [ + 397, + 119 + ], + [ + 352, + 135 + ] + ], + [ + [ + 374, + 139 + ], + [ + 372, + 123 + ], + [ + 350, + 134 + ], + [ + 399, + 144 + ], + [ + 353, + 148 + ], + [ + 399, + 158 + ] + ], + [ + [ + 372, + 152 + ], + [ + 367, + 171 + ], + [ + 349, + 156 + ], + [ + 397, + 148 + ], + [ + 355, + 139 + ], + [ + 401, + 130 + ] + ], + [ + [ + 392, + 129 + ], + [ + 393, + 154 + ], + [ + 396, + 138 + ], + [ + 388, + 122 + ], + [ + 396, + 110 + ], + [ + 388, + 96 + ] + ] + ] + }, + { + "task_id": "0003945", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0003_cfg_0006_104122064161#214", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003945/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 366, + 122 + ], + [ + 363, + 108 + ], + [ + 389, + 123 + ], + [ + 342, + 121 + ], + [ + 393, + 139 + ], + [ + 344, + 137 + ] + ], + [ + [ + 373, + 100 + ], + [ + 396, + 98 + ], + [ + 378, + 108 + ], + [ + 368, + 92 + ], + [ + 354, + 111 + ], + [ + 345, + 94 + ] + ], + [ + [ + 356, + 101 + ], + [ + 360, + 77 + ], + [ + 356, + 108 + ], + [ + 356, + 95 + ], + [ + 352, + 132 + ], + [ + 353, + 117 + ] + ], + [ + [ + 369, + 89 + ], + [ + 363, + 113 + ], + [ + 345, + 83 + ], + [ + 392, + 94 + ], + [ + 351, + 58 + ], + [ + 398, + 69 + ] + ] + ] + }, + { + "task_id": "0003946", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0003_cfg_0006_104122064161#515", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003946/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 364, + 164 + ], + [ + 390, + 176 + ], + [ + 355, + 183 + ], + [ + 373, + 144 + ], + [ + 330, + 172 + ], + [ + 346, + 131 + ] + ], + [ + [ + 372, + 184 + ], + [ + 366, + 195 + ], + [ + 383, + 160 + ], + [ + 359, + 209 + ], + [ + 391, + 147 + ], + [ + 365, + 198 + ] + ], + [ + [ + 347, + 165 + ], + [ + 327, + 162 + ], + [ + 358, + 141 + ], + [ + 335, + 191 + ], + [ + 379, + 144 + ], + [ + 356, + 196 + ] + ], + [ + [ + 359, + 185 + ], + [ + 364, + 166 + ], + [ + 357, + 207 + ], + [ + 360, + 164 + ], + [ + 352, + 224 + ], + [ + 355, + 181 + ] + ] + ] + }, + { + "task_id": "0003948", + "id": "RH20T_cfg1_task_0061_user_0015_scene_0001_cfg_0001_039422060546#449", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003948/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 317, + 116 + ], + [ + 322, + 90 + ], + [ + 317, + 130 + ], + [ + 317, + 100 + ], + [ + 313, + 156 + ], + [ + 312, + 128 + ] + ], + [ + [ + 293, + 102 + ], + [ + 290, + 81 + ], + [ + 294, + 125 + ], + [ + 291, + 81 + ], + [ + 297, + 143 + ], + [ + 293, + 100 + ] + ], + [ + [ + 298, + 74 + ], + [ + 306, + 91 + ], + [ + 328, + 66 + ], + [ + 269, + 82 + ], + [ + 321, + 47 + ], + [ + 259, + 64 + ] + ], + [ + [ + 289, + 76 + ], + [ + 309, + 72 + ], + [ + 309, + 97 + ], + [ + 270, + 56 + ], + [ + 288, + 103 + ], + [ + 248, + 59 + ] + ] + ] + }, + { + "task_id": "0003950", + "id": "RH20T_cfg5_task_0042_user_0007_scene_0003_cfg_0005_104422070044#126", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003950/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 435, + 161 + ], + [ + 435, + 149 + ], + [ + 450, + 169 + ], + [ + 421, + 153 + ], + [ + 449, + 180 + ], + [ + 421, + 164 + ] + ], + [ + [ + 449, + 163 + ], + [ + 447, + 160 + ], + [ + 445, + 145 + ], + [ + 452, + 180 + ], + [ + 447, + 148 + ], + [ + 454, + 184 + ] + ], + [ + [ + 447, + 154 + ], + [ + 461, + 143 + ], + [ + 456, + 167 + ], + [ + 438, + 142 + ], + [ + 442, + 178 + ], + [ + 424, + 153 + ] + ], + [ + [ + 430, + 168 + ], + [ + 440, + 158 + ], + [ + 433, + 158 + ], + [ + 427, + 178 + ], + [ + 423, + 169 + ], + [ + 418, + 188 + ] + ] + ] + }, + { + "task_id": "0003952", + "id": "RH20T_cfg5_task_0042_user_0007_scene_0003_cfg_0005_104422070044#357", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003952/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 407, + 181 + ], + [ + 405, + 169 + ], + [ + 424, + 188 + ], + [ + 390, + 174 + ], + [ + 425, + 200 + ], + [ + 392, + 185 + ] + ], + [ + [ + 419, + 194 + ], + [ + 405, + 190 + ], + [ + 408, + 184 + ], + [ + 429, + 203 + ], + [ + 424, + 187 + ], + [ + 444, + 207 + ] + ], + [ + [ + 414, + 197 + ], + [ + 402, + 207 + ], + [ + 417, + 182 + ], + [ + 412, + 211 + ], + [ + 428, + 173 + ], + [ + 423, + 202 + ] + ], + [ + [ + 401, + 169 + ], + [ + 403, + 156 + ], + [ + 398, + 182 + ], + [ + 404, + 156 + ], + [ + 397, + 196 + ], + [ + 403, + 170 + ] + ] + ] + }, + { + "task_id": "0003955", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_038522062288#288", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003955/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 207, + 190 + ], + [ + 203, + 170 + ], + [ + 177, + 189 + ], + [ + 236, + 191 + ], + [ + 182, + 208 + ], + [ + 238, + 209 + ] + ], + [ + [ + 201, + 183 + ], + [ + 202, + 213 + ], + [ + 223, + 188 + ], + [ + 177, + 178 + ], + [ + 222, + 161 + ], + [ + 177, + 147 + ] + ], + [ + [ + 183, + 217 + ], + [ + 174, + 204 + ], + [ + 205, + 196 + ], + [ + 161, + 238 + ], + [ + 215, + 209 + ], + [ + 169, + 254 + ] + ], + [ + [ + 228, + 189 + ], + [ + 222, + 186 + ], + [ + 252, + 176 + ], + [ + 203, + 201 + ], + [ + 256, + 180 + ], + [ + 209, + 203 + ] + ] + ] + }, + { + "task_id": "0003957", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_038522062288#516", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003957/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 217, + 164 + ], + [ + 216, + 139 + ], + [ + 242, + 157 + ], + [ + 190, + 172 + ], + [ + 242, + 180 + ], + [ + 192, + 196 + ] + ], + [ + [ + 215, + 182 + ], + [ + 215, + 160 + ], + [ + 188, + 192 + ], + [ + 244, + 172 + ], + [ + 187, + 216 + ], + [ + 246, + 195 + ] + ], + [ + [ + 169, + 189 + ], + [ + 172, + 178 + ], + [ + 199, + 197 + ], + [ + 138, + 181 + ], + [ + 198, + 210 + ], + [ + 133, + 194 + ] + ], + [ + [ + 196, + 169 + ], + [ + 193, + 149 + ], + [ + 165, + 168 + ], + [ + 227, + 170 + ], + [ + 169, + 186 + ], + [ + 228, + 187 + ] + ] + ] + }, + { + "task_id": "0003958", + "id": "RH20T_cfg4_task_0127_user_0010_scene_0001_cfg_0004_104122062295#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003958/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 273, + 312 + ], + [ + 279, + 262 + ], + [ + 304, + 344 + ], + [ + 248, + 286 + ], + [ + 297, + 360 + ], + [ + 243, + 329 + ] + ], + [ + [ + 288, + 260 + ], + [ + 314, + 250 + ], + [ + 241, + 246 + ], + [ + 340, + 277 + ], + [ + 204, + 257 + ], + [ + 311, + 294 + ] + ], + [ + [ + 238, + 320 + ], + [ + 224, + 270 + ], + [ + 190, + 353 + ], + [ + 278, + 293 + ], + [ + 205, + 360 + ], + [ + 290, + 338 + ] + ], + [ + [ + 190, + 339 + ], + [ + 213, + 310 + ], + [ + 203, + 360 + ], + [ + 176, + 283 + ], + [ + 174, + 360 + ], + [ + 142, + 313 + ] + ] + ] + }, + { + "task_id": "0003959", + "id": "RH20T_cfg4_task_0127_user_0010_scene_0001_cfg_0004_104122062295#372", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003959/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 222, + 167 + ], + [ + 219, + 134 + ], + [ + 178, + 171 + ], + [ + 264, + 164 + ], + [ + 183, + 200 + ], + [ + 264, + 193 + ] + ], + [ + [ + 222, + 174 + ], + [ + 256, + 194 + ], + [ + 197, + 200 + ], + [ + 245, + 149 + ], + [ + 162, + 179 + ], + [ + 213, + 131 + ] + ], + [ + [ + 227, + 170 + ], + [ + 227, + 140 + ], + [ + 269, + 169 + ], + [ + 184, + 171 + ], + [ + 267, + 194 + ], + [ + 186, + 197 + ] + ], + [ + [ + 209, + 194 + ], + [ + 202, + 239 + ], + [ + 163, + 187 + ], + [ + 251, + 199 + ], + [ + 171, + 140 + ], + [ + 257, + 157 + ] + ] + ] + }, + { + "task_id": "0003960", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0005_cfg_0005_036422060215#190", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003960/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 318, + 159 + ], + [ + 312, + 177 + ], + [ + 338, + 167 + ], + [ + 298, + 151 + ], + [ + 345, + 149 + ], + [ + 304, + 133 + ] + ], + [ + [ + 311, + 129 + ], + [ + 327, + 120 + ], + [ + 298, + 111 + ], + [ + 323, + 146 + ], + [ + 283, + 120 + ], + [ + 308, + 154 + ] + ], + [ + [ + 319, + 140 + ], + [ + 325, + 127 + ], + [ + 341, + 146 + ], + [ + 298, + 134 + ], + [ + 336, + 160 + ], + [ + 291, + 147 + ] + ], + [ + [ + 304, + 142 + ], + [ + 308, + 124 + ], + [ + 321, + 138 + ], + [ + 286, + 146 + ], + [ + 317, + 155 + ], + [ + 282, + 164 + ] + ] + ] + }, + { + "task_id": "0003961", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0005_cfg_0005_036422060215#280", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003961/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 291, + 154 + ], + [ + 281, + 144 + ], + [ + 309, + 158 + ], + [ + 272, + 149 + ], + [ + 320, + 169 + ], + [ + 283, + 160 + ] + ], + [ + [ + 274, + 149 + ], + [ + 277, + 131 + ], + [ + 293, + 145 + ], + [ + 253, + 153 + ], + [ + 289, + 162 + ], + [ + 251, + 171 + ] + ], + [ + [ + 273, + 168 + ], + [ + 284, + 178 + ], + [ + 267, + 153 + ], + [ + 280, + 183 + ], + [ + 255, + 141 + ], + [ + 268, + 173 + ] + ], + [ + [ + 275, + 147 + ], + [ + 295, + 142 + ], + [ + 269, + 160 + ], + [ + 281, + 134 + ], + [ + 249, + 164 + ], + [ + 262, + 139 + ] + ] + ] + }, + { + "task_id": "0003963", + "id": "RH20T_cfg4_task_0089_user_0014_scene_0001_cfg_0004_104122063550#44", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push an object with obstacles'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003963/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 272, + 278 + ], + [ + 243, + 265 + ], + [ + 285, + 247 + ], + [ + 258, + 309 + ], + [ + 315, + 260 + ], + [ + 289, + 323 + ] + ], + [ + [ + 277, + 317 + ], + [ + 283, + 310 + ], + [ + 269, + 284 + ], + [ + 284, + 350 + ], + [ + 262, + 290 + ], + [ + 278, + 360 + ] + ], + [ + [ + 280, + 274 + ], + [ + 256, + 258 + ], + [ + 294, + 282 + ], + [ + 267, + 266 + ], + [ + 317, + 297 + ], + [ + 288, + 281 + ] + ], + [ + [ + 278, + 313 + ], + [ + 279, + 307 + ], + [ + 296, + 341 + ], + [ + 261, + 287 + ], + [ + 295, + 345 + ], + [ + 262, + 293 + ] + ] + ] + }, + { + "task_id": "0003965", + "id": "RH20T_cfg1_task_0046_user_0001_scene_0003_cfg_0001_038522062547#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003965/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 517, + 256 + ], + [ + 525, + 242 + ], + [ + 547, + 251 + ], + [ + 486, + 260 + ], + [ + 538, + 264 + ], + [ + 481, + 272 + ] + ], + [ + [ + 502, + 274 + ], + [ + 494, + 249 + ], + [ + 509, + 256 + ], + [ + 495, + 290 + ], + [ + 518, + 283 + ], + [ + 503, + 317 + ] + ], + [ + [ + 487, + 228 + ], + [ + 505, + 206 + ], + [ + 468, + 234 + ], + [ + 504, + 223 + ], + [ + 452, + 255 + ], + [ + 488, + 243 + ] + ], + [ + [ + 533, + 254 + ], + [ + 507, + 270 + ], + [ + 550, + 263 + ], + [ + 515, + 244 + ], + [ + 578, + 247 + ], + [ + 543, + 227 + ] + ] + ] + }, + { + "task_id": "0003968", + "id": "RH20T_cfg5_task_0015_user_0007_scene_0006_cfg_0005_104122061850#249", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003968/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 365, + 163 + ], + [ + 337, + 152 + ], + [ + 379, + 134 + ], + [ + 352, + 192 + ], + [ + 409, + 144 + ], + [ + 381, + 204 + ] + ], + [ + [ + 352, + 170 + ], + [ + 346, + 188 + ], + [ + 342, + 142 + ], + [ + 360, + 196 + ], + [ + 347, + 127 + ], + [ + 364, + 179 + ] + ], + [ + [ + 362, + 119 + ], + [ + 371, + 140 + ], + [ + 390, + 107 + ], + [ + 334, + 131 + ], + [ + 380, + 88 + ], + [ + 326, + 111 + ] + ], + [ + [ + 359, + 156 + ], + [ + 354, + 138 + ], + [ + 329, + 161 + ], + [ + 389, + 151 + ], + [ + 335, + 176 + ], + [ + 391, + 167 + ] + ] + ] + }, + { + "task_id": "0003970", + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_819112071037#167", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003970/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 46, + 196 + ], + [ + 39, + 196 + ], + [ + 60, + 252 + ], + [ + 32, + 140 + ], + [ + 64, + 246 + ], + [ + 39, + 146 + ] + ], + [ + [ + 48, + 145 + ], + [ + 79, + 195 + ], + [ + 104, + 123 + ], + [ + 0, + 170 + ], + [ + 75, + 77 + ], + [ + 0, + 115 + ] + ], + [ + [ + 54, + 147 + ], + [ + 44, + 94 + ], + [ + 0, + 159 + ], + [ + 103, + 137 + ], + [ + 6, + 212 + ], + [ + 110, + 182 + ] + ], + [ + [ + 9, + 188 + ], + [ + 0, + 181 + ], + [ + 37, + 142 + ], + [ + 0, + 236 + ], + [ + 53, + 147 + ], + [ + 0, + 250 + ] + ] + ] + }, + { + "task_id": "0003974", + "id": "RH20T_cfg3_task_0007_user_0016_scene_0004_cfg_0003_036422060909#214", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003974/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 91 + ], + [ + 321, + 110 + ], + [ + 316, + 91 + ], + [ + 303, + 90 + ], + [ + 304, + 70 + ], + [ + 292, + 72 + ] + ], + [ + [ + 312, + 106 + ], + [ + 302, + 126 + ], + [ + 297, + 96 + ], + [ + 327, + 117 + ], + [ + 307, + 75 + ], + [ + 337, + 98 + ] + ], + [ + [ + 314, + 110 + ], + [ + 315, + 93 + ], + [ + 330, + 98 + ], + [ + 298, + 122 + ], + [ + 329, + 114 + ], + [ + 299, + 137 + ] + ], + [ + [ + 325, + 89 + ], + [ + 302, + 84 + ], + [ + 327, + 86 + ], + [ + 323, + 92 + ], + [ + 351, + 91 + ], + [ + 344, + 97 + ] + ] + ] + }, + { + "task_id": "0003977", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_104422070044#220", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003977/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 206 + ], + [ + 334, + 216 + ], + [ + 341, + 204 + ], + [ + 355, + 208 + ], + [ + 354, + 195 + ], + [ + 369, + 199 + ] + ], + [ + [ + 342, + 197 + ], + [ + 348, + 187 + ], + [ + 344, + 185 + ], + [ + 340, + 209 + ], + [ + 338, + 195 + ], + [ + 334, + 219 + ] + ], + [ + [ + 340, + 217 + ], + [ + 331, + 207 + ], + [ + 353, + 206 + ], + [ + 327, + 229 + ], + [ + 364, + 216 + ], + [ + 337, + 240 + ] + ], + [ + [ + 347, + 203 + ], + [ + 344, + 192 + ], + [ + 357, + 213 + ], + [ + 337, + 193 + ], + [ + 360, + 223 + ], + [ + 339, + 204 + ] + ] + ] + }, + { + "task_id": "0003979", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_104422070044#419", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003979/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 350, + 184 + ], + [ + 348, + 173 + ], + [ + 360, + 194 + ], + [ + 340, + 175 + ], + [ + 361, + 205 + ], + [ + 342, + 185 + ] + ], + [ + [ + 351, + 183 + ], + [ + 336, + 185 + ], + [ + 351, + 167 + ], + [ + 352, + 198 + ], + [ + 365, + 165 + ], + [ + 366, + 195 + ] + ], + [ + [ + 371, + 192 + ], + [ + 388, + 193 + ], + [ + 371, + 184 + ], + [ + 372, + 200 + ], + [ + 354, + 183 + ], + [ + 355, + 199 + ] + ], + [ + [ + 345, + 179 + ], + [ + 352, + 190 + ], + [ + 359, + 169 + ], + [ + 330, + 189 + ], + [ + 351, + 158 + ], + [ + 322, + 178 + ] + ] + ] + }, + { + "task_id": "0003980", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_104422070044#577", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003980/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 336, + 172 + ], + [ + 352, + 167 + ], + [ + 333, + 159 + ], + [ + 339, + 184 + ], + [ + 317, + 163 + ], + [ + 324, + 189 + ] + ], + [ + [ + 337, + 180 + ], + [ + 335, + 169 + ], + [ + 342, + 191 + ], + [ + 333, + 169 + ], + [ + 344, + 202 + ], + [ + 335, + 180 + ] + ], + [ + [ + 329, + 187 + ], + [ + 314, + 181 + ], + [ + 328, + 194 + ], + [ + 330, + 180 + ], + [ + 343, + 199 + ], + [ + 345, + 185 + ] + ], + [ + [ + 345, + 187 + ], + [ + 334, + 194 + ], + [ + 334, + 178 + ], + [ + 356, + 196 + ], + [ + 344, + 171 + ], + [ + 367, + 189 + ] + ] + ] + }, + { + "task_id": "0003982", + "id": "RH20T_cfg7_task_0044_user_0020_scene_0001_cfg_0007_104122061330#117", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003982/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 218 + ], + [ + 316, + 197 + ], + [ + 329, + 224 + ], + [ + 284, + 213 + ], + [ + 319, + 247 + ], + [ + 273, + 235 + ] + ], + [ + [ + 318, + 201 + ], + [ + 297, + 204 + ], + [ + 308, + 179 + ], + [ + 329, + 224 + ], + [ + 328, + 177 + ], + [ + 349, + 220 + ] + ], + [ + [ + 282, + 236 + ], + [ + 293, + 227 + ], + [ + 259, + 229 + ], + [ + 306, + 243 + ], + [ + 246, + 239 + ], + [ + 295, + 255 + ] + ], + [ + [ + 293, + 213 + ], + [ + 284, + 202 + ], + [ + 282, + 237 + ], + [ + 304, + 189 + ], + [ + 290, + 246 + ], + [ + 312, + 200 + ] + ] + ] + }, + { + "task_id": "0003983", + "id": "RH20T_cfg3_task_0007_user_0016_scene_0008_cfg_0003_104122062295#175", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003983/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 91, + 231 + ], + [ + 77, + 192 + ], + [ + 123, + 219 + ], + [ + 53, + 246 + ], + [ + 135, + 254 + ], + [ + 68, + 287 + ] + ], + [ + [ + 92, + 247 + ], + [ + 87, + 217 + ], + [ + 72, + 219 + ], + [ + 115, + 277 + ], + [ + 77, + 246 + ], + [ + 119, + 303 + ] + ], + [ + [ + 114, + 231 + ], + [ + 152, + 216 + ], + [ + 93, + 206 + ], + [ + 137, + 258 + ], + [ + 50, + 222 + ], + [ + 94, + 279 + ] + ], + [ + [ + 100, + 228 + ], + [ + 105, + 193 + ], + [ + 87, + 209 + ], + [ + 111, + 244 + ], + [ + 80, + 250 + ], + [ + 106, + 280 + ] + ] + ] + }, + { + "task_id": "0003986", + "id": "RH20T_cfg1_task_0015_user_0001_scene_0006_cfg_0001_038522062547#278", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003986/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 434, + 122 + ], + [ + 451, + 140 + ], + [ + 429, + 115 + ], + [ + 439, + 127 + ], + [ + 412, + 96 + ], + [ + 423, + 110 + ] + ], + [ + [ + 406, + 136 + ], + [ + 409, + 124 + ], + [ + 428, + 128 + ], + [ + 383, + 144 + ], + [ + 424, + 139 + ], + [ + 381, + 155 + ] + ], + [ + [ + 424, + 128 + ], + [ + 441, + 113 + ], + [ + 426, + 112 + ], + [ + 423, + 142 + ], + [ + 408, + 128 + ], + [ + 406, + 157 + ] + ], + [ + [ + 394, + 122 + ], + [ + 402, + 127 + ], + [ + 402, + 99 + ], + [ + 387, + 146 + ], + [ + 394, + 92 + ], + [ + 378, + 141 + ] + ] + ] + }, + { + "task_id": "0003987", + "id": "RH20T_cfg5_task_0025_user_0010_scene_0006_cfg_0005_104122063678#197", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold a piece of paper'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003987/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 223 + ], + [ + 347, + 203 + ], + [ + 344, + 240 + ], + [ + 332, + 207 + ], + [ + 335, + 259 + ], + [ + 324, + 225 + ] + ], + [ + [ + 336, + 233 + ], + [ + 359, + 235 + ], + [ + 342, + 211 + ], + [ + 331, + 253 + ], + [ + 319, + 208 + ], + [ + 308, + 251 + ] + ], + [ + [ + 334, + 249 + ], + [ + 320, + 249 + ], + [ + 355, + 260 + ], + [ + 315, + 239 + ], + [ + 368, + 260 + ], + [ + 329, + 240 + ] + ], + [ + [ + 326, + 227 + ], + [ + 306, + 216 + ], + [ + 342, + 223 + ], + [ + 312, + 230 + ], + [ + 362, + 234 + ], + [ + 330, + 239 + ] + ] + ] + }, + { + "task_id": "0003988", + "id": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_819112071037#110", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Insert the tip of a medium pipette into the holder for medium pipette tips'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003988/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 269, + 203 + ], + [ + 258, + 248 + ], + [ + 266, + 261 + ], + [ + 272, + 137 + ], + [ + 274, + 223 + ], + [ + 280, + 109 + ] + ], + [ + [ + 268, + 153 + ], + [ + 211, + 133 + ], + [ + 259, + 104 + ], + [ + 275, + 191 + ], + [ + 329, + 127 + ], + [ + 331, + 212 + ] + ], + [ + [ + 305, + 190 + ], + [ + 321, + 162 + ], + [ + 370, + 213 + ], + [ + 241, + 167 + ], + [ + 359, + 253 + ], + [ + 211, + 199 + ] + ], + [ + [ + 280, + 215 + ], + [ + 280, + 154 + ], + [ + 218, + 205 + ], + [ + 345, + 225 + ], + [ + 222, + 258 + ], + [ + 342, + 280 + ] + ] + ] + }, + { + "task_id": "0003990", + "id": "RH20T_cfg1_task_0201_user_0015_scene_0007_cfg_0001_819112071037#293", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Insert the tip of a medium pipette into the holder for medium pipette tips'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003990/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 121 + ], + [ + 344, + 130 + ], + [ + 324, + 91 + ], + [ + 403, + 153 + ], + [ + 343, + 87 + ], + [ + 415, + 143 + ] + ], + [ + [ + 358, + 86 + ], + [ + 353, + 29 + ], + [ + 313, + 100 + ], + [ + 397, + 74 + ], + [ + 320, + 154 + ], + [ + 400, + 124 + ] + ], + [ + [ + 404, + 27 + ], + [ + 448, + 7 + ], + [ + 399, + 0 + ], + [ + 410, + 85 + ], + [ + 363, + 0 + ], + [ + 373, + 95 + ] + ], + [ + [ + 396, + 54 + ], + [ + 443, + 65 + ], + [ + 384, + 0 + ], + [ + 406, + 107 + ], + [ + 345, + 0 + ], + [ + 370, + 94 + ] + ] + ] + }, + { + "task_id": "0003991", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522063145#109", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003991/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 484, + 148 + ], + [ + 490, + 131 + ], + [ + 505, + 158 + ], + [ + 463, + 138 + ], + [ + 498, + 173 + ], + [ + 458, + 154 + ] + ], + [ + [ + 494, + 171 + ], + [ + 484, + 185 + ], + [ + 475, + 175 + ], + [ + 514, + 167 + ], + [ + 485, + 163 + ], + [ + 523, + 154 + ] + ], + [ + [ + 468, + 145 + ], + [ + 457, + 163 + ], + [ + 467, + 134 + ], + [ + 469, + 156 + ], + [ + 479, + 116 + ], + [ + 482, + 136 + ] + ], + [ + [ + 509, + 143 + ], + [ + 491, + 160 + ], + [ + 496, + 133 + ], + [ + 523, + 155 + ], + [ + 513, + 116 + ], + [ + 542, + 137 + ] + ] + ] + }, + { + "task_id": "0003994", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_038522063145#652", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003994/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 452, + 79 + ], + [ + 439, + 91 + ], + [ + 439, + 63 + ], + [ + 465, + 95 + ], + [ + 452, + 52 + ], + [ + 477, + 83 + ] + ], + [ + [ + 427, + 94 + ], + [ + 414, + 102 + ], + [ + 419, + 76 + ], + [ + 435, + 113 + ], + [ + 433, + 69 + ], + [ + 448, + 105 + ] + ], + [ + [ + 445, + 92 + ], + [ + 454, + 77 + ], + [ + 430, + 97 + ], + [ + 462, + 86 + ], + [ + 421, + 111 + ], + [ + 453, + 101 + ] + ], + [ + [ + 433, + 94 + ], + [ + 436, + 77 + ], + [ + 451, + 102 + ], + [ + 417, + 86 + ], + [ + 448, + 118 + ], + [ + 415, + 102 + ] + ] + ] + }, + { + "task_id": "0003995", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0006_cfg_0004_036422060909#105", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0003995/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 377, + 86 + ], + [ + 381, + 62 + ], + [ + 357, + 72 + ], + [ + 399, + 101 + ], + [ + 355, + 94 + ], + [ + 395, + 123 + ] + ], + [ + [ + 374, + 120 + ], + [ + 351, + 114 + ], + [ + 359, + 108 + ], + [ + 390, + 133 + ], + [ + 380, + 113 + ], + [ + 412, + 138 + ] + ], + [ + [ + 359, + 81 + ], + [ + 381, + 68 + ], + [ + 339, + 75 + ], + [ + 381, + 89 + ], + [ + 317, + 88 + ], + [ + 358, + 104 + ] + ], + [ + [ + 379, + 111 + ], + [ + 366, + 135 + ], + [ + 378, + 125 + ], + [ + 380, + 99 + ], + [ + 393, + 98 + ], + [ + 394, + 75 + ] + ] + ] + }, + { + "task_id": "0004000", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122064161#641", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004000/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 430, + 180 + ], + [ + 435, + 164 + ], + [ + 432, + 197 + ], + [ + 427, + 164 + ], + [ + 426, + 212 + ], + [ + 422, + 179 + ] + ], + [ + [ + 438, + 179 + ], + [ + 418, + 176 + ], + [ + 430, + 173 + ], + [ + 446, + 184 + ], + [ + 453, + 175 + ], + [ + 467, + 186 + ] + ], + [ + [ + 413, + 171 + ], + [ + 412, + 186 + ], + [ + 426, + 158 + ], + [ + 398, + 185 + ], + [ + 428, + 143 + ], + [ + 399, + 170 + ] + ], + [ + [ + 438, + 180 + ], + [ + 459, + 168 + ], + [ + 432, + 164 + ], + [ + 445, + 197 + ], + [ + 413, + 175 + ], + [ + 424, + 208 + ] + ] + ] + }, + { + "task_id": "0004002", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0007_cfg_0007_104122064161#881", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004002/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 425, + 149 + ], + [ + 420, + 132 + ], + [ + 414, + 161 + ], + [ + 438, + 135 + ], + [ + 418, + 179 + ], + [ + 444, + 152 + ] + ], + [ + [ + 448, + 148 + ], + [ + 451, + 165 + ], + [ + 447, + 160 + ], + [ + 449, + 136 + ], + [ + 444, + 142 + ], + [ + 446, + 118 + ] + ], + [ + [ + 429, + 135 + ], + [ + 432, + 122 + ], + [ + 428, + 153 + ], + [ + 429, + 117 + ], + [ + 426, + 164 + ], + [ + 426, + 129 + ] + ], + [ + [ + 407, + 155 + ], + [ + 416, + 163 + ], + [ + 410, + 135 + ], + [ + 404, + 174 + ], + [ + 402, + 128 + ], + [ + 396, + 167 + ] + ] + ] + }, + { + "task_id": "0004004", + "id": "RH20T_cfg3_task_0008_user_0016_scene_0009_cfg_0003_104122063550#201", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004004/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 340, + 141 + ], + [ + 340, + 130 + ], + [ + 344, + 164 + ], + [ + 336, + 120 + ], + [ + 344, + 174 + ], + [ + 337, + 131 + ] + ], + [ + [ + 346, + 133 + ], + [ + 330, + 115 + ], + [ + 354, + 115 + ], + [ + 338, + 152 + ], + [ + 368, + 132 + ], + [ + 354, + 169 + ] + ], + [ + [ + 325, + 146 + ], + [ + 303, + 155 + ], + [ + 330, + 162 + ], + [ + 319, + 128 + ], + [ + 351, + 154 + ], + [ + 342, + 119 + ] + ], + [ + [ + 357, + 151 + ], + [ + 344, + 160 + ], + [ + 371, + 171 + ], + [ + 343, + 131 + ], + [ + 386, + 161 + ], + [ + 357, + 120 + ] + ] + ] + }, + { + "task_id": "0004005", + "id": "RH20T_cfg2_task_0016_user_0003_scene_0009_cfg_0002_104422070044#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004005/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 191 + ], + [ + 335, + 179 + ], + [ + 332, + 205 + ], + [ + 345, + 178 + ], + [ + 335, + 216 + ], + [ + 348, + 189 + ] + ], + [ + [ + 345, + 178 + ], + [ + 338, + 170 + ], + [ + 361, + 180 + ], + [ + 329, + 175 + ], + [ + 366, + 188 + ], + [ + 335, + 183 + ] + ], + [ + [ + 330, + 183 + ], + [ + 322, + 196 + ], + [ + 324, + 170 + ], + [ + 336, + 196 + ], + [ + 332, + 158 + ], + [ + 344, + 184 + ] + ], + [ + [ + 323, + 194 + ], + [ + 337, + 202 + ], + [ + 317, + 189 + ], + [ + 329, + 199 + ], + [ + 303, + 181 + ], + [ + 316, + 191 + ] + ] + ] + }, + { + "task_id": "0004006", + "id": "RH20T_cfg1_task_0051_user_0015_scene_0002_cfg_0001_039422060546#129", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the dish off the dish rack'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004006/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 207 + ], + [ + 342, + 221 + ], + [ + 319, + 240 + ], + [ + 308, + 170 + ], + [ + 295, + 226 + ], + [ + 282, + 159 + ] + ], + [ + [ + 284, + 163 + ], + [ + 252, + 182 + ], + [ + 276, + 135 + ], + [ + 293, + 193 + ], + [ + 308, + 116 + ], + [ + 329, + 173 + ] + ], + [ + [ + 277, + 187 + ], + [ + 280, + 177 + ], + [ + 298, + 221 + ], + [ + 257, + 153 + ], + [ + 293, + 227 + ], + [ + 256, + 165 + ] + ], + [ + [ + 282, + 203 + ], + [ + 267, + 168 + ], + [ + 313, + 185 + ], + [ + 248, + 221 + ], + [ + 326, + 217 + ], + [ + 263, + 254 + ] + ] + ] + }, + { + "task_id": "0004009", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0003_cfg_0003_104122062823#213", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004009/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 420, + 181 + ], + [ + 399, + 151 + ], + [ + 426, + 199 + ], + [ + 415, + 166 + ], + [ + 447, + 229 + ], + [ + 433, + 193 + ] + ], + [ + [ + 370, + 125 + ], + [ + 394, + 116 + ], + [ + 381, + 161 + ], + [ + 358, + 89 + ], + [ + 359, + 167 + ], + [ + 338, + 99 + ] + ], + [ + [ + 393, + 133 + ], + [ + 400, + 101 + ], + [ + 432, + 146 + ], + [ + 356, + 121 + ], + [ + 424, + 174 + ], + [ + 352, + 150 + ] + ], + [ + [ + 383, + 134 + ], + [ + 369, + 162 + ], + [ + 347, + 140 + ], + [ + 423, + 128 + ], + [ + 361, + 116 + ], + [ + 434, + 102 + ] + ] + ] + }, + { + "task_id": "0004010", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0003_cfg_0003_104122062823#273", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004010/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 250, + 115 + ], + [ + 221, + 115 + ], + [ + 240, + 80 + ], + [ + 259, + 147 + ], + [ + 272, + 79 + ], + [ + 290, + 148 + ] + ], + [ + [ + 272, + 70 + ], + [ + 268, + 50 + ], + [ + 306, + 57 + ], + [ + 239, + 81 + ], + [ + 313, + 79 + ], + [ + 241, + 104 + ] + ], + [ + [ + 266, + 103 + ], + [ + 262, + 81 + ], + [ + 293, + 86 + ], + [ + 238, + 121 + ], + [ + 294, + 106 + ], + [ + 243, + 140 + ] + ], + [ + [ + 270, + 87 + ], + [ + 272, + 56 + ], + [ + 300, + 102 + ], + [ + 243, + 74 + ], + [ + 297, + 130 + ], + [ + 243, + 101 + ] + ] + ] + }, + { + "task_id": "0004011", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0003_cfg_0003_104122062823#311", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004011/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 216, + 64 + ], + [ + 194, + 74 + ], + [ + 238, + 53 + ], + [ + 196, + 73 + ], + [ + 258, + 44 + ], + [ + 215, + 64 + ] + ], + [ + [ + 188, + 54 + ], + [ + 200, + 85 + ], + [ + 156, + 64 + ], + [ + 219, + 44 + ], + [ + 142, + 31 + ], + [ + 207, + 12 + ] + ], + [ + [ + 203, + 25 + ], + [ + 175, + 4 + ], + [ + 201, + 32 + ], + [ + 205, + 17 + ], + [ + 226, + 51 + ], + [ + 234, + 39 + ] + ], + [ + [ + 195, + 53 + ], + [ + 193, + 24 + ], + [ + 220, + 66 + ], + [ + 172, + 42 + ], + [ + 221, + 93 + ], + [ + 176, + 69 + ] + ] + ] + }, + { + "task_id": "0004012", + "id": "RH20T_cfg4_task_0081_user_0007_scene_0004_cfg_0004_104422070011#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004012/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 286, + 241 + ], + [ + 307, + 263 + ], + [ + 266, + 258 + ], + [ + 305, + 226 + ], + [ + 245, + 236 + ], + [ + 286, + 206 + ] + ], + [ + [ + 287, + 211 + ], + [ + 296, + 197 + ], + [ + 266, + 227 + ], + [ + 307, + 197 + ], + [ + 259, + 239 + ], + [ + 299, + 210 + ] + ], + [ + [ + 276, + 191 + ], + [ + 299, + 207 + ], + [ + 286, + 177 + ], + [ + 267, + 204 + ], + [ + 261, + 160 + ], + [ + 244, + 189 + ] + ], + [ + [ + 297, + 214 + ], + [ + 310, + 218 + ], + [ + 288, + 190 + ], + [ + 307, + 237 + ], + [ + 277, + 187 + ], + [ + 295, + 232 + ] + ] + ] + }, + { + "task_id": "0004015", + "id": "RH20T_cfg4_task_0081_user_0007_scene_0004_cfg_0004_104422070011#271", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004015/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 117 + ], + [ + 322, + 146 + ], + [ + 291, + 120 + ], + [ + 346, + 115 + ], + [ + 288, + 91 + ], + [ + 344, + 87 + ] + ], + [ + [ + 310, + 170 + ], + [ + 334, + 153 + ], + [ + 298, + 154 + ], + [ + 321, + 185 + ], + [ + 272, + 172 + ], + [ + 297, + 202 + ] + ], + [ + [ + 315, + 136 + ], + [ + 327, + 122 + ], + [ + 293, + 148 + ], + [ + 335, + 125 + ], + [ + 283, + 160 + ], + [ + 324, + 138 + ] + ], + [ + [ + 311, + 121 + ], + [ + 294, + 100 + ], + [ + 297, + 137 + ], + [ + 326, + 104 + ], + [ + 313, + 157 + ], + [ + 343, + 126 + ] + ] + ] + }, + { + "task_id": "0004017", + "id": "RH20T_cfg5_task_0004_user_0010_scene_0001_cfg_0005_037522062165#417", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004017/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 147, + 182 + ], + [ + 141, + 167 + ], + [ + 126, + 171 + ], + [ + 168, + 193 + ], + [ + 133, + 185 + ], + [ + 173, + 206 + ] + ], + [ + [ + 150, + 210 + ], + [ + 132, + 230 + ], + [ + 130, + 200 + ], + [ + 171, + 220 + ], + [ + 147, + 181 + ], + [ + 188, + 200 + ] + ], + [ + [ + 144, + 165 + ], + [ + 135, + 178 + ], + [ + 155, + 187 + ], + [ + 133, + 142 + ], + [ + 163, + 175 + ], + [ + 142, + 132 + ] + ], + [ + [ + 136, + 178 + ], + [ + 152, + 161 + ], + [ + 155, + 194 + ], + [ + 116, + 162 + ], + [ + 139, + 212 + ], + [ + 98, + 179 + ] + ] + ] + }, + { + "task_id": "0004018", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_036422060215#380", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004018/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 198 + ], + [ + 273, + 215 + ], + [ + 254, + 191 + ], + [ + 278, + 206 + ], + [ + 245, + 174 + ], + [ + 270, + 188 + ] + ], + [ + [ + 255, + 197 + ], + [ + 259, + 180 + ], + [ + 269, + 212 + ], + [ + 242, + 182 + ], + [ + 265, + 228 + ], + [ + 239, + 197 + ] + ], + [ + [ + 267, + 197 + ], + [ + 289, + 196 + ], + [ + 263, + 178 + ], + [ + 271, + 216 + ], + [ + 241, + 179 + ], + [ + 248, + 217 + ] + ], + [ + [ + 247, + 211 + ], + [ + 268, + 203 + ], + [ + 261, + 229 + ], + [ + 235, + 193 + ], + [ + 239, + 238 + ], + [ + 213, + 201 + ] + ] + ] + }, + { + "task_id": "0004020", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_036422060215#584", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004020/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 193 + ], + [ + 317, + 173 + ], + [ + 287, + 182 + ], + [ + 325, + 203 + ], + [ + 276, + 202 + ], + [ + 315, + 223 + ] + ], + [ + [ + 283, + 174 + ], + [ + 283, + 171 + ], + [ + 283, + 195 + ], + [ + 284, + 154 + ], + [ + 283, + 197 + ], + [ + 284, + 157 + ] + ], + [ + [ + 273, + 182 + ], + [ + 254, + 177 + ], + [ + 277, + 161 + ], + [ + 268, + 204 + ], + [ + 297, + 166 + ], + [ + 287, + 210 + ] + ], + [ + [ + 280, + 185 + ], + [ + 285, + 169 + ], + [ + 290, + 201 + ], + [ + 271, + 170 + ], + [ + 285, + 216 + ], + [ + 267, + 185 + ] + ] + ] + }, + { + "task_id": "0004021", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_036422060215#631", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004021/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 312, + 189 + ], + [ + 321, + 171 + ], + [ + 309, + 180 + ], + [ + 316, + 200 + ], + [ + 300, + 197 + ], + [ + 307, + 219 + ] + ], + [ + [ + 307, + 180 + ], + [ + 295, + 181 + ], + [ + 297, + 163 + ], + [ + 317, + 197 + ], + [ + 308, + 162 + ], + [ + 328, + 195 + ] + ], + [ + [ + 299, + 184 + ], + [ + 304, + 169 + ], + [ + 308, + 200 + ], + [ + 289, + 169 + ], + [ + 303, + 215 + ], + [ + 284, + 184 + ] + ], + [ + [ + 296, + 171 + ], + [ + 315, + 179 + ], + [ + 289, + 179 + ], + [ + 303, + 163 + ], + [ + 270, + 172 + ], + [ + 283, + 155 + ] + ] + ] + }, + { + "task_id": "0004022", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_750612070853#96", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004022/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 256, + 107 + ], + [ + 239, + 81 + ], + [ + 260, + 111 + ], + [ + 250, + 102 + ], + [ + 275, + 135 + ], + [ + 268, + 129 + ] + ], + [ + [ + 278, + 92 + ], + [ + 273, + 63 + ], + [ + 247, + 97 + ], + [ + 308, + 88 + ], + [ + 253, + 124 + ], + [ + 311, + 116 + ] + ], + [ + [ + 243, + 69 + ], + [ + 248, + 53 + ], + [ + 250, + 94 + ], + [ + 236, + 41 + ], + [ + 246, + 113 + ], + [ + 231, + 59 + ] + ], + [ + [ + 272, + 96 + ], + [ + 281, + 77 + ], + [ + 288, + 118 + ], + [ + 257, + 74 + ], + [ + 279, + 134 + ], + [ + 250, + 91 + ] + ] + ] + }, + { + "task_id": "0004025", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_750612070853#291", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004025/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 254, + 129 + ], + [ + 218, + 121 + ], + [ + 257, + 98 + ], + [ + 250, + 157 + ], + [ + 294, + 106 + ], + [ + 284, + 164 + ] + ], + [ + [ + 276, + 155 + ], + [ + 295, + 139 + ], + [ + 298, + 180 + ], + [ + 254, + 129 + ], + [ + 279, + 199 + ], + [ + 232, + 145 + ] + ], + [ + [ + 263, + 99 + ], + [ + 260, + 74 + ], + [ + 297, + 102 + ], + [ + 228, + 95 + ], + [ + 302, + 129 + ], + [ + 230, + 123 + ] + ], + [ + [ + 262, + 140 + ], + [ + 259, + 108 + ], + [ + 226, + 141 + ], + [ + 297, + 138 + ], + [ + 231, + 170 + ], + [ + 298, + 167 + ] + ] + ] + }, + { + "task_id": "0004027", + "id": "RH20T_cfg4_task_0057_user_0014_scene_0009_cfg_0004_038522062288#178", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004027/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 241, + 163 + ], + [ + 240, + 140 + ], + [ + 217, + 153 + ], + [ + 265, + 173 + ], + [ + 219, + 174 + ], + [ + 265, + 194 + ] + ], + [ + [ + 221, + 176 + ], + [ + 219, + 189 + ], + [ + 197, + 187 + ], + [ + 246, + 164 + ], + [ + 201, + 175 + ], + [ + 248, + 153 + ] + ], + [ + [ + 251, + 145 + ], + [ + 266, + 159 + ], + [ + 246, + 170 + ], + [ + 257, + 119 + ], + [ + 232, + 156 + ], + [ + 242, + 107 + ] + ], + [ + [ + 269, + 144 + ], + [ + 255, + 128 + ], + [ + 283, + 161 + ], + [ + 257, + 127 + ], + [ + 296, + 175 + ], + [ + 270, + 142 + ] + ] + ] + }, + { + "task_id": "0004029", + "id": "RH20T_cfg4_task_0057_user_0014_scene_0009_cfg_0004_038522062288#287", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004029/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 180, + 165 + ], + [ + 156, + 178 + ], + [ + 169, + 137 + ], + [ + 192, + 192 + ], + [ + 192, + 127 + ], + [ + 213, + 180 + ] + ], + [ + [ + 230, + 185 + ], + [ + 216, + 158 + ], + [ + 216, + 186 + ], + [ + 242, + 183 + ], + [ + 231, + 215 + ], + [ + 255, + 208 + ] + ], + [ + [ + 198, + 193 + ], + [ + 183, + 188 + ], + [ + 219, + 172 + ], + [ + 178, + 215 + ], + [ + 231, + 177 + ], + [ + 192, + 218 + ] + ], + [ + [ + 207, + 185 + ], + [ + 202, + 162 + ], + [ + 182, + 176 + ], + [ + 232, + 194 + ], + [ + 188, + 197 + ], + [ + 236, + 216 + ] + ] + ] + }, + { + "task_id": "0004036", + "id": "RH20T_cfg1_task_0037_user_0016_scene_0009_cfg_0001_035622060973#644", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004036/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 198, + 232 + ], + [ + 197, + 223 + ], + [ + 188, + 213 + ], + [ + 208, + 251 + ], + [ + 189, + 222 + ], + [ + 208, + 258 + ] + ], + [ + [ + 199, + 234 + ], + [ + 204, + 226 + ], + [ + 218, + 224 + ], + [ + 180, + 243 + ], + [ + 213, + 232 + ], + [ + 176, + 250 + ] + ], + [ + [ + 212, + 225 + ], + [ + 231, + 214 + ], + [ + 225, + 232 + ], + [ + 198, + 217 + ], + [ + 207, + 242 + ], + [ + 179, + 229 + ] + ], + [ + [ + 214, + 213 + ], + [ + 216, + 228 + ], + [ + 228, + 199 + ], + [ + 199, + 228 + ], + [ + 226, + 183 + ], + [ + 196, + 213 + ] + ] + ] + }, + { + "task_id": "0004037", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_104122061602#197", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004037/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 508, + 163 + ], + [ + 509, + 139 + ], + [ + 496, + 148 + ], + [ + 518, + 177 + ], + [ + 494, + 175 + ], + [ + 517, + 202 + ] + ], + [ + [ + 512, + 200 + ], + [ + 512, + 228 + ], + [ + 481, + 196 + ], + [ + 543, + 204 + ], + [ + 482, + 170 + ], + [ + 541, + 177 + ] + ], + [ + [ + 498, + 195 + ], + [ + 522, + 196 + ], + [ + 512, + 214 + ], + [ + 484, + 175 + ], + [ + 491, + 212 + ], + [ + 463, + 175 + ] + ], + [ + [ + 515, + 223 + ], + [ + 537, + 208 + ], + [ + 506, + 238 + ], + [ + 523, + 210 + ], + [ + 486, + 252 + ], + [ + 504, + 224 + ] + ] + ] + }, + { + "task_id": "0004038", + "id": "RH20T_cfg5_task_0096_user_0007_scene_0008_cfg_0005_104422070044#108", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004038/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 451, + 150 + ], + [ + 434, + 158 + ], + [ + 449, + 164 + ], + [ + 454, + 136 + ], + [ + 465, + 156 + ], + [ + 471, + 127 + ] + ], + [ + [ + 457, + 162 + ], + [ + 458, + 145 + ], + [ + 471, + 173 + ], + [ + 443, + 151 + ], + [ + 469, + 189 + ], + [ + 442, + 167 + ] + ], + [ + [ + 466, + 170 + ], + [ + 457, + 181 + ], + [ + 470, + 189 + ], + [ + 462, + 153 + ], + [ + 480, + 177 + ], + [ + 471, + 141 + ] + ], + [ + [ + 437, + 150 + ], + [ + 434, + 164 + ], + [ + 437, + 165 + ], + [ + 437, + 133 + ], + [ + 440, + 152 + ], + [ + 440, + 120 + ] + ] + ] + }, + { + "task_id": "0004041", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0010_cfg_0005_104422070044#91", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004041/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 422, + 144 + ], + [ + 437, + 134 + ], + [ + 434, + 157 + ], + [ + 410, + 130 + ], + [ + 419, + 167 + ], + [ + 394, + 139 + ] + ], + [ + [ + 437, + 135 + ], + [ + 427, + 150 + ], + [ + 447, + 146 + ], + [ + 426, + 123 + ], + [ + 456, + 132 + ], + [ + 436, + 108 + ] + ], + [ + [ + 430, + 142 + ], + [ + 430, + 129 + ], + [ + 442, + 152 + ], + [ + 418, + 133 + ], + [ + 442, + 165 + ], + [ + 419, + 145 + ] + ], + [ + [ + 427, + 131 + ], + [ + 438, + 120 + ], + [ + 411, + 122 + ], + [ + 443, + 139 + ], + [ + 402, + 132 + ], + [ + 433, + 149 + ] + ] + ] + }, + { + "task_id": "0004043", + "id": "RH20T_cfg7_task_0013_user_0020_scene_0004_cfg_0007_104122064161#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004043/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 372, + 153 + ], + [ + 374, + 140 + ], + [ + 371, + 171 + ], + [ + 373, + 135 + ], + [ + 369, + 183 + ], + [ + 371, + 148 + ] + ], + [ + [ + 378, + 168 + ], + [ + 396, + 162 + ], + [ + 365, + 160 + ], + [ + 393, + 177 + ], + [ + 346, + 165 + ], + [ + 374, + 184 + ] + ], + [ + [ + 382, + 138 + ], + [ + 368, + 130 + ], + [ + 376, + 123 + ], + [ + 389, + 152 + ], + [ + 391, + 132 + ], + [ + 404, + 160 + ] + ], + [ + [ + 359, + 167 + ], + [ + 352, + 183 + ], + [ + 364, + 155 + ], + [ + 353, + 179 + ], + [ + 371, + 138 + ], + [ + 361, + 161 + ] + ] + ] + }, + { + "task_id": "0004045", + "id": "RH20T_cfg6_task_0006_user_0014_scene_0005_cfg_0006_104122063633#207", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004045/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 201, + 153 + ], + [ + 222, + 161 + ], + [ + 199, + 149 + ], + [ + 204, + 158 + ], + [ + 179, + 142 + ], + [ + 181, + 150 + ] + ], + [ + [ + 202, + 143 + ], + [ + 218, + 153 + ], + [ + 207, + 156 + ], + [ + 196, + 128 + ], + [ + 191, + 146 + ], + [ + 180, + 118 + ] + ], + [ + [ + 203, + 161 + ], + [ + 202, + 141 + ], + [ + 210, + 151 + ], + [ + 196, + 172 + ], + [ + 211, + 169 + ], + [ + 197, + 192 + ] + ], + [ + [ + 218, + 169 + ], + [ + 222, + 170 + ], + [ + 211, + 192 + ], + [ + 224, + 147 + ], + [ + 205, + 193 + ], + [ + 219, + 145 + ] + ] + ] + }, + { + "task_id": "0004046", + "id": "RH20T_cfg4_task_0069_user_0007_scene_0006_cfg_0004_036422060909#91", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Assemble one piece of a puzzle'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004046/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 277, + 127 + ], + [ + 260, + 145 + ], + [ + 259, + 110 + ], + [ + 295, + 143 + ], + [ + 275, + 93 + ], + [ + 311, + 126 + ] + ], + [ + [ + 278, + 123 + ], + [ + 281, + 105 + ], + [ + 265, + 109 + ], + [ + 292, + 139 + ], + [ + 264, + 126 + ], + [ + 289, + 156 + ] + ], + [ + [ + 287, + 133 + ], + [ + 309, + 127 + ], + [ + 281, + 116 + ], + [ + 293, + 149 + ], + [ + 259, + 123 + ], + [ + 271, + 155 + ] + ], + [ + [ + 289, + 114 + ], + [ + 274, + 131 + ], + [ + 292, + 127 + ], + [ + 287, + 100 + ], + [ + 306, + 111 + ], + [ + 302, + 83 + ] + ] + ] + }, + { + "task_id": "0004050", + "id": "RH20T_cfg2_task_0008_user_0007_scene_0004_cfg_0002_036422060215#104", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004050/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 289, + 221 + ], + [ + 292, + 204 + ], + [ + 269, + 221 + ], + [ + 310, + 221 + ], + [ + 266, + 239 + ], + [ + 308, + 239 + ] + ], + [ + [ + 283, + 217 + ], + [ + 284, + 229 + ], + [ + 303, + 215 + ], + [ + 262, + 220 + ], + [ + 302, + 202 + ], + [ + 260, + 207 + ] + ], + [ + [ + 281, + 209 + ], + [ + 286, + 195 + ], + [ + 300, + 205 + ], + [ + 262, + 213 + ], + [ + 295, + 218 + ], + [ + 258, + 227 + ] + ], + [ + [ + 281, + 208 + ], + [ + 269, + 201 + ], + [ + 282, + 187 + ], + [ + 280, + 227 + ], + [ + 295, + 193 + ], + [ + 292, + 235 + ] + ] + ] + }, + { + "task_id": "0004054", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122061018#313", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004054/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 313, + 169 + ], + [ + 301, + 143 + ], + [ + 308, + 184 + ], + [ + 319, + 151 + ], + [ + 319, + 210 + ], + [ + 333, + 179 + ] + ], + [ + [ + 324, + 161 + ], + [ + 343, + 136 + ], + [ + 328, + 161 + ], + [ + 320, + 161 + ], + [ + 310, + 184 + ], + [ + 299, + 187 + ] + ], + [ + [ + 313, + 175 + ], + [ + 325, + 203 + ], + [ + 334, + 163 + ], + [ + 289, + 189 + ], + [ + 322, + 135 + ], + [ + 275, + 160 + ] + ], + [ + [ + 308, + 166 + ], + [ + 298, + 172 + ], + [ + 337, + 171 + ], + [ + 278, + 162 + ], + [ + 349, + 165 + ], + [ + 287, + 154 + ] + ] + ] + }, + { + "task_id": "0004055", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122061018#466", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004055/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 282, + 146 + ], + [ + 295, + 134 + ], + [ + 271, + 118 + ], + [ + 294, + 175 + ], + [ + 261, + 130 + ], + [ + 283, + 184 + ] + ], + [ + [ + 312, + 152 + ], + [ + 283, + 156 + ], + [ + 318, + 141 + ], + [ + 306, + 164 + ], + [ + 347, + 138 + ], + [ + 338, + 161 + ] + ], + [ + [ + 281, + 165 + ], + [ + 275, + 196 + ], + [ + 302, + 177 + ], + [ + 261, + 154 + ], + [ + 309, + 143 + ], + [ + 266, + 122 + ] + ], + [ + [ + 277, + 134 + ], + [ + 257, + 156 + ], + [ + 271, + 132 + ], + [ + 283, + 137 + ], + [ + 290, + 112 + ], + [ + 304, + 113 + ] + ] + ] + }, + { + "task_id": "0004056", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0002_cfg_0007_104122061018#567", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004056/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 137 + ], + [ + 314, + 107 + ], + [ + 302, + 129 + ], + [ + 331, + 146 + ], + [ + 303, + 155 + ], + [ + 331, + 175 + ] + ], + [ + [ + 316, + 123 + ], + [ + 299, + 143 + ], + [ + 295, + 128 + ], + [ + 339, + 116 + ], + [ + 311, + 110 + ], + [ + 355, + 97 + ] + ], + [ + [ + 294, + 128 + ], + [ + 311, + 110 + ], + [ + 269, + 117 + ], + [ + 317, + 138 + ], + [ + 253, + 136 + ], + [ + 300, + 154 + ] + ], + [ + [ + 333, + 150 + ], + [ + 328, + 157 + ], + [ + 320, + 125 + ], + [ + 346, + 175 + ], + [ + 325, + 114 + ], + [ + 352, + 168 + ] + ] + ] + }, + { + "task_id": "0004058", + "id": "RH20T_cfg1_task_0061_user_0015_scene_0001_cfg_0001_819112070044#770", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004058/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 312, + 236 + ], + [ + 308, + 252 + ], + [ + 314, + 255 + ], + [ + 310, + 219 + ], + [ + 319, + 238 + ], + [ + 314, + 202 + ] + ], + [ + [ + 322, + 259 + ], + [ + 326, + 241 + ], + [ + 333, + 246 + ], + [ + 309, + 274 + ], + [ + 330, + 262 + ], + [ + 306, + 292 + ] + ], + [ + [ + 326, + 240 + ], + [ + 344, + 224 + ], + [ + 324, + 230 + ], + [ + 327, + 251 + ], + [ + 307, + 245 + ], + [ + 308, + 268 + ] + ], + [ + [ + 347, + 250 + ], + [ + 336, + 236 + ], + [ + 341, + 270 + ], + [ + 354, + 229 + ], + [ + 352, + 286 + ], + [ + 367, + 244 + ] + ] + ] + }, + { + "task_id": "0004059", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0002_cfg_0005_037522062165#109", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004059/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 353, + 221 + ], + [ + 326, + 206 + ], + [ + 336, + 229 + ], + [ + 369, + 215 + ], + [ + 366, + 247 + ], + [ + 397, + 230 + ] + ], + [ + [ + 341, + 241 + ], + [ + 344, + 227 + ], + [ + 326, + 215 + ], + [ + 357, + 269 + ], + [ + 324, + 228 + ], + [ + 353, + 280 + ] + ], + [ + [ + 321, + 231 + ], + [ + 290, + 239 + ], + [ + 315, + 222 + ], + [ + 328, + 241 + ], + [ + 344, + 215 + ], + [ + 360, + 232 + ] + ], + [ + [ + 355, + 255 + ], + [ + 323, + 257 + ], + [ + 354, + 286 + ], + [ + 356, + 223 + ], + [ + 384, + 285 + ], + [ + 387, + 222 + ] + ] + ] + }, + { + "task_id": "0004060", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0002_cfg_0005_037522062165#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004060/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 273 + ], + [ + 335, + 260 + ], + [ + 318, + 303 + ], + [ + 292, + 245 + ], + [ + 286, + 318 + ], + [ + 262, + 258 + ] + ], + [ + [ + 357, + 260 + ], + [ + 389, + 265 + ], + [ + 344, + 269 + ], + [ + 372, + 250 + ], + [ + 313, + 265 + ], + [ + 336, + 245 + ] + ], + [ + [ + 350, + 250 + ], + [ + 368, + 266 + ], + [ + 378, + 243 + ], + [ + 321, + 258 + ], + [ + 361, + 230 + ], + [ + 306, + 243 + ] + ], + [ + [ + 332, + 273 + ], + [ + 336, + 261 + ], + [ + 318, + 244 + ], + [ + 347, + 304 + ], + [ + 315, + 256 + ], + [ + 342, + 313 + ] + ] + ] + }, + { + "task_id": "0004064", + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_104122064161#83", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004064/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 374, + 199 + ], + [ + 380, + 196 + ], + [ + 373, + 225 + ], + [ + 375, + 174 + ], + [ + 368, + 227 + ], + [ + 370, + 178 + ] + ], + [ + [ + 373, + 201 + ], + [ + 348, + 204 + ], + [ + 380, + 218 + ], + [ + 366, + 185 + ], + [ + 405, + 216 + ], + [ + 389, + 183 + ] + ], + [ + [ + 377, + 190 + ], + [ + 357, + 198 + ], + [ + 380, + 175 + ], + [ + 373, + 207 + ], + [ + 401, + 166 + ], + [ + 396, + 198 + ] + ], + [ + [ + 348, + 208 + ], + [ + 332, + 192 + ], + [ + 338, + 227 + ], + [ + 359, + 188 + ], + [ + 354, + 244 + ], + [ + 376, + 205 + ] + ] + ] + }, + { + "task_id": "0004070", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0004_cfg_0005_104122063678#1127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004070/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 328, + 135 + ], + [ + 321, + 155 + ], + [ + 311, + 124 + ], + [ + 345, + 145 + ], + [ + 319, + 104 + ], + [ + 352, + 125 + ] + ], + [ + [ + 335, + 168 + ], + [ + 356, + 173 + ], + [ + 339, + 150 + ], + [ + 331, + 185 + ], + [ + 317, + 145 + ], + [ + 311, + 181 + ] + ], + [ + [ + 334, + 147 + ], + [ + 337, + 132 + ], + [ + 331, + 163 + ], + [ + 336, + 132 + ], + [ + 328, + 177 + ], + [ + 333, + 146 + ] + ], + [ + [ + 333, + 152 + ], + [ + 313, + 155 + ], + [ + 329, + 133 + ], + [ + 337, + 171 + ], + [ + 351, + 131 + ], + [ + 358, + 169 + ] + ] + ] + }, + { + "task_id": "0004072", + "id": "RH20T_cfg4_task_0069_user_0007_scene_0006_cfg_0004_104122062823#183", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Assemble one piece of a puzzle'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004072/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 86 + ], + [ + 337, + 60 + ], + [ + 350, + 73 + ], + [ + 317, + 100 + ], + [ + 347, + 96 + ], + [ + 315, + 124 + ] + ], + [ + [ + 316, + 111 + ], + [ + 298, + 110 + ], + [ + 318, + 81 + ], + [ + 315, + 141 + ], + [ + 339, + 80 + ], + [ + 335, + 144 + ] + ], + [ + [ + 329, + 91 + ], + [ + 310, + 75 + ], + [ + 321, + 70 + ], + [ + 336, + 109 + ], + [ + 342, + 87 + ], + [ + 355, + 125 + ] + ], + [ + [ + 324, + 80 + ], + [ + 321, + 102 + ], + [ + 295, + 74 + ], + [ + 352, + 85 + ], + [ + 300, + 53 + ], + [ + 355, + 64 + ] + ] + ] + }, + { + "task_id": "0004073", + "id": "RH20T_cfg4_task_0069_user_0007_scene_0006_cfg_0004_104122062823#219", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Assemble one piece of a puzzle'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004073/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 44 + ], + [ + 311, + 31 + ], + [ + 318, + 47 + ], + [ + 347, + 42 + ], + [ + 343, + 61 + ], + [ + 370, + 55 + ] + ], + [ + [ + 324, + 78 + ], + [ + 328, + 53 + ], + [ + 341, + 67 + ], + [ + 305, + 90 + ], + [ + 337, + 90 + ], + [ + 302, + 113 + ] + ], + [ + [ + 318, + 76 + ], + [ + 334, + 64 + ], + [ + 308, + 50 + ], + [ + 327, + 101 + ], + [ + 289, + 62 + ], + [ + 310, + 114 + ] + ], + [ + [ + 333, + 112 + ], + [ + 343, + 85 + ], + [ + 355, + 117 + ], + [ + 312, + 107 + ], + [ + 344, + 145 + ], + [ + 303, + 132 + ] + ] + ] + }, + { + "task_id": "0004080", + "id": "RH20T_cfg1_task_0038_user_0016_scene_0009_cfg_0001_750612070851#123", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004080/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 224 + ], + [ + 365, + 245 + ], + [ + 418, + 240 + ], + [ + 366, + 207 + ], + [ + 448, + 220 + ], + [ + 396, + 183 + ] + ], + [ + [ + 386, + 232 + ], + [ + 386, + 201 + ], + [ + 346, + 237 + ], + [ + 426, + 227 + ], + [ + 348, + 264 + ], + [ + 424, + 255 + ] + ], + [ + [ + 359, + 250 + ], + [ + 317, + 261 + ], + [ + 352, + 219 + ], + [ + 368, + 285 + ], + [ + 391, + 209 + ], + [ + 414, + 274 + ] + ], + [ + [ + 345, + 241 + ], + [ + 320, + 227 + ], + [ + 369, + 211 + ], + [ + 323, + 270 + ], + [ + 391, + 225 + ], + [ + 347, + 281 + ] + ] + ] + }, + { + "task_id": "0004081", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0007_cfg_0003_104122062295#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004081/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 345, + 241 + ], + [ + 339, + 268 + ], + [ + 385, + 245 + ], + [ + 307, + 237 + ], + [ + 387, + 221 + ], + [ + 314, + 214 + ] + ], + [ + [ + 326, + 242 + ], + [ + 343, + 235 + ], + [ + 311, + 208 + ], + [ + 341, + 277 + ], + [ + 291, + 214 + ], + [ + 322, + 287 + ] + ], + [ + [ + 325, + 215 + ], + [ + 333, + 188 + ], + [ + 294, + 198 + ], + [ + 358, + 233 + ], + [ + 288, + 223 + ], + [ + 349, + 257 + ] + ], + [ + [ + 343, + 228 + ], + [ + 324, + 253 + ], + [ + 372, + 213 + ], + [ + 318, + 241 + ], + [ + 388, + 190 + ], + [ + 335, + 219 + ] + ] + ] + }, + { + "task_id": "0004082", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0007_cfg_0003_104122062295#176", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004082/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 235 + ], + [ + 322, + 206 + ], + [ + 281, + 217 + ], + [ + 348, + 254 + ], + [ + 275, + 243 + ], + [ + 339, + 279 + ] + ], + [ + [ + 358, + 206 + ], + [ + 383, + 209 + ], + [ + 358, + 167 + ], + [ + 358, + 246 + ], + [ + 331, + 162 + ], + [ + 329, + 246 + ] + ], + [ + [ + 306, + 230 + ], + [ + 324, + 242 + ], + [ + 297, + 200 + ], + [ + 316, + 263 + ], + [ + 277, + 185 + ], + [ + 294, + 250 + ] + ], + [ + [ + 310, + 233 + ], + [ + 301, + 212 + ], + [ + 327, + 255 + ], + [ + 291, + 210 + ], + [ + 337, + 279 + ], + [ + 299, + 233 + ] + ] + ] + }, + { + "task_id": "0004084", + "id": "RH20T_cfg3_task_0057_user_0010_scene_0007_cfg_0003_104122062295#240", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004084/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 318, + 298 + ], + [ + 352, + 271 + ], + [ + 319, + 298 + ], + [ + 318, + 299 + ], + [ + 280, + 330 + ], + [ + 285, + 325 + ] + ], + [ + [ + 283, + 344 + ], + [ + 304, + 332 + ], + [ + 248, + 310 + ], + [ + 317, + 360 + ], + [ + 232, + 323 + ], + [ + 296, + 360 + ] + ], + [ + [ + 286, + 281 + ], + [ + 294, + 246 + ], + [ + 249, + 261 + ], + [ + 326, + 302 + ], + [ + 244, + 292 + ], + [ + 317, + 333 + ] + ], + [ + [ + 245, + 289 + ], + [ + 268, + 261 + ], + [ + 275, + 281 + ], + [ + 218, + 295 + ], + [ + 249, + 315 + ], + [ + 191, + 326 + ] + ] + ] + }, + { + "task_id": "0004085", + "id": "RH20T_cfg6_task_0008_user_0014_scene_0007_cfg_0006_104122061018#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004085/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 297, + 89 + ], + [ + 302, + 66 + ], + [ + 284, + 70 + ], + [ + 311, + 109 + ], + [ + 280, + 91 + ], + [ + 306, + 129 + ] + ], + [ + [ + 306, + 76 + ], + [ + 303, + 87 + ], + [ + 283, + 58 + ], + [ + 327, + 94 + ], + [ + 287, + 49 + ], + [ + 329, + 83 + ] + ], + [ + [ + 280, + 82 + ], + [ + 305, + 71 + ], + [ + 292, + 110 + ], + [ + 267, + 54 + ], + [ + 268, + 120 + ], + [ + 242, + 65 + ] + ], + [ + [ + 311, + 73 + ], + [ + 292, + 51 + ], + [ + 307, + 75 + ], + [ + 315, + 70 + ], + [ + 324, + 95 + ], + [ + 335, + 93 + ] + ] + ] + }, + { + "task_id": "0004087", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_750612070853#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004087/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 429, + 285 + ], + [ + 484, + 280 + ], + [ + 443, + 332 + ], + [ + 418, + 246 + ], + [ + 383, + 338 + ], + [ + 368, + 251 + ] + ], + [ + [ + 401, + 236 + ], + [ + 399, + 197 + ], + [ + 357, + 219 + ], + [ + 449, + 256 + ], + [ + 361, + 252 + ], + [ + 447, + 290 + ] + ], + [ + [ + 399, + 226 + ], + [ + 369, + 263 + ], + [ + 366, + 179 + ], + [ + 431, + 271 + ], + [ + 394, + 151 + ], + [ + 453, + 237 + ] + ], + [ + [ + 431, + 198 + ], + [ + 404, + 232 + ], + [ + 384, + 175 + ], + [ + 482, + 224 + ], + [ + 409, + 148 + ], + [ + 502, + 191 + ] + ] + ] + }, + { + "task_id": "0004088", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_750612070853#307", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004088/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 369, + 185 + ], + [ + 389, + 231 + ], + [ + 324, + 199 + ], + [ + 410, + 173 + ], + [ + 306, + 153 + ], + [ + 390, + 131 + ] + ], + [ + [ + 363, + 248 + ], + [ + 391, + 260 + ], + [ + 372, + 207 + ], + [ + 353, + 291 + ], + [ + 342, + 191 + ], + [ + 320, + 279 + ] + ], + [ + [ + 347, + 176 + ], + [ + 355, + 159 + ], + [ + 309, + 189 + ], + [ + 388, + 163 + ], + [ + 296, + 209 + ], + [ + 382, + 183 + ] + ], + [ + [ + 375, + 196 + ], + [ + 376, + 159 + ], + [ + 331, + 192 + ], + [ + 420, + 200 + ], + [ + 333, + 225 + ], + [ + 416, + 234 + ] + ] + ] + }, + { + "task_id": "0004091", + "id": "RH20T_cfg1_task_0120_user_0014_scene_0002_cfg_0001_038522062547#114", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004091/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 398, + 136 + ], + [ + 399, + 146 + ], + [ + 422, + 147 + ], + [ + 374, + 124 + ], + [ + 420, + 138 + ], + [ + 374, + 115 + ] + ], + [ + [ + 408, + 121 + ], + [ + 406, + 99 + ], + [ + 431, + 131 + ], + [ + 388, + 113 + ], + [ + 432, + 151 + ], + [ + 391, + 132 + ] + ], + [ + [ + 412, + 123 + ], + [ + 396, + 115 + ], + [ + 404, + 104 + ], + [ + 420, + 141 + ], + [ + 422, + 112 + ], + [ + 437, + 150 + ] + ], + [ + [ + 420, + 110 + ], + [ + 438, + 97 + ], + [ + 444, + 115 + ], + [ + 398, + 105 + ], + [ + 424, + 129 + ], + [ + 379, + 117 + ] + ] + ] + }, + { + "task_id": "0004093", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0004_cfg_0006_104122060811#463", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004093/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 374, + 122 + ], + [ + 387, + 117 + ], + [ + 369, + 109 + ], + [ + 379, + 135 + ], + [ + 356, + 114 + ], + [ + 367, + 140 + ] + ], + [ + [ + 375, + 132 + ], + [ + 381, + 140 + ], + [ + 363, + 134 + ], + [ + 388, + 129 + ], + [ + 356, + 126 + ], + [ + 381, + 121 + ] + ], + [ + [ + 359, + 139 + ], + [ + 367, + 127 + ], + [ + 357, + 137 + ], + [ + 361, + 141 + ], + [ + 349, + 148 + ], + [ + 353, + 154 + ] + ], + [ + [ + 368, + 131 + ], + [ + 382, + 129 + ], + [ + 366, + 139 + ], + [ + 370, + 123 + ], + [ + 353, + 141 + ], + [ + 356, + 125 + ] + ] + ] + }, + { + "task_id": "0004094", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0004_cfg_0006_104122060811#523", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004094/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 125 + ], + [ + 377, + 121 + ], + [ + 362, + 134 + ], + [ + 364, + 115 + ], + [ + 349, + 138 + ], + [ + 349, + 119 + ] + ], + [ + [ + 366, + 109 + ], + [ + 369, + 95 + ], + [ + 360, + 104 + ], + [ + 371, + 115 + ], + [ + 357, + 118 + ], + [ + 367, + 129 + ] + ], + [ + [ + 354, + 125 + ], + [ + 358, + 137 + ], + [ + 357, + 116 + ], + [ + 351, + 134 + ], + [ + 352, + 105 + ], + [ + 346, + 122 + ] + ], + [ + [ + 350, + 118 + ], + [ + 345, + 104 + ], + [ + 350, + 122 + ], + [ + 350, + 113 + ], + [ + 354, + 135 + ], + [ + 354, + 128 + ] + ] + ] + }, + { + "task_id": "0004098", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0002_cfg_0005_036422060215#410", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004098/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 166, + 180 + ], + [ + 185, + 196 + ], + [ + 169, + 182 + ], + [ + 162, + 178 + ], + [ + 150, + 166 + ], + [ + 142, + 160 + ] + ], + [ + [ + 180, + 139 + ], + [ + 167, + 119 + ], + [ + 198, + 124 + ], + [ + 162, + 154 + ], + [ + 212, + 145 + ], + [ + 175, + 174 + ] + ], + [ + [ + 188, + 171 + ], + [ + 187, + 151 + ], + [ + 213, + 174 + ], + [ + 162, + 168 + ], + [ + 213, + 193 + ], + [ + 164, + 187 + ] + ], + [ + [ + 169, + 176 + ], + [ + 189, + 180 + ], + [ + 183, + 159 + ], + [ + 155, + 193 + ], + [ + 164, + 156 + ], + [ + 135, + 189 + ] + ] + ] + }, + { + "task_id": "0004100", + "id": "RH20T_cfg7_task_0025_user_0020_scene_0008_cfg_0007_104122060811#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold a piece of paper'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004100/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 351, + 151 + ], + [ + 357, + 140 + ], + [ + 362, + 159 + ], + [ + 340, + 142 + ], + [ + 356, + 170 + ], + [ + 335, + 153 + ] + ], + [ + [ + 358, + 142 + ], + [ + 358, + 154 + ], + [ + 346, + 146 + ], + [ + 371, + 139 + ], + [ + 346, + 135 + ], + [ + 371, + 128 + ] + ], + [ + [ + 361, + 150 + ], + [ + 358, + 146 + ], + [ + 360, + 164 + ], + [ + 362, + 136 + ], + [ + 363, + 169 + ], + [ + 365, + 140 + ] + ], + [ + [ + 364, + 143 + ], + [ + 374, + 136 + ], + [ + 373, + 145 + ], + [ + 354, + 140 + ], + [ + 363, + 152 + ], + [ + 344, + 147 + ] + ] + ] + }, + { + "task_id": "0004101", + "id": "RH20T_cfg3_task_0008_user_0016_scene_0005_cfg_0003_104122062295#319", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004101/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 43, + 200 + ], + [ + 37, + 163 + ], + [ + 28, + 177 + ], + [ + 59, + 227 + ], + [ + 33, + 209 + ], + [ + 63, + 261 + ] + ], + [ + [ + 29, + 212 + ], + [ + 80, + 206 + ], + [ + 18, + 178 + ], + [ + 42, + 249 + ], + [ + 0, + 181 + ], + [ + 0, + 259 + ] + ], + [ + [ + 52, + 207 + ], + [ + 3, + 219 + ], + [ + 40, + 173 + ], + [ + 64, + 243 + ], + [ + 83, + 165 + ], + [ + 109, + 231 + ] + ], + [ + [ + 19, + 176 + ], + [ + 0, + 133 + ], + [ + 0, + 194 + ], + [ + 67, + 158 + ], + [ + 0, + 237 + ], + [ + 86, + 199 + ] + ] + ] + }, + { + "task_id": "0004102", + "id": "RH20T_cfg4_task_0091_user_0007_scene_0006_cfg_0004_104122062823#104", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004102/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 407, + 92 + ], + [ + 430, + 108 + ], + [ + 439, + 71 + ], + [ + 375, + 113 + ], + [ + 417, + 58 + ], + [ + 357, + 97 + ] + ], + [ + [ + 378, + 123 + ], + [ + 384, + 92 + ], + [ + 415, + 121 + ], + [ + 340, + 126 + ], + [ + 409, + 149 + ], + [ + 338, + 154 + ] + ], + [ + [ + 396, + 136 + ], + [ + 430, + 114 + ], + [ + 383, + 110 + ], + [ + 410, + 165 + ], + [ + 353, + 131 + ], + [ + 376, + 186 + ] + ], + [ + [ + 375, + 97 + ], + [ + 370, + 70 + ], + [ + 368, + 67 + ], + [ + 382, + 123 + ], + [ + 374, + 97 + ], + [ + 387, + 151 + ] + ] + ] + }, + { + "task_id": "0004103", + "id": "RH20T_cfg4_task_0009_user_0010_scene_0004_cfg_0004_104122062295#106", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004103/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 185, + 247 + ], + [ + 195, + 202 + ], + [ + 136, + 240 + ], + [ + 233, + 253 + ], + [ + 122, + 289 + ], + [ + 224, + 302 + ] + ], + [ + [ + 178, + 150 + ], + [ + 224, + 147 + ], + [ + 163, + 124 + ], + [ + 196, + 180 + ], + [ + 114, + 126 + ], + [ + 141, + 186 + ] + ], + [ + [ + 200, + 181 + ], + [ + 203, + 142 + ], + [ + 166, + 161 + ], + [ + 236, + 203 + ], + [ + 165, + 196 + ], + [ + 232, + 237 + ] + ], + [ + [ + 217, + 214 + ], + [ + 264, + 203 + ], + [ + 211, + 169 + ], + [ + 224, + 261 + ], + [ + 165, + 179 + ], + [ + 175, + 272 + ] + ] + ] + }, + { + "task_id": "0004108", + "id": "RH20T_cfg2_task_0042_user_0007_scene_0002_cfg_0002_104122063678#228", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004108/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 328, + 186 + ], + [ + 306, + 192 + ], + [ + 333, + 176 + ], + [ + 322, + 196 + ], + [ + 355, + 169 + ], + [ + 345, + 190 + ] + ], + [ + [ + 348, + 170 + ], + [ + 324, + 178 + ], + [ + 340, + 146 + ], + [ + 355, + 194 + ], + [ + 364, + 139 + ], + [ + 378, + 187 + ] + ], + [ + [ + 342, + 191 + ], + [ + 339, + 167 + ], + [ + 347, + 199 + ], + [ + 339, + 183 + ], + [ + 350, + 223 + ], + [ + 342, + 205 + ] + ], + [ + [ + 358, + 174 + ], + [ + 367, + 190 + ], + [ + 335, + 174 + ], + [ + 381, + 174 + ], + [ + 325, + 158 + ], + [ + 372, + 157 + ] + ] + ] + }, + { + "task_id": "0004109", + "id": "RH20T_cfg1_task_0015_user_0008_scene_0008_cfg_0001_038522063145#128", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004109/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 524, + 163 + ], + [ + 531, + 142 + ], + [ + 505, + 165 + ], + [ + 546, + 161 + ], + [ + 498, + 186 + ], + [ + 539, + 184 + ] + ], + [ + [ + 523, + 157 + ], + [ + 542, + 175 + ], + [ + 523, + 146 + ], + [ + 523, + 167 + ], + [ + 504, + 129 + ], + [ + 506, + 151 + ] + ], + [ + [ + 512, + 153 + ], + [ + 519, + 135 + ], + [ + 529, + 146 + ], + [ + 494, + 161 + ], + [ + 522, + 163 + ], + [ + 487, + 178 + ] + ], + [ + [ + 519, + 142 + ], + [ + 518, + 129 + ], + [ + 531, + 157 + ], + [ + 506, + 127 + ], + [ + 532, + 172 + ], + [ + 506, + 141 + ] + ] + ] + }, + { + "task_id": "0004110", + "id": "RH20T_cfg1_task_0015_user_0008_scene_0008_cfg_0001_038522063145#222", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004110/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 480, + 103 + ], + [ + 500, + 93 + ], + [ + 491, + 106 + ], + [ + 469, + 99 + ], + [ + 472, + 115 + ], + [ + 450, + 108 + ] + ], + [ + [ + 455, + 123 + ], + [ + 446, + 143 + ], + [ + 468, + 129 + ], + [ + 442, + 117 + ], + [ + 477, + 110 + ], + [ + 451, + 96 + ] + ], + [ + [ + 477, + 119 + ], + [ + 484, + 102 + ], + [ + 468, + 105 + ], + [ + 486, + 133 + ], + [ + 463, + 121 + ], + [ + 480, + 149 + ] + ], + [ + [ + 465, + 134 + ], + [ + 457, + 153 + ], + [ + 476, + 135 + ], + [ + 456, + 134 + ], + [ + 485, + 115 + ], + [ + 465, + 116 + ] + ] + ] + }, + { + "task_id": "0004111", + "id": "RH20T_cfg2_task_0101_user_0007_scene_0002_cfg_0002_036422060215#187", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the table with a cloth'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004111/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 170 + ], + [ + 285, + 161 + ], + [ + 301, + 160 + ], + [ + 268, + 180 + ], + [ + 302, + 170 + ], + [ + 267, + 190 + ] + ], + [ + [ + 278, + 153 + ], + [ + 265, + 140 + ], + [ + 265, + 160 + ], + [ + 291, + 146 + ], + [ + 279, + 174 + ], + [ + 305, + 159 + ] + ], + [ + [ + 286, + 151 + ], + [ + 302, + 145 + ], + [ + 290, + 146 + ], + [ + 282, + 155 + ], + [ + 272, + 153 + ], + [ + 266, + 161 + ] + ], + [ + [ + 291, + 155 + ], + [ + 294, + 141 + ], + [ + 307, + 151 + ], + [ + 274, + 159 + ], + [ + 304, + 164 + ], + [ + 271, + 173 + ] + ] + ] + }, + { + "task_id": "0004114", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104122063678#835", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004114/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 407, + 220 + ], + [ + 410, + 195 + ], + [ + 425, + 220 + ], + [ + 388, + 219 + ], + [ + 421, + 245 + ], + [ + 384, + 245 + ] + ], + [ + [ + 406, + 259 + ], + [ + 384, + 249 + ], + [ + 411, + 260 + ], + [ + 401, + 257 + ], + [ + 434, + 270 + ], + [ + 426, + 268 + ] + ], + [ + [ + 427, + 243 + ], + [ + 437, + 269 + ], + [ + 410, + 249 + ], + [ + 442, + 238 + ], + [ + 399, + 224 + ], + [ + 432, + 214 + ] + ], + [ + [ + 406, + 236 + ], + [ + 413, + 222 + ], + [ + 389, + 249 + ], + [ + 423, + 224 + ], + [ + 383, + 262 + ], + [ + 417, + 238 + ] + ] + ] + }, + { + "task_id": "0004115", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104122063678#1199", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004115/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 310, + 129 + ], + [ + 312, + 113 + ], + [ + 289, + 133 + ], + [ + 330, + 126 + ], + [ + 288, + 148 + ], + [ + 327, + 141 + ] + ], + [ + [ + 287, + 118 + ], + [ + 275, + 108 + ], + [ + 304, + 113 + ], + [ + 268, + 124 + ], + [ + 316, + 123 + ], + [ + 280, + 136 + ] + ], + [ + [ + 322, + 120 + ], + [ + 322, + 139 + ], + [ + 318, + 128 + ], + [ + 325, + 111 + ], + [ + 319, + 111 + ], + [ + 326, + 92 + ] + ], + [ + [ + 302, + 150 + ], + [ + 311, + 151 + ], + [ + 285, + 162 + ], + [ + 319, + 137 + ], + [ + 275, + 160 + ], + [ + 310, + 135 + ] + ] + ] + }, + { + "task_id": "0004118", + "id": "RH20T_cfg3_task_0043_user_0010_scene_0006_cfg_0003_036422060909#179", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004118/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 293, + 55 + ], + [ + 313, + 72 + ], + [ + 301, + 40 + ], + [ + 287, + 68 + ], + [ + 281, + 22 + ], + [ + 268, + 52 + ] + ], + [ + [ + 279, + 58 + ], + [ + 278, + 36 + ], + [ + 281, + 43 + ], + [ + 277, + 75 + ], + [ + 282, + 63 + ], + [ + 278, + 95 + ] + ], + [ + [ + 281, + 63 + ], + [ + 262, + 67 + ], + [ + 278, + 38 + ], + [ + 284, + 88 + ], + [ + 298, + 33 + ], + [ + 305, + 86 + ] + ], + [ + [ + 262, + 49 + ], + [ + 258, + 73 + ], + [ + 240, + 40 + ], + [ + 285, + 57 + ], + [ + 244, + 15 + ], + [ + 291, + 32 + ] + ] + ] + }, + { + "task_id": "0004120", + "id": "RH20T_cfg3_task_0043_user_0010_scene_0006_cfg_0003_036422060909#344", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004120/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 276, + 48 + ], + [ + 299, + 56 + ], + [ + 271, + 43 + ], + [ + 282, + 53 + ], + [ + 249, + 34 + ], + [ + 258, + 44 + ] + ], + [ + [ + 275, + 58 + ], + [ + 293, + 64 + ], + [ + 261, + 74 + ], + [ + 291, + 41 + ], + [ + 242, + 68 + ], + [ + 272, + 34 + ] + ], + [ + [ + 284, + 62 + ], + [ + 285, + 41 + ], + [ + 285, + 47 + ], + [ + 282, + 78 + ], + [ + 284, + 67 + ], + [ + 281, + 98 + ] + ], + [ + [ + 266, + 57 + ], + [ + 290, + 58 + ], + [ + 269, + 37 + ], + [ + 264, + 75 + ], + [ + 244, + 35 + ], + [ + 241, + 75 + ] + ] + ] + }, + { + "task_id": "0004122", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0008_cfg_0005_104422070044#264", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004122/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 561, + 149 + ], + [ + 570, + 132 + ], + [ + 563, + 141 + ], + [ + 559, + 156 + ], + [ + 554, + 157 + ], + [ + 550, + 172 + ] + ], + [ + [ + 560, + 126 + ], + [ + 579, + 118 + ], + [ + 571, + 136 + ], + [ + 550, + 116 + ], + [ + 552, + 143 + ], + [ + 532, + 124 + ] + ], + [ + [ + 543, + 148 + ], + [ + 525, + 144 + ], + [ + 552, + 133 + ], + [ + 533, + 164 + ], + [ + 571, + 137 + ], + [ + 553, + 168 + ] + ], + [ + [ + 559, + 132 + ], + [ + 558, + 118 + ], + [ + 560, + 146 + ], + [ + 558, + 119 + ], + [ + 561, + 160 + ], + [ + 559, + 133 + ] + ] + ] + }, + { + "task_id": "0004124", + "id": "RH20T_cfg4_task_0120_user_0007_scene_0004_cfg_0004_104122062295#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004124/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 247, + 143 + ], + [ + 247, + 97 + ], + [ + 202, + 136 + ], + [ + 293, + 150 + ], + [ + 204, + 178 + ], + [ + 290, + 192 + ] + ], + [ + [ + 293, + 100 + ], + [ + 290, + 143 + ], + [ + 278, + 120 + ], + [ + 305, + 82 + ], + [ + 280, + 68 + ], + [ + 309, + 37 + ] + ], + [ + [ + 288, + 121 + ], + [ + 277, + 72 + ], + [ + 315, + 117 + ], + [ + 255, + 127 + ], + [ + 325, + 161 + ], + [ + 267, + 180 + ] + ], + [ + [ + 281, + 144 + ], + [ + 312, + 130 + ], + [ + 290, + 188 + ], + [ + 272, + 101 + ], + [ + 261, + 197 + ], + [ + 245, + 115 + ] + ] + ] + }, + { + "task_id": "0004127", + "id": "RH20T_cfg5_task_0082_user_0007_scene_0008_cfg_0005_037522062165#189", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004127/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 263, + 228 + ], + [ + 263, + 211 + ], + [ + 237, + 212 + ], + [ + 289, + 246 + ], + [ + 238, + 228 + ], + [ + 287, + 261 + ] + ], + [ + [ + 225, + 245 + ], + [ + 196, + 241 + ], + [ + 231, + 262 + ], + [ + 219, + 227 + ], + [ + 259, + 267 + ], + [ + 250, + 232 + ] + ], + [ + [ + 255, + 196 + ], + [ + 244, + 187 + ], + [ + 229, + 213 + ], + [ + 281, + 179 + ], + [ + 240, + 224 + ], + [ + 296, + 187 + ] + ], + [ + [ + 275, + 254 + ], + [ + 247, + 277 + ], + [ + 257, + 235 + ], + [ + 294, + 275 + ], + [ + 283, + 213 + ], + [ + 322, + 251 + ] + ] + ] + }, + { + "task_id": "0004129", + "id": "RH20T_cfg6_task_0038_user_0020_scene_0010_cfg_0006_104122061602#349", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004129/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 496, + 169 + ], + [ + 486, + 190 + ], + [ + 485, + 179 + ], + [ + 508, + 157 + ], + [ + 494, + 160 + ], + [ + 518, + 135 + ] + ], + [ + [ + 527, + 184 + ], + [ + 546, + 175 + ], + [ + 541, + 210 + ], + [ + 514, + 160 + ], + [ + 523, + 217 + ], + [ + 498, + 170 + ] + ], + [ + [ + 533, + 196 + ], + [ + 530, + 182 + ], + [ + 532, + 221 + ], + [ + 536, + 169 + ], + [ + 535, + 237 + ], + [ + 539, + 183 + ] + ], + [ + [ + 528, + 182 + ], + [ + 503, + 168 + ], + [ + 542, + 181 + ], + [ + 515, + 182 + ], + [ + 568, + 195 + ], + [ + 538, + 195 + ] + ] + ] + }, + { + "task_id": "0004130", + "id": "RH20T_cfg7_task_0013_user_0020_scene_0004_cfg_0007_104122060811#149", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004130/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 349, + 136 + ], + [ + 335, + 133 + ], + [ + 351, + 145 + ], + [ + 346, + 127 + ], + [ + 364, + 148 + ], + [ + 360, + 130 + ] + ], + [ + [ + 339, + 138 + ], + [ + 328, + 140 + ], + [ + 335, + 125 + ], + [ + 344, + 151 + ], + [ + 345, + 124 + ], + [ + 354, + 149 + ] + ], + [ + [ + 345, + 124 + ], + [ + 337, + 135 + ], + [ + 355, + 135 + ], + [ + 336, + 114 + ], + [ + 364, + 124 + ], + [ + 344, + 104 + ] + ], + [ + [ + 337, + 145 + ], + [ + 342, + 137 + ], + [ + 350, + 143 + ], + [ + 324, + 146 + ], + [ + 346, + 150 + ], + [ + 320, + 153 + ] + ] + ] + }, + { + "task_id": "0004131", + "id": "RH20T_cfg7_task_0015_user_0020_scene_0007_cfg_0007_037522061512#223", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004131/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 205, + 198 + ], + [ + 215, + 219 + ], + [ + 211, + 208 + ], + [ + 197, + 187 + ], + [ + 201, + 188 + ], + [ + 187, + 165 + ] + ], + [ + [ + 185, + 197 + ], + [ + 176, + 176 + ], + [ + 162, + 210 + ], + [ + 207, + 185 + ], + [ + 171, + 230 + ], + [ + 214, + 204 + ] + ], + [ + [ + 202, + 183 + ], + [ + 214, + 185 + ], + [ + 190, + 205 + ], + [ + 214, + 161 + ], + [ + 176, + 205 + ], + [ + 202, + 158 + ] + ], + [ + [ + 236, + 176 + ], + [ + 240, + 175 + ], + [ + 223, + 198 + ], + [ + 249, + 155 + ], + [ + 220, + 198 + ], + [ + 244, + 158 + ] + ] + ] + }, + { + "task_id": "0004137", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_038522062288#128", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004137/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 233, + 162 + ], + [ + 266, + 176 + ], + [ + 225, + 160 + ], + [ + 243, + 165 + ], + [ + 192, + 146 + ], + [ + 205, + 150 + ] + ], + [ + [ + 191, + 170 + ], + [ + 205, + 153 + ], + [ + 221, + 161 + ], + [ + 159, + 180 + ], + [ + 208, + 175 + ], + [ + 148, + 195 + ] + ], + [ + [ + 217, + 145 + ], + [ + 215, + 120 + ], + [ + 187, + 129 + ], + [ + 247, + 161 + ], + [ + 191, + 152 + ], + [ + 247, + 183 + ] + ], + [ + [ + 239, + 138 + ], + [ + 216, + 108 + ], + [ + 238, + 131 + ], + [ + 240, + 147 + ], + [ + 259, + 158 + ], + [ + 264, + 177 + ] + ] + ] + }, + { + "task_id": "0004139", + "id": "RH20T_cfg1_task_0080_user_0015_scene_0005_cfg_0001_819112070044#126", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Flip over and spread out the paper that is laid flat on the table'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004139/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 345, + 218 + ], + [ + 366, + 224 + ], + [ + 340, + 239 + ], + [ + 349, + 197 + ], + [ + 319, + 233 + ], + [ + 329, + 191 + ] + ], + [ + [ + 317, + 246 + ], + [ + 330, + 262 + ], + [ + 333, + 238 + ], + [ + 299, + 255 + ], + [ + 320, + 223 + ], + [ + 286, + 239 + ] + ], + [ + [ + 320, + 228 + ], + [ + 318, + 213 + ], + [ + 337, + 218 + ], + [ + 301, + 239 + ], + [ + 338, + 232 + ], + [ + 303, + 253 + ] + ], + [ + [ + 308, + 220 + ], + [ + 293, + 210 + ], + [ + 307, + 237 + ], + [ + 309, + 203 + ], + [ + 321, + 247 + ], + [ + 322, + 213 + ] + ] + ] + }, + { + "task_id": "0004141", + "id": "RH20T_cfg2_task_0057_user_0016_scene_0006_cfg_0002_104122063678#301", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004141/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 446, + 265 + ], + [ + 440, + 290 + ], + [ + 433, + 258 + ], + [ + 457, + 272 + ], + [ + 440, + 234 + ], + [ + 463, + 249 + ] + ], + [ + [ + 439, + 243 + ], + [ + 444, + 231 + ], + [ + 415, + 233 + ], + [ + 464, + 253 + ], + [ + 412, + 244 + ], + [ + 459, + 264 + ] + ], + [ + [ + 436, + 232 + ], + [ + 427, + 257 + ], + [ + 457, + 243 + ], + [ + 414, + 221 + ], + [ + 465, + 220 + ], + [ + 423, + 197 + ] + ], + [ + [ + 440, + 246 + ], + [ + 443, + 241 + ], + [ + 419, + 228 + ], + [ + 460, + 265 + ], + [ + 415, + 233 + ], + [ + 458, + 273 + ] + ] + ] + }, + { + "task_id": "0004143", + "id": "RH20T_cfg2_task_0057_user_0016_scene_0006_cfg_0002_104122063678#670", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004143/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 480, + 197 + ], + [ + 497, + 216 + ], + [ + 498, + 191 + ], + [ + 463, + 203 + ], + [ + 481, + 171 + ], + [ + 447, + 186 + ] + ], + [ + [ + 464, + 194 + ], + [ + 467, + 172 + ], + [ + 482, + 204 + ], + [ + 445, + 184 + ], + [ + 480, + 227 + ], + [ + 441, + 207 + ] + ], + [ + [ + 476, + 187 + ], + [ + 490, + 166 + ], + [ + 478, + 193 + ], + [ + 474, + 183 + ], + [ + 464, + 215 + ], + [ + 462, + 203 + ] + ], + [ + [ + 455, + 216 + ], + [ + 463, + 205 + ], + [ + 431, + 214 + ], + [ + 479, + 218 + ], + [ + 424, + 223 + ], + [ + 470, + 227 + ] + ] + ] + }, + { + "task_id": "0004144", + "id": "RH20T_cfg4_task_0014_user_0010_scene_0005_cfg_0004_104122062823#112", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the object from the scale'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004144/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 233, + 34 + ], + [ + 266, + 39 + ], + [ + 233, + 47 + ], + [ + 233, + 19 + ], + [ + 202, + 42 + ], + [ + 196, + 12 + ] + ], + [ + [ + 266, + 69 + ], + [ + 281, + 60 + ], + [ + 241, + 74 + ], + [ + 293, + 63 + ], + [ + 223, + 84 + ], + [ + 277, + 73 + ] + ], + [ + [ + 255, + 68 + ], + [ + 271, + 40 + ], + [ + 259, + 80 + ], + [ + 251, + 56 + ], + [ + 242, + 108 + ], + [ + 237, + 82 + ] + ], + [ + [ + 254, + 73 + ], + [ + 254, + 41 + ], + [ + 281, + 83 + ], + [ + 229, + 64 + ], + [ + 280, + 112 + ], + [ + 230, + 92 + ] + ] + ] + }, + { + "task_id": "0004146", + "id": "RH20T_cfg1_task_0103_user_0010_scene_0005_cfg_0001_819112070044#91", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004146/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 426, + 212 + ], + [ + 427, + 198 + ], + [ + 443, + 203 + ], + [ + 409, + 221 + ], + [ + 442, + 216 + ], + [ + 409, + 234 + ] + ], + [ + [ + 418, + 201 + ], + [ + 429, + 184 + ], + [ + 403, + 193 + ], + [ + 433, + 211 + ], + [ + 391, + 209 + ], + [ + 421, + 228 + ] + ], + [ + [ + 417, + 211 + ], + [ + 403, + 197 + ], + [ + 400, + 221 + ], + [ + 433, + 201 + ], + [ + 415, + 237 + ], + [ + 448, + 216 + ] + ], + [ + [ + 408, + 218 + ], + [ + 393, + 209 + ], + [ + 421, + 213 + ], + [ + 395, + 224 + ], + [ + 437, + 222 + ], + [ + 411, + 234 + ] + ] + ] + }, + { + "task_id": "0004150", + "id": "RH20T_cfg1_task_0085_user_0015_scene_0002_cfg_0001_819112070044#117", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the bubble ring from the assembled bubble ring and ball'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004150/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 344, + 204 + ], + [ + 338, + 190 + ], + [ + 346, + 218 + ], + [ + 341, + 189 + ], + [ + 353, + 232 + ], + [ + 348, + 204 + ] + ], + [ + [ + 340, + 201 + ], + [ + 342, + 185 + ], + [ + 355, + 193 + ], + [ + 324, + 208 + ], + [ + 352, + 208 + ], + [ + 322, + 223 + ] + ], + [ + [ + 328, + 200 + ], + [ + 338, + 182 + ], + [ + 310, + 190 + ], + [ + 345, + 210 + ], + [ + 300, + 207 + ], + [ + 334, + 228 + ] + ], + [ + [ + 332, + 183 + ], + [ + 345, + 192 + ], + [ + 344, + 166 + ], + [ + 320, + 200 + ], + [ + 332, + 158 + ], + [ + 308, + 191 + ] + ] + ] + }, + { + "task_id": "0004153", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_037522062165#373", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004153/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 236, + 247 + ], + [ + 249, + 232 + ], + [ + 221, + 254 + ], + [ + 252, + 240 + ], + [ + 207, + 270 + ], + [ + 239, + 257 + ] + ], + [ + [ + 243, + 263 + ], + [ + 269, + 258 + ], + [ + 247, + 290 + ], + [ + 238, + 235 + ], + [ + 219, + 296 + ], + [ + 210, + 240 + ] + ], + [ + [ + 226, + 282 + ], + [ + 217, + 306 + ], + [ + 244, + 298 + ], + [ + 207, + 266 + ], + [ + 252, + 276 + ], + [ + 217, + 244 + ] + ], + [ + [ + 238, + 262 + ], + [ + 238, + 249 + ], + [ + 224, + 242 + ], + [ + 252, + 283 + ], + [ + 224, + 254 + ], + [ + 251, + 294 + ] + ] + ] + }, + { + "task_id": "0004155", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0007_cfg_0002_037522062165#699", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004155/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 229, + 295 + ], + [ + 229, + 282 + ], + [ + 215, + 273 + ], + [ + 243, + 318 + ], + [ + 216, + 285 + ], + [ + 243, + 329 + ] + ], + [ + [ + 225, + 281 + ], + [ + 207, + 298 + ], + [ + 227, + 304 + ], + [ + 223, + 256 + ], + [ + 243, + 288 + ], + [ + 240, + 242 + ] + ], + [ + [ + 226, + 260 + ], + [ + 250, + 268 + ], + [ + 237, + 266 + ], + [ + 216, + 254 + ], + [ + 210, + 257 + ], + [ + 191, + 246 + ] + ], + [ + [ + 217, + 294 + ], + [ + 189, + 303 + ], + [ + 219, + 318 + ], + [ + 215, + 270 + ], + [ + 246, + 309 + ], + [ + 243, + 262 + ] + ] + ] + }, + { + "task_id": "0004157", + "id": "RH20T_cfg5_task_0011_user_0007_scene_0004_cfg_0005_104122061850#236", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004157/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 398, + 58 + ], + [ + 394, + 37 + ], + [ + 371, + 64 + ], + [ + 427, + 52 + ], + [ + 376, + 82 + ], + [ + 429, + 72 + ] + ], + [ + [ + 412, + 58 + ], + [ + 428, + 37 + ], + [ + 393, + 56 + ], + [ + 433, + 60 + ], + [ + 377, + 77 + ], + [ + 416, + 83 + ] + ], + [ + [ + 378, + 75 + ], + [ + 358, + 57 + ], + [ + 395, + 52 + ], + [ + 362, + 97 + ], + [ + 416, + 71 + ], + [ + 381, + 116 + ] + ], + [ + [ + 378, + 61 + ], + [ + 374, + 32 + ], + [ + 353, + 65 + ], + [ + 405, + 57 + ], + [ + 357, + 91 + ], + [ + 408, + 86 + ] + ] + ] + }, + { + "task_id": "0004158", + "id": "RH20T_cfg4_task_0047_user_0014_scene_0012_cfg_0004_104122062295#199", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004158/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 132 + ], + [ + 286, + 139 + ], + [ + 336, + 168 + ], + [ + 290, + 93 + ], + [ + 369, + 163 + ], + [ + 321, + 81 + ] + ], + [ + [ + 351, + 151 + ], + [ + 322, + 156 + ], + [ + 359, + 199 + ], + [ + 344, + 102 + ], + [ + 394, + 198 + ], + [ + 379, + 91 + ] + ], + [ + [ + 315, + 118 + ], + [ + 306, + 69 + ], + [ + 284, + 131 + ], + [ + 341, + 106 + ], + [ + 294, + 179 + ], + [ + 348, + 149 + ] + ], + [ + [ + 317, + 76 + ], + [ + 367, + 65 + ], + [ + 301, + 45 + ], + [ + 335, + 111 + ], + [ + 252, + 55 + ], + [ + 278, + 124 + ] + ] + ] + }, + { + "task_id": "0004159", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_036422060215#262", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004159/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 289, + 179 + ], + [ + 275, + 168 + ], + [ + 273, + 194 + ], + [ + 304, + 165 + ], + [ + 288, + 207 + ], + [ + 319, + 176 + ] + ], + [ + [ + 293, + 163 + ], + [ + 298, + 147 + ], + [ + 310, + 158 + ], + [ + 275, + 167 + ], + [ + 305, + 173 + ], + [ + 270, + 182 + ] + ], + [ + [ + 292, + 162 + ], + [ + 281, + 150 + ], + [ + 302, + 168 + ], + [ + 280, + 156 + ], + [ + 313, + 181 + ], + [ + 291, + 169 + ] + ], + [ + [ + 296, + 166 + ], + [ + 286, + 175 + ], + [ + 301, + 183 + ], + [ + 292, + 149 + ], + [ + 312, + 173 + ], + [ + 302, + 138 + ] + ] + ] + }, + { + "task_id": "0004160", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_036422060215#487", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004160/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 209, + 132 + ], + [ + 194, + 116 + ], + [ + 223, + 118 + ], + [ + 196, + 144 + ], + [ + 239, + 134 + ], + [ + 211, + 159 + ] + ], + [ + [ + 213, + 154 + ], + [ + 216, + 170 + ], + [ + 228, + 144 + ], + [ + 196, + 165 + ], + [ + 225, + 127 + ], + [ + 191, + 148 + ] + ], + [ + [ + 214, + 151 + ], + [ + 218, + 131 + ], + [ + 233, + 150 + ], + [ + 196, + 152 + ], + [ + 229, + 170 + ], + [ + 193, + 172 + ] + ], + [ + [ + 206, + 132 + ], + [ + 209, + 116 + ], + [ + 228, + 135 + ], + [ + 184, + 130 + ], + [ + 224, + 150 + ], + [ + 181, + 145 + ] + ] + ] + }, + { + "task_id": "0004166", + "id": "RH20T_cfg1_task_0046_user_0001_scene_0003_cfg_0001_819112070044#192", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004166/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 392, + 151 + ], + [ + 399, + 155 + ], + [ + 408, + 144 + ], + [ + 375, + 157 + ], + [ + 401, + 139 + ], + [ + 367, + 153 + ] + ], + [ + [ + 406, + 140 + ], + [ + 407, + 127 + ], + [ + 394, + 130 + ], + [ + 419, + 151 + ], + [ + 393, + 143 + ], + [ + 417, + 163 + ] + ], + [ + [ + 389, + 135 + ], + [ + 403, + 140 + ], + [ + 393, + 147 + ], + [ + 384, + 121 + ], + [ + 380, + 142 + ], + [ + 370, + 116 + ] + ], + [ + [ + 404, + 164 + ], + [ + 421, + 156 + ], + [ + 399, + 149 + ], + [ + 408, + 179 + ], + [ + 384, + 158 + ], + [ + 392, + 187 + ] + ] + ] + }, + { + "task_id": "0004171", + "id": "RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_104122060811#224", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004171/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 377, + 127 + ], + [ + 373, + 140 + ], + [ + 367, + 121 + ], + [ + 386, + 133 + ], + [ + 371, + 108 + ], + [ + 390, + 121 + ] + ], + [ + [ + 368, + 106 + ], + [ + 363, + 95 + ], + [ + 364, + 114 + ], + [ + 373, + 98 + ], + [ + 369, + 125 + ], + [ + 379, + 110 + ] + ], + [ + [ + 374, + 113 + ], + [ + 378, + 105 + ], + [ + 387, + 110 + ], + [ + 361, + 117 + ], + [ + 384, + 117 + ], + [ + 359, + 125 + ] + ], + [ + [ + 371, + 126 + ], + [ + 363, + 114 + ], + [ + 367, + 124 + ], + [ + 374, + 129 + ], + [ + 374, + 135 + ], + [ + 382, + 140 + ] + ] + ] + }, + { + "task_id": "0004172", + "id": "RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_104122060811#469", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004172/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 332, + 120 + ], + [ + 325, + 107 + ], + [ + 318, + 126 + ], + [ + 345, + 115 + ], + [ + 325, + 139 + ], + [ + 352, + 127 + ] + ], + [ + [ + 338, + 106 + ], + [ + 334, + 116 + ], + [ + 328, + 97 + ], + [ + 347, + 115 + ], + [ + 331, + 87 + ], + [ + 350, + 106 + ] + ], + [ + [ + 337, + 112 + ], + [ + 338, + 103 + ], + [ + 351, + 108 + ], + [ + 323, + 117 + ], + [ + 350, + 117 + ], + [ + 323, + 125 + ] + ], + [ + [ + 340, + 119 + ], + [ + 344, + 106 + ], + [ + 336, + 112 + ], + [ + 344, + 127 + ], + [ + 332, + 125 + ], + [ + 339, + 140 + ] + ] + ] + }, + { + "task_id": "0004173", + "id": "RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_104122060811#639", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004173/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 130 + ], + [ + 334, + 133 + ], + [ + 338, + 116 + ], + [ + 330, + 143 + ], + [ + 338, + 113 + ], + [ + 330, + 140 + ] + ], + [ + [ + 336, + 127 + ], + [ + 335, + 117 + ], + [ + 321, + 129 + ], + [ + 351, + 125 + ], + [ + 323, + 139 + ], + [ + 353, + 135 + ] + ], + [ + [ + 338, + 113 + ], + [ + 333, + 106 + ], + [ + 351, + 106 + ], + [ + 325, + 120 + ], + [ + 356, + 113 + ], + [ + 330, + 126 + ] + ], + [ + [ + 343, + 125 + ], + [ + 353, + 131 + ], + [ + 340, + 115 + ], + [ + 346, + 135 + ], + [ + 330, + 109 + ], + [ + 336, + 129 + ] + ] + ] + }, + { + "task_id": "0004174", + "id": "RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_104122060811#779", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004174/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 335, + 109 + ], + [ + 330, + 107 + ], + [ + 349, + 109 + ], + [ + 321, + 109 + ], + [ + 354, + 110 + ], + [ + 327, + 110 + ] + ], + [ + [ + 334, + 99 + ], + [ + 328, + 99 + ], + [ + 331, + 114 + ], + [ + 337, + 85 + ], + [ + 337, + 113 + ], + [ + 343, + 85 + ] + ], + [ + [ + 336, + 109 + ], + [ + 333, + 102 + ], + [ + 350, + 103 + ], + [ + 323, + 116 + ], + [ + 353, + 110 + ], + [ + 327, + 122 + ] + ], + [ + [ + 337, + 109 + ], + [ + 342, + 122 + ], + [ + 323, + 111 + ], + [ + 351, + 107 + ], + [ + 318, + 99 + ], + [ + 345, + 96 + ] + ] + ] + }, + { + "task_id": "0004175", + "id": "RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_104122060811#931", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004175/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 104 + ], + [ + 355, + 97 + ], + [ + 335, + 100 + ], + [ + 362, + 109 + ], + [ + 328, + 108 + ], + [ + 355, + 117 + ] + ], + [ + [ + 346, + 114 + ], + [ + 345, + 115 + ], + [ + 342, + 127 + ], + [ + 351, + 100 + ], + [ + 343, + 126 + ], + [ + 352, + 99 + ] + ], + [ + [ + 336, + 111 + ], + [ + 332, + 104 + ], + [ + 350, + 104 + ], + [ + 323, + 118 + ], + [ + 353, + 111 + ], + [ + 328, + 124 + ] + ], + [ + [ + 329, + 120 + ], + [ + 315, + 121 + ], + [ + 326, + 130 + ], + [ + 332, + 111 + ], + [ + 341, + 128 + ], + [ + 346, + 110 + ] + ] + ] + }, + { + "task_id": "0004177", + "id": "RH20T_cfg1_task_0120_user_0014_scene_0002_cfg_0001_038522063145#105", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004177/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 412, + 110 + ], + [ + 407, + 91 + ], + [ + 435, + 104 + ], + [ + 389, + 116 + ], + [ + 439, + 123 + ], + [ + 394, + 135 + ] + ], + [ + [ + 413, + 127 + ], + [ + 393, + 131 + ], + [ + 406, + 139 + ], + [ + 421, + 113 + ], + [ + 425, + 134 + ], + [ + 441, + 109 + ] + ], + [ + [ + 398, + 122 + ], + [ + 389, + 111 + ], + [ + 377, + 121 + ], + [ + 418, + 122 + ], + [ + 385, + 134 + ], + [ + 427, + 134 + ] + ], + [ + [ + 426, + 92 + ], + [ + 436, + 82 + ], + [ + 427, + 72 + ], + [ + 425, + 113 + ], + [ + 418, + 82 + ], + [ + 416, + 122 + ] + ] + ] + }, + { + "task_id": "0004180", + "id": "RH20T_cfg1_task_0120_user_0014_scene_0002_cfg_0001_038522063145#442", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004180/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 414, + 96 + ], + [ + 398, + 80 + ], + [ + 429, + 76 + ], + [ + 398, + 114 + ], + [ + 447, + 92 + ], + [ + 414, + 130 + ] + ], + [ + [ + 410, + 113 + ], + [ + 389, + 115 + ], + [ + 405, + 102 + ], + [ + 416, + 124 + ], + [ + 428, + 99 + ], + [ + 437, + 122 + ] + ], + [ + [ + 439, + 117 + ], + [ + 425, + 136 + ], + [ + 429, + 117 + ], + [ + 451, + 116 + ], + [ + 442, + 99 + ], + [ + 466, + 95 + ] + ], + [ + [ + 433, + 109 + ], + [ + 440, + 90 + ], + [ + 444, + 100 + ], + [ + 421, + 119 + ], + [ + 437, + 117 + ], + [ + 414, + 137 + ] + ] + ] + }, + { + "task_id": "0004183", + "id": "RH20T_cfg1_task_0100_user_0015_scene_0008_cfg_0001_038522062547#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stir the beaker with a glass rod'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004183/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 467, + 37 + ], + [ + 499, + 24 + ], + [ + 474, + 52 + ], + [ + 461, + 23 + ], + [ + 441, + 64 + ], + [ + 432, + 35 + ] + ], + [ + [ + 471, + 52 + ], + [ + 482, + 29 + ], + [ + 481, + 78 + ], + [ + 461, + 27 + ], + [ + 471, + 96 + ], + [ + 452, + 48 + ] + ], + [ + [ + 446, + 26 + ], + [ + 427, + 54 + ], + [ + 471, + 37 + ], + [ + 424, + 17 + ], + [ + 493, + 6 + ], + [ + 443, + 0 + ] + ], + [ + [ + 498, + 61 + ], + [ + 519, + 31 + ], + [ + 528, + 71 + ], + [ + 472, + 51 + ], + [ + 506, + 101 + ], + [ + 453, + 78 + ] + ] + ] + }, + { + "task_id": "0004184", + "id": "RH20T_cfg1_task_0002_user_0014_scene_0004_cfg_0001_038522062547#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004184/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 431, + 152 + ], + [ + 430, + 130 + ], + [ + 415, + 165 + ], + [ + 449, + 138 + ], + [ + 416, + 189 + ], + [ + 452, + 163 + ] + ], + [ + [ + 428, + 175 + ], + [ + 431, + 163 + ], + [ + 435, + 201 + ], + [ + 421, + 150 + ], + [ + 432, + 210 + ], + [ + 419, + 162 + ] + ], + [ + [ + 408, + 184 + ], + [ + 403, + 184 + ], + [ + 411, + 212 + ], + [ + 404, + 155 + ], + [ + 416, + 214 + ], + [ + 409, + 153 + ] + ], + [ + [ + 423, + 171 + ], + [ + 436, + 149 + ], + [ + 409, + 153 + ], + [ + 436, + 187 + ], + [ + 395, + 176 + ], + [ + 423, + 208 + ] + ] + ] + }, + { + "task_id": "0004185", + "id": "RH20T_cfg6_task_0039_user_0014_scene_0002_cfg_0006_037522061512#117", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004185/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 294, + 124 + ], + [ + 312, + 135 + ], + [ + 307, + 104 + ], + [ + 281, + 144 + ], + [ + 291, + 94 + ], + [ + 265, + 133 + ] + ], + [ + [ + 271, + 114 + ], + [ + 268, + 96 + ], + [ + 282, + 97 + ], + [ + 261, + 131 + ], + [ + 284, + 114 + ], + [ + 264, + 148 + ] + ], + [ + [ + 290, + 107 + ], + [ + 280, + 83 + ], + [ + 276, + 109 + ], + [ + 307, + 105 + ], + [ + 286, + 130 + ], + [ + 317, + 129 + ] + ], + [ + [ + 295, + 119 + ], + [ + 285, + 108 + ], + [ + 317, + 115 + ], + [ + 272, + 123 + ], + [ + 328, + 126 + ], + [ + 282, + 134 + ] + ] + ] + }, + { + "task_id": "0004188", + "id": "RH20T_cfg7_task_0042_user_0020_scene_0003_cfg_0007_104122064161#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004188/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 370, + 263 + ], + [ + 368, + 249 + ], + [ + 387, + 249 + ], + [ + 352, + 277 + ], + [ + 387, + 262 + ], + [ + 354, + 290 + ] + ], + [ + [ + 366, + 215 + ], + [ + 344, + 219 + ], + [ + 360, + 219 + ], + [ + 372, + 212 + ], + [ + 384, + 214 + ], + [ + 394, + 208 + ] + ], + [ + [ + 363, + 223 + ], + [ + 381, + 236 + ], + [ + 360, + 215 + ], + [ + 365, + 231 + ], + [ + 342, + 202 + ], + [ + 348, + 218 + ] + ], + [ + [ + 386, + 231 + ], + [ + 389, + 219 + ], + [ + 387, + 254 + ], + [ + 386, + 210 + ], + [ + 384, + 264 + ], + [ + 383, + 221 + ] + ] + ] + }, + { + "task_id": "0004190", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0003_cfg_0004_104422070011#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004190/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 337, + 159 + ], + [ + 347, + 132 + ], + [ + 329, + 138 + ], + [ + 346, + 182 + ], + [ + 320, + 162 + ], + [ + 336, + 207 + ] + ], + [ + [ + 358, + 160 + ], + [ + 362, + 151 + ], + [ + 328, + 147 + ], + [ + 387, + 173 + ], + [ + 327, + 157 + ], + [ + 382, + 181 + ] + ], + [ + [ + 359, + 129 + ], + [ + 331, + 129 + ], + [ + 368, + 104 + ], + [ + 352, + 151 + ], + [ + 397, + 104 + ], + [ + 378, + 150 + ] + ], + [ + [ + 349, + 146 + ], + [ + 359, + 130 + ], + [ + 324, + 158 + ], + [ + 373, + 134 + ], + [ + 317, + 172 + ], + [ + 364, + 148 + ] + ] + ] + }, + { + "task_id": "0004191", + "id": "RH20T_cfg4_task_0096_user_0015_scene_0003_cfg_0004_104422070011#370", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004191/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 374, + 159 + ], + [ + 343, + 167 + ], + [ + 376, + 143 + ], + [ + 371, + 172 + ], + [ + 409, + 134 + ], + [ + 400, + 164 + ] + ], + [ + [ + 337, + 146 + ], + [ + 341, + 159 + ], + [ + 359, + 162 + ], + [ + 315, + 130 + ], + [ + 355, + 147 + ], + [ + 308, + 114 + ] + ], + [ + [ + 353, + 162 + ], + [ + 364, + 146 + ], + [ + 329, + 179 + ], + [ + 375, + 147 + ], + [ + 321, + 193 + ], + [ + 365, + 161 + ] + ], + [ + [ + 350, + 138 + ], + [ + 325, + 121 + ], + [ + 336, + 166 + ], + [ + 365, + 108 + ], + [ + 362, + 185 + ], + [ + 393, + 125 + ] + ] + ] + }, + { + "task_id": "0004192", + "id": "RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_104122063633#218", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004192/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 148, + 159 + ], + [ + 172, + 160 + ], + [ + 155, + 147 + ], + [ + 141, + 173 + ], + [ + 133, + 146 + ], + [ + 116, + 173 + ] + ], + [ + [ + 182, + 155 + ], + [ + 206, + 159 + ], + [ + 181, + 157 + ], + [ + 184, + 152 + ], + [ + 157, + 153 + ], + [ + 158, + 147 + ] + ], + [ + [ + 185, + 175 + ], + [ + 163, + 182 + ], + [ + 189, + 195 + ], + [ + 181, + 154 + ], + [ + 209, + 188 + ], + [ + 202, + 148 + ] + ], + [ + [ + 166, + 154 + ], + [ + 165, + 135 + ], + [ + 170, + 139 + ], + [ + 162, + 169 + ], + [ + 171, + 156 + ], + [ + 163, + 187 + ] + ] + ] + }, + { + "task_id": "0004193", + "id": "RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_104122063633#456", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004193/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 149, + 164 + ], + [ + 174, + 165 + ], + [ + 151, + 189 + ], + [ + 146, + 138 + ], + [ + 127, + 188 + ], + [ + 120, + 137 + ] + ], + [ + [ + 137, + 188 + ], + [ + 131, + 170 + ], + [ + 143, + 169 + ], + [ + 131, + 209 + ], + [ + 148, + 186 + ], + [ + 137, + 225 + ] + ], + [ + [ + 119, + 204 + ], + [ + 97, + 200 + ], + [ + 126, + 227 + ], + [ + 111, + 180 + ], + [ + 149, + 231 + ], + [ + 134, + 184 + ] + ], + [ + [ + 167, + 179 + ], + [ + 192, + 186 + ], + [ + 172, + 162 + ], + [ + 161, + 197 + ], + [ + 147, + 154 + ], + [ + 133, + 190 + ] + ] + ] + }, + { + "task_id": "0004194", + "id": "RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_104122063633#636", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004194/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 147, + 189 + ], + [ + 172, + 179 + ], + [ + 144, + 182 + ], + [ + 149, + 196 + ], + [ + 116, + 193 + ], + [ + 124, + 206 + ] + ], + [ + [ + 139, + 188 + ], + [ + 131, + 179 + ], + [ + 143, + 163 + ], + [ + 134, + 213 + ], + [ + 150, + 173 + ], + [ + 142, + 220 + ] + ], + [ + [ + 147, + 176 + ], + [ + 152, + 152 + ], + [ + 175, + 177 + ], + [ + 119, + 175 + ], + [ + 170, + 203 + ], + [ + 113, + 201 + ] + ], + [ + [ + 162, + 185 + ], + [ + 154, + 182 + ], + [ + 135, + 189 + ], + [ + 189, + 182 + ], + [ + 143, + 193 + ], + [ + 195, + 185 + ] + ] + ] + }, + { + "task_id": "0004195", + "id": "RH20T_cfg6_task_0120_user_0020_scene_0005_cfg_0006_104122063633#784", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004195/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 162, + 164 + ], + [ + 189, + 162 + ], + [ + 159, + 165 + ], + [ + 164, + 162 + ], + [ + 129, + 168 + ], + [ + 138, + 164 + ] + ], + [ + [ + 138, + 204 + ], + [ + 165, + 208 + ], + [ + 146, + 200 + ], + [ + 128, + 209 + ], + [ + 120, + 195 + ], + [ + 98, + 204 + ] + ], + [ + [ + 161, + 173 + ], + [ + 138, + 184 + ], + [ + 152, + 150 + ], + [ + 170, + 196 + ], + [ + 175, + 140 + ], + [ + 191, + 185 + ] + ], + [ + [ + 133, + 185 + ], + [ + 127, + 171 + ], + [ + 139, + 162 + ], + [ + 127, + 209 + ], + [ + 143, + 176 + ], + [ + 132, + 221 + ] + ] + ] + }, + { + "task_id": "0004197", + "id": "RH20T_cfg4_task_0083_user_0015_scene_0004_cfg_0004_104122062823#197", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Swat with a flyswatter'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004197/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 346, + 65 + ], + [ + 357, + 39 + ], + [ + 346, + 58 + ], + [ + 346, + 72 + ], + [ + 334, + 85 + ], + [ + 336, + 96 + ] + ], + [ + [ + 362, + 72 + ], + [ + 384, + 59 + ], + [ + 347, + 49 + ], + [ + 377, + 95 + ], + [ + 324, + 62 + ], + [ + 355, + 109 + ] + ], + [ + [ + 347, + 70 + ], + [ + 345, + 43 + ], + [ + 375, + 75 + ], + [ + 322, + 65 + ], + [ + 376, + 99 + ], + [ + 325, + 88 + ] + ], + [ + [ + 341, + 53 + ], + [ + 325, + 35 + ], + [ + 341, + 28 + ], + [ + 341, + 80 + ], + [ + 355, + 46 + ], + [ + 356, + 95 + ] + ] + ] + }, + { + "task_id": "0004199", + "id": "RH20T_cfg1_task_0042_user_0016_scene_0007_cfg_0001_039422060546#107", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004199/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 380, + 22 + ], + [ + 355, + 36 + ], + [ + 393, + 45 + ], + [ + 366, + 0 + ], + [ + 418, + 32 + ], + [ + 393, + 0 + ] + ], + [ + [ + 365, + 24 + ], + [ + 360, + 0 + ], + [ + 361, + 43 + ], + [ + 369, + 7 + ], + [ + 366, + 68 + ], + [ + 373, + 32 + ] + ], + [ + [ + 358, + 43 + ], + [ + 329, + 36 + ], + [ + 369, + 33 + ], + [ + 349, + 51 + ], + [ + 400, + 40 + ], + [ + 376, + 57 + ] + ], + [ + [ + 354, + 0 + ], + [ + 348, + 6 + ], + [ + 382, + 0 + ], + [ + 327, + 6 + ], + [ + 386, + 0 + ], + [ + 333, + 0 + ] + ] + ] + }, + { + "task_id": "0004205", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0003_cfg_0004_104122062295#381", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004205/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 183, + 292 + ], + [ + 189, + 253 + ], + [ + 161, + 260 + ], + [ + 209, + 329 + ], + [ + 158, + 294 + ], + [ + 203, + 360 + ] + ], + [ + [ + 151, + 358 + ], + [ + 197, + 342 + ], + [ + 138, + 360 + ], + [ + 165, + 313 + ], + [ + 84, + 360 + ], + [ + 109, + 327 + ] + ], + [ + [ + 212, + 298 + ], + [ + 260, + 265 + ], + [ + 222, + 298 + ], + [ + 199, + 298 + ], + [ + 177, + 329 + ], + [ + 141, + 338 + ] + ], + [ + [ + 179, + 348 + ], + [ + 138, + 360 + ], + [ + 153, + 311 + ], + [ + 209, + 360 + ], + [ + 190, + 276 + ], + [ + 249, + 347 + ] + ] + ] + }, + { + "task_id": "0004206", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0003_cfg_0004_104122062295#539", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004206/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 202, + 322 + ], + [ + 257, + 307 + ], + [ + 231, + 328 + ], + [ + 178, + 317 + ], + [ + 159, + 348 + ], + [ + 117, + 332 + ] + ], + [ + [ + 178, + 353 + ], + [ + 222, + 360 + ], + [ + 158, + 360 + ], + [ + 198, + 301 + ], + [ + 120, + 360 + ], + [ + 157, + 272 + ] + ], + [ + [ + 207, + 360 + ], + [ + 256, + 360 + ], + [ + 214, + 360 + ], + [ + 202, + 318 + ], + [ + 153, + 360 + ], + [ + 150, + 315 + ] + ], + [ + [ + 225, + 319 + ], + [ + 236, + 278 + ], + [ + 204, + 283 + ], + [ + 249, + 360 + ], + [ + 195, + 319 + ], + [ + 237, + 360 + ] + ] + ] + }, + { + "task_id": "0004207", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0003_cfg_0005_037522062165#199", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004207/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 233 + ], + [ + 278, + 232 + ], + [ + 305, + 260 + ], + [ + 294, + 207 + ], + [ + 325, + 259 + ], + [ + 314, + 209 + ] + ], + [ + [ + 327, + 229 + ], + [ + 330, + 211 + ], + [ + 308, + 211 + ], + [ + 348, + 248 + ], + [ + 306, + 228 + ], + [ + 344, + 263 + ] + ], + [ + [ + 342, + 239 + ], + [ + 353, + 236 + ], + [ + 318, + 250 + ], + [ + 367, + 226 + ], + [ + 305, + 254 + ], + [ + 356, + 229 + ] + ], + [ + [ + 330, + 220 + ], + [ + 352, + 204 + ], + [ + 346, + 240 + ], + [ + 314, + 202 + ], + [ + 323, + 257 + ], + [ + 292, + 218 + ] + ] + ] + }, + { + "task_id": "0004212", + "id": "RH20T_cfg5_task_0039_user_0007_scene_0007_cfg_0005_104422070044#91", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004212/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 129 + ], + [ + 336, + 142 + ], + [ + 347, + 132 + ], + [ + 348, + 125 + ], + [ + 358, + 119 + ], + [ + 359, + 112 + ] + ], + [ + [ + 352, + 151 + ], + [ + 365, + 144 + ], + [ + 341, + 150 + ], + [ + 363, + 153 + ], + [ + 327, + 156 + ], + [ + 350, + 161 + ] + ], + [ + [ + 349, + 148 + ], + [ + 347, + 135 + ], + [ + 360, + 157 + ], + [ + 338, + 140 + ], + [ + 362, + 169 + ], + [ + 341, + 152 + ] + ], + [ + [ + 341, + 148 + ], + [ + 326, + 154 + ], + [ + 349, + 147 + ], + [ + 333, + 150 + ], + [ + 364, + 142 + ], + [ + 349, + 144 + ] + ] + ] + }, + { + "task_id": "0004213", + "id": "RH20T_cfg1_task_0103_user_0010_scene_0005_cfg_0001_035622060973#87", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004213/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 242, + 146 + ], + [ + 251, + 161 + ], + [ + 246, + 135 + ], + [ + 237, + 159 + ], + [ + 236, + 119 + ], + [ + 226, + 143 + ] + ], + [ + [ + 245, + 137 + ], + [ + 243, + 129 + ], + [ + 236, + 121 + ], + [ + 254, + 153 + ], + [ + 237, + 130 + ], + [ + 254, + 161 + ] + ], + [ + [ + 260, + 136 + ], + [ + 246, + 134 + ], + [ + 268, + 152 + ], + [ + 253, + 120 + ], + [ + 282, + 153 + ], + [ + 267, + 122 + ] + ], + [ + [ + 242, + 126 + ], + [ + 226, + 137 + ], + [ + 233, + 122 + ], + [ + 252, + 130 + ], + [ + 248, + 111 + ], + [ + 268, + 119 + ] + ] + ] + }, + { + "task_id": "0004214", + "id": "RH20T_cfg6_task_0104_user_0020_scene_0006_cfg_0006_104122061602#221", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate back after holding it down'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004214/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 391, + 123 + ], + [ + 406, + 130 + ], + [ + 383, + 143 + ], + [ + 398, + 103 + ], + [ + 370, + 136 + ], + [ + 384, + 97 + ] + ], + [ + [ + 373, + 113 + ], + [ + 367, + 93 + ], + [ + 352, + 118 + ], + [ + 394, + 108 + ], + [ + 358, + 137 + ], + [ + 399, + 128 + ] + ], + [ + [ + 382, + 99 + ], + [ + 362, + 105 + ], + [ + 386, + 112 + ], + [ + 379, + 86 + ], + [ + 408, + 106 + ], + [ + 399, + 80 + ] + ], + [ + [ + 393, + 122 + ], + [ + 414, + 114 + ], + [ + 392, + 110 + ], + [ + 394, + 134 + ], + [ + 373, + 117 + ], + [ + 373, + 141 + ] + ] + ] + }, + { + "task_id": "0004216", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0008_cfg_0006_104122060811#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004216/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 404, + 147 + ], + [ + 393, + 157 + ], + [ + 404, + 141 + ], + [ + 405, + 153 + ], + [ + 415, + 131 + ], + [ + 415, + 143 + ] + ], + [ + [ + 395, + 152 + ], + [ + 385, + 141 + ], + [ + 383, + 161 + ], + [ + 407, + 143 + ], + [ + 393, + 172 + ], + [ + 417, + 154 + ] + ], + [ + [ + 393, + 143 + ], + [ + 407, + 135 + ], + [ + 398, + 148 + ], + [ + 388, + 138 + ], + [ + 385, + 155 + ], + [ + 374, + 146 + ] + ], + [ + [ + 392, + 143 + ], + [ + 380, + 135 + ], + [ + 395, + 146 + ], + [ + 389, + 139 + ], + [ + 407, + 153 + ], + [ + 402, + 147 + ] + ] + ] + }, + { + "task_id": "0004217", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0008_cfg_0006_104122060811#228", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004217/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 411, + 142 + ], + [ + 409, + 149 + ], + [ + 399, + 150 + ], + [ + 423, + 134 + ], + [ + 402, + 143 + ], + [ + 427, + 127 + ] + ], + [ + [ + 402, + 126 + ], + [ + 403, + 115 + ], + [ + 407, + 136 + ], + [ + 397, + 115 + ], + [ + 407, + 147 + ], + [ + 397, + 126 + ] + ], + [ + [ + 398, + 144 + ], + [ + 413, + 140 + ], + [ + 396, + 151 + ], + [ + 400, + 137 + ], + [ + 382, + 156 + ], + [ + 385, + 142 + ] + ], + [ + [ + 381, + 142 + ], + [ + 381, + 127 + ], + [ + 371, + 143 + ], + [ + 391, + 141 + ], + [ + 372, + 158 + ], + [ + 391, + 156 + ] + ] + ] + }, + { + "task_id": "0004218", + "id": "RH20T_cfg2_task_0071_user_0014_scene_0003_cfg_0002_036422060215#177", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004218/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 298, + 147 + ], + [ + 302, + 140 + ], + [ + 303, + 165 + ], + [ + 294, + 129 + ], + [ + 299, + 172 + ], + [ + 290, + 137 + ] + ], + [ + [ + 286, + 185 + ], + [ + 292, + 193 + ], + [ + 267, + 184 + ], + [ + 303, + 187 + ], + [ + 262, + 177 + ], + [ + 297, + 180 + ] + ], + [ + [ + 285, + 162 + ], + [ + 281, + 149 + ], + [ + 299, + 170 + ], + [ + 270, + 154 + ], + [ + 303, + 184 + ], + [ + 274, + 169 + ] + ], + [ + [ + 298, + 165 + ], + [ + 303, + 151 + ], + [ + 316, + 166 + ], + [ + 280, + 164 + ], + [ + 311, + 180 + ], + [ + 275, + 178 + ] + ] + ] + }, + { + "task_id": "0004219", + "id": "RH20T_cfg2_task_0071_user_0014_scene_0003_cfg_0002_036422060215#200", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004219/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 281, + 177 + ], + [ + 282, + 182 + ], + [ + 275, + 196 + ], + [ + 286, + 158 + ], + [ + 274, + 191 + ], + [ + 284, + 155 + ] + ], + [ + [ + 292, + 175 + ], + [ + 282, + 167 + ], + [ + 307, + 167 + ], + [ + 277, + 184 + ], + [ + 318, + 175 + ], + [ + 287, + 193 + ] + ], + [ + [ + 303, + 179 + ], + [ + 314, + 165 + ], + [ + 308, + 175 + ], + [ + 297, + 183 + ], + [ + 298, + 189 + ], + [ + 286, + 198 + ] + ], + [ + [ + 297, + 165 + ], + [ + 302, + 150 + ], + [ + 315, + 167 + ], + [ + 278, + 162 + ], + [ + 309, + 180 + ], + [ + 273, + 176 + ] + ] + ] + }, + { + "task_id": "0004220", + "id": "RH20T_cfg2_task_0071_user_0014_scene_0003_cfg_0002_036422060215#264", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004220/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 297, + 157 + ], + [ + 300, + 161 + ], + [ + 284, + 144 + ], + [ + 309, + 170 + ], + [ + 281, + 141 + ], + [ + 306, + 166 + ] + ], + [ + [ + 298, + 164 + ], + [ + 293, + 153 + ], + [ + 282, + 159 + ], + [ + 313, + 169 + ], + [ + 286, + 172 + ], + [ + 318, + 181 + ] + ], + [ + [ + 294, + 165 + ], + [ + 298, + 151 + ], + [ + 312, + 168 + ], + [ + 275, + 163 + ], + [ + 307, + 181 + ], + [ + 271, + 177 + ] + ], + [ + [ + 288, + 154 + ], + [ + 270, + 146 + ], + [ + 284, + 156 + ], + [ + 292, + 153 + ], + [ + 301, + 163 + ], + [ + 311, + 162 + ] + ] + ] + }, + { + "task_id": "0004221", + "id": "RH20T_cfg3_task_0059_user_0010_scene_0002_cfg_0003_104422070011#122", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004221/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 163 + ], + [ + 282, + 172 + ], + [ + 279, + 190 + ], + [ + 269, + 135 + ], + [ + 273, + 181 + ], + [ + 263, + 129 + ] + ], + [ + [ + 279, + 210 + ], + [ + 294, + 187 + ], + [ + 298, + 217 + ], + [ + 260, + 205 + ], + [ + 282, + 243 + ], + [ + 244, + 228 + ] + ], + [ + [ + 272, + 195 + ], + [ + 286, + 191 + ], + [ + 259, + 220 + ], + [ + 284, + 171 + ], + [ + 247, + 223 + ], + [ + 272, + 176 + ] + ], + [ + [ + 292, + 178 + ], + [ + 295, + 149 + ], + [ + 274, + 173 + ], + [ + 307, + 183 + ], + [ + 271, + 205 + ], + [ + 304, + 212 + ] + ] + ] + }, + { + "task_id": "0004226", + "id": "RH20T_cfg1_task_0043_user_0016_scene_0005_cfg_0001_038522063145#164", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004226/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 353, + 31 + ], + [ + 358, + 27 + ], + [ + 373, + 31 + ], + [ + 333, + 32 + ], + [ + 368, + 35 + ], + [ + 329, + 36 + ] + ], + [ + [ + 363, + 45 + ], + [ + 367, + 63 + ], + [ + 382, + 44 + ], + [ + 344, + 46 + ], + [ + 378, + 26 + ], + [ + 340, + 28 + ] + ], + [ + [ + 366, + 51 + ], + [ + 368, + 32 + ], + [ + 382, + 60 + ], + [ + 350, + 42 + ], + [ + 380, + 77 + ], + [ + 349, + 60 + ] + ], + [ + [ + 382, + 42 + ], + [ + 369, + 47 + ], + [ + 365, + 37 + ], + [ + 399, + 46 + ], + [ + 377, + 32 + ], + [ + 411, + 40 + ] + ] + ] + }, + { + "task_id": "0004228", + "id": "RH20T_cfg7_task_0017_user_0014_scene_0002_cfg_0007_104122061602#183", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the pen into the pen holder'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004228/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 418, + 167 + ], + [ + 426, + 154 + ], + [ + 429, + 187 + ], + [ + 408, + 148 + ], + [ + 422, + 198 + ], + [ + 401, + 160 + ] + ], + [ + [ + 397, + 149 + ], + [ + 384, + 135 + ], + [ + 378, + 160 + ], + [ + 416, + 138 + ], + [ + 391, + 176 + ], + [ + 431, + 153 + ] + ], + [ + [ + 448, + 166 + ], + [ + 469, + 161 + ], + [ + 460, + 162 + ], + [ + 436, + 170 + ], + [ + 438, + 169 + ], + [ + 416, + 176 + ] + ], + [ + [ + 442, + 173 + ], + [ + 420, + 182 + ], + [ + 435, + 155 + ], + [ + 449, + 192 + ], + [ + 457, + 146 + ], + [ + 473, + 183 + ] + ] + ] + }, + { + "task_id": "0004229", + "id": "RH20T_cfg6_task_0104_user_0020_scene_0004_cfg_0006_104122061602#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate back after holding it down'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004229/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 402, + 137 + ], + [ + 384, + 135 + ], + [ + 393, + 140 + ], + [ + 410, + 135 + ], + [ + 413, + 143 + ], + [ + 429, + 138 + ] + ], + [ + [ + 406, + 122 + ], + [ + 415, + 109 + ], + [ + 414, + 138 + ], + [ + 397, + 104 + ], + [ + 404, + 152 + ], + [ + 387, + 118 + ] + ], + [ + [ + 379, + 111 + ], + [ + 399, + 113 + ], + [ + 379, + 132 + ], + [ + 380, + 89 + ], + [ + 360, + 129 + ], + [ + 361, + 88 + ] + ], + [ + [ + 398, + 127 + ], + [ + 411, + 133 + ], + [ + 396, + 148 + ], + [ + 400, + 106 + ], + [ + 384, + 141 + ], + [ + 388, + 101 + ] + ] + ] + }, + { + "task_id": "0004233", + "id": "RH20T_cfg7_task_0033_user_0020_scene_0001_cfg_0007_104122060811#228", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004233/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 308, + 148 + ], + [ + 301, + 137 + ], + [ + 317, + 148 + ], + [ + 297, + 147 + ], + [ + 324, + 159 + ], + [ + 304, + 158 + ] + ], + [ + [ + 307, + 123 + ], + [ + 312, + 125 + ], + [ + 294, + 129 + ], + [ + 320, + 117 + ], + [ + 288, + 126 + ], + [ + 315, + 114 + ] + ], + [ + [ + 307, + 138 + ], + [ + 306, + 152 + ], + [ + 292, + 137 + ], + [ + 322, + 139 + ], + [ + 294, + 123 + ], + [ + 324, + 125 + ] + ], + [ + [ + 302, + 142 + ], + [ + 316, + 140 + ], + [ + 301, + 149 + ], + [ + 303, + 135 + ], + [ + 287, + 152 + ], + [ + 288, + 137 + ] + ] + ] + }, + { + "task_id": "0004234", + "id": "RH20T_cfg1_task_0020_user_0008_scene_0002_cfg_0001_038522063145#107", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004234/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 523, + 180 + ], + [ + 528, + 167 + ], + [ + 548, + 173 + ], + [ + 500, + 185 + ], + [ + 543, + 187 + ], + [ + 495, + 199 + ] + ], + [ + [ + 522, + 191 + ], + [ + 542, + 190 + ], + [ + 516, + 168 + ], + [ + 528, + 214 + ], + [ + 496, + 169 + ], + [ + 507, + 216 + ] + ], + [ + [ + 539, + 203 + ], + [ + 554, + 214 + ], + [ + 535, + 184 + ], + [ + 544, + 223 + ], + [ + 520, + 172 + ], + [ + 529, + 210 + ] + ], + [ + [ + 521, + 191 + ], + [ + 529, + 175 + ], + [ + 539, + 210 + ], + [ + 504, + 174 + ], + [ + 531, + 224 + ], + [ + 497, + 189 + ] + ] + ] + }, + { + "task_id": "0004235", + "id": "RH20T_cfg5_task_0001_user_0007_scene_0001_cfg_0005_104422070044#51", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004235/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 507, + 135 + ], + [ + 507, + 124 + ], + [ + 494, + 148 + ], + [ + 521, + 121 + ], + [ + 494, + 160 + ], + [ + 521, + 133 + ] + ], + [ + [ + 525, + 146 + ], + [ + 544, + 137 + ], + [ + 514, + 138 + ], + [ + 538, + 155 + ], + [ + 496, + 148 + ], + [ + 518, + 165 + ] + ], + [ + [ + 509, + 152 + ], + [ + 510, + 134 + ], + [ + 528, + 157 + ], + [ + 491, + 147 + ], + [ + 526, + 176 + ], + [ + 490, + 164 + ] + ], + [ + [ + 505, + 160 + ], + [ + 496, + 168 + ], + [ + 502, + 179 + ], + [ + 508, + 141 + ], + [ + 511, + 171 + ], + [ + 517, + 135 + ] + ] + ] + }, + { + "task_id": "0004236", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0009_cfg_0006_104122060811#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004236/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 379, + 148 + ], + [ + 384, + 161 + ], + [ + 386, + 152 + ], + [ + 372, + 144 + ], + [ + 381, + 140 + ], + [ + 367, + 132 + ] + ], + [ + [ + 388, + 169 + ], + [ + 390, + 181 + ], + [ + 390, + 160 + ], + [ + 385, + 177 + ], + [ + 388, + 148 + ], + [ + 383, + 166 + ] + ], + [ + [ + 406, + 167 + ], + [ + 421, + 167 + ], + [ + 405, + 181 + ], + [ + 406, + 152 + ], + [ + 391, + 181 + ], + [ + 391, + 152 + ] + ], + [ + [ + 386, + 159 + ], + [ + 399, + 152 + ], + [ + 393, + 161 + ], + [ + 379, + 156 + ], + [ + 381, + 168 + ], + [ + 367, + 163 + ] + ] + ] + }, + { + "task_id": "0004240", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0002_cfg_0006_104122064161#607", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004240/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 372, + 108 + ], + [ + 393, + 112 + ], + [ + 373, + 107 + ], + [ + 371, + 110 + ], + [ + 353, + 103 + ], + [ + 350, + 106 + ] + ], + [ + [ + 366, + 137 + ], + [ + 385, + 129 + ], + [ + 361, + 130 + ], + [ + 371, + 144 + ], + [ + 342, + 138 + ], + [ + 351, + 152 + ] + ], + [ + [ + 357, + 117 + ], + [ + 357, + 117 + ], + [ + 356, + 137 + ], + [ + 359, + 96 + ], + [ + 356, + 136 + ], + [ + 360, + 96 + ] + ], + [ + [ + 354, + 136 + ], + [ + 336, + 139 + ], + [ + 345, + 132 + ], + [ + 364, + 140 + ], + [ + 362, + 129 + ], + [ + 382, + 137 + ] + ] + ] + }, + { + "task_id": "0004241", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0002_cfg_0006_104122064161#753", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004241/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 342, + 125 + ], + [ + 360, + 136 + ], + [ + 334, + 143 + ], + [ + 351, + 107 + ], + [ + 315, + 132 + ], + [ + 333, + 95 + ] + ], + [ + [ + 356, + 137 + ], + [ + 364, + 129 + ], + [ + 351, + 115 + ], + [ + 361, + 158 + ], + [ + 341, + 123 + ], + [ + 353, + 167 + ] + ], + [ + [ + 360, + 141 + ], + [ + 359, + 128 + ], + [ + 360, + 160 + ], + [ + 360, + 123 + ], + [ + 360, + 171 + ], + [ + 360, + 136 + ] + ], + [ + [ + 361, + 153 + ], + [ + 382, + 145 + ], + [ + 370, + 158 + ], + [ + 353, + 147 + ], + [ + 349, + 166 + ], + [ + 333, + 154 + ] + ] + ] + }, + { + "task_id": "0004243", + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_035622060973#247", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004243/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 216, + 47 + ], + [ + 210, + 64 + ], + [ + 220, + 59 + ], + [ + 213, + 36 + ], + [ + 226, + 41 + ], + [ + 219, + 18 + ] + ], + [ + [ + 229, + 60 + ], + [ + 226, + 44 + ], + [ + 209, + 56 + ], + [ + 249, + 65 + ], + [ + 212, + 71 + ], + [ + 251, + 79 + ] + ], + [ + [ + 218, + 67 + ], + [ + 202, + 73 + ], + [ + 226, + 56 + ], + [ + 209, + 79 + ], + [ + 242, + 50 + ], + [ + 226, + 74 + ] + ], + [ + [ + 212, + 42 + ], + [ + 214, + 47 + ], + [ + 191, + 39 + ], + [ + 233, + 46 + ], + [ + 187, + 33 + ], + [ + 232, + 41 + ] + ] + ] + }, + { + "task_id": "0004245", + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_035622060973#543", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004245/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 239, + 68 + ], + [ + 236, + 51 + ], + [ + 221, + 61 + ], + [ + 256, + 74 + ], + [ + 225, + 77 + ], + [ + 258, + 89 + ] + ], + [ + [ + 239, + 90 + ], + [ + 240, + 72 + ], + [ + 257, + 86 + ], + [ + 220, + 94 + ], + [ + 255, + 102 + ], + [ + 220, + 111 + ] + ], + [ + [ + 234, + 74 + ], + [ + 219, + 77 + ], + [ + 232, + 53 + ], + [ + 235, + 94 + ], + [ + 247, + 51 + ], + [ + 249, + 91 + ] + ], + [ + [ + 251, + 93 + ], + [ + 261, + 81 + ], + [ + 257, + 109 + ], + [ + 243, + 76 + ], + [ + 248, + 122 + ], + [ + 233, + 89 + ] + ] + ] + }, + { + "task_id": "0004248", + "id": "RH20T_cfg5_task_0063_user_0007_scene_0007_cfg_0005_036422060215#178", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a shovel to scoop up an object'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004248/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 263, + 142 + ], + [ + 266, + 125 + ], + [ + 277, + 137 + ], + [ + 248, + 148 + ], + [ + 275, + 153 + ], + [ + 246, + 165 + ] + ], + [ + [ + 262, + 158 + ], + [ + 260, + 177 + ], + [ + 242, + 157 + ], + [ + 281, + 159 + ], + [ + 243, + 137 + ], + [ + 283, + 139 + ] + ], + [ + [ + 289, + 144 + ], + [ + 306, + 133 + ], + [ + 299, + 157 + ], + [ + 278, + 130 + ], + [ + 283, + 167 + ], + [ + 261, + 140 + ] + ], + [ + [ + 276, + 136 + ], + [ + 265, + 131 + ], + [ + 264, + 152 + ], + [ + 288, + 121 + ], + [ + 276, + 158 + ], + [ + 300, + 126 + ] + ] + ] + }, + { + "task_id": "0004254", + "id": "RH20T_cfg5_task_0091_user_0007_scene_0004_cfg_0005_105422061350#223", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004254/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 197, + 45 + ], + [ + 202, + 59 + ], + [ + 205, + 68 + ], + [ + 187, + 18 + ], + [ + 200, + 55 + ], + [ + 182, + 6 + ] + ], + [ + [ + 231, + 63 + ], + [ + 212, + 59 + ], + [ + 219, + 43 + ], + [ + 242, + 81 + ], + [ + 241, + 47 + ], + [ + 262, + 86 + ] + ], + [ + [ + 197, + 64 + ], + [ + 178, + 82 + ], + [ + 219, + 85 + ], + [ + 173, + 43 + ], + [ + 236, + 67 + ], + [ + 192, + 27 + ] + ], + [ + [ + 212, + 61 + ], + [ + 214, + 42 + ], + [ + 238, + 58 + ], + [ + 185, + 65 + ], + [ + 235, + 76 + ], + [ + 184, + 83 + ] + ] + ] + }, + { + "task_id": "0004258", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122061018#216", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004258/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 280, + 185 + ], + [ + 255, + 206 + ], + [ + 284, + 179 + ], + [ + 276, + 191 + ], + [ + 309, + 157 + ], + [ + 298, + 172 + ] + ], + [ + [ + 261, + 158 + ], + [ + 288, + 160 + ], + [ + 261, + 191 + ], + [ + 261, + 124 + ], + [ + 237, + 187 + ], + [ + 236, + 123 + ] + ], + [ + [ + 262, + 168 + ], + [ + 265, + 146 + ], + [ + 251, + 146 + ], + [ + 275, + 191 + ], + [ + 248, + 166 + ], + [ + 271, + 210 + ] + ], + [ + [ + 240, + 155 + ], + [ + 267, + 158 + ], + [ + 231, + 174 + ], + [ + 250, + 134 + ], + [ + 202, + 171 + ], + [ + 220, + 130 + ] + ] + ] + }, + { + "task_id": "0004259", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122061018#565", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004259/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 108 + ], + [ + 268, + 79 + ], + [ + 267, + 110 + ], + [ + 282, + 106 + ], + [ + 273, + 137 + ], + [ + 288, + 137 + ] + ], + [ + [ + 287, + 97 + ], + [ + 275, + 86 + ], + [ + 311, + 86 + ], + [ + 262, + 109 + ], + [ + 326, + 98 + ], + [ + 274, + 122 + ] + ], + [ + [ + 273, + 93 + ], + [ + 250, + 92 + ], + [ + 287, + 97 + ], + [ + 257, + 89 + ], + [ + 310, + 99 + ], + [ + 282, + 91 + ] + ], + [ + [ + 296, + 132 + ], + [ + 292, + 153 + ], + [ + 313, + 148 + ], + [ + 280, + 116 + ], + [ + 318, + 125 + ], + [ + 283, + 93 + ] + ] + ] + }, + { + "task_id": "0004262", + "id": "RH20T_cfg6_task_0225_user_0014_scene_0006_cfg_0006_037522061512#452", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004262/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 233, + 0 + ], + [ + 243, + 0 + ], + [ + 249, + 0 + ], + [ + 216, + 0 + ], + [ + 240, + 0 + ], + [ + 207, + 0 + ] + ], + [ + [ + 250, + 0 + ], + [ + 232, + 0 + ], + [ + 257, + 0 + ], + [ + 242, + 0 + ], + [ + 274, + 0 + ], + [ + 262, + 0 + ] + ], + [ + [ + 225, + 0 + ], + [ + 210, + 0 + ], + [ + 250, + 0 + ], + [ + 201, + 0 + ], + [ + 262, + 0 + ], + [ + 215, + 0 + ] + ], + [ + [ + 240, + 0 + ], + [ + 267, + 0 + ], + [ + 252, + 0 + ], + [ + 228, + 0 + ], + [ + 226, + 0 + ], + [ + 200, + 0 + ] + ] + ] + }, + { + "task_id": "0004265", + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_038522063145#147", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004265/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 440, + 135 + ], + [ + 438, + 152 + ], + [ + 442, + 115 + ], + [ + 438, + 157 + ], + [ + 444, + 96 + ], + [ + 440, + 139 + ] + ], + [ + [ + 453, + 103 + ], + [ + 449, + 114 + ], + [ + 442, + 78 + ], + [ + 463, + 127 + ], + [ + 446, + 69 + ], + [ + 465, + 117 + ] + ], + [ + [ + 479, + 129 + ], + [ + 485, + 156 + ], + [ + 484, + 130 + ], + [ + 473, + 128 + ], + [ + 478, + 104 + ], + [ + 467, + 100 + ] + ], + [ + [ + 450, + 142 + ], + [ + 463, + 118 + ], + [ + 476, + 153 + ], + [ + 424, + 130 + ], + [ + 463, + 174 + ], + [ + 412, + 152 + ] + ] + ] + }, + { + "task_id": "0004266", + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_038522063145#206", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004266/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 494, + 134 + ], + [ + 501, + 109 + ], + [ + 521, + 137 + ], + [ + 467, + 132 + ], + [ + 513, + 160 + ], + [ + 461, + 156 + ] + ], + [ + [ + 491, + 154 + ], + [ + 504, + 134 + ], + [ + 481, + 163 + ], + [ + 503, + 143 + ], + [ + 468, + 182 + ], + [ + 489, + 164 + ] + ], + [ + [ + 518, + 126 + ], + [ + 521, + 116 + ], + [ + 527, + 151 + ], + [ + 509, + 102 + ], + [ + 525, + 158 + ], + [ + 507, + 112 + ] + ], + [ + [ + 471, + 152 + ], + [ + 477, + 146 + ], + [ + 492, + 169 + ], + [ + 451, + 135 + ], + [ + 486, + 174 + ], + [ + 447, + 142 + ] + ] + ] + }, + { + "task_id": "0004268", + "id": "RH20T_cfg6_task_0127_user_0020_scene_0006_cfg_0006_104122061330#288", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004268/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 409, + 146 + ], + [ + 406, + 170 + ], + [ + 424, + 142 + ], + [ + 394, + 151 + ], + [ + 428, + 118 + ], + [ + 398, + 125 + ] + ], + [ + [ + 414, + 148 + ], + [ + 416, + 126 + ], + [ + 407, + 164 + ], + [ + 421, + 134 + ], + [ + 406, + 185 + ], + [ + 420, + 154 + ] + ], + [ + [ + 441, + 151 + ], + [ + 465, + 142 + ], + [ + 453, + 155 + ], + [ + 431, + 148 + ], + [ + 427, + 164 + ], + [ + 408, + 156 + ] + ], + [ + [ + 396, + 140 + ], + [ + 384, + 132 + ], + [ + 374, + 152 + ], + [ + 419, + 127 + ], + [ + 386, + 161 + ], + [ + 433, + 135 + ] + ] + ] + }, + { + "task_id": "0004269", + "id": "RH20T_cfg6_task_0127_user_0020_scene_0006_cfg_0006_104122061330#545", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004269/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 382, + 178 + ], + [ + 403, + 161 + ], + [ + 369, + 158 + ], + [ + 395, + 199 + ], + [ + 350, + 174 + ], + [ + 373, + 216 + ] + ], + [ + [ + 399, + 155 + ], + [ + 381, + 139 + ], + [ + 418, + 138 + ], + [ + 381, + 173 + ], + [ + 437, + 155 + ], + [ + 400, + 190 + ] + ], + [ + [ + 368, + 147 + ], + [ + 365, + 126 + ], + [ + 356, + 164 + ], + [ + 379, + 130 + ], + [ + 360, + 183 + ], + [ + 382, + 149 + ] + ], + [ + [ + 382, + 169 + ], + [ + 356, + 165 + ], + [ + 388, + 153 + ], + [ + 374, + 186 + ], + [ + 414, + 157 + ], + [ + 402, + 192 + ] + ] + ] + }, + { + "task_id": "0004270", + "id": "RH20T_cfg6_task_0127_user_0020_scene_0006_cfg_0006_104122061330#660", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004270/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 463, + 95 + ], + [ + 448, + 108 + ], + [ + 452, + 107 + ], + [ + 474, + 84 + ], + [ + 469, + 93 + ], + [ + 491, + 71 + ] + ], + [ + [ + 447, + 102 + ], + [ + 474, + 99 + ], + [ + 445, + 123 + ], + [ + 448, + 82 + ], + [ + 418, + 124 + ], + [ + 423, + 85 + ] + ], + [ + [ + 469, + 84 + ], + [ + 461, + 60 + ], + [ + 489, + 83 + ], + [ + 452, + 84 + ], + [ + 497, + 107 + ], + [ + 460, + 105 + ] + ], + [ + [ + 444, + 87 + ], + [ + 436, + 78 + ], + [ + 433, + 110 + ], + [ + 455, + 64 + ], + [ + 440, + 118 + ], + [ + 462, + 74 + ] + ] + ] + }, + { + "task_id": "0004271", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061330#124", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004271/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 131 + ], + [ + 339, + 139 + ], + [ + 352, + 122 + ], + [ + 374, + 138 + ], + [ + 378, + 113 + ], + [ + 398, + 131 + ] + ], + [ + [ + 357, + 142 + ], + [ + 375, + 161 + ], + [ + 377, + 126 + ], + [ + 339, + 157 + ], + [ + 359, + 106 + ], + [ + 322, + 138 + ] + ], + [ + [ + 357, + 144 + ], + [ + 367, + 160 + ], + [ + 368, + 123 + ], + [ + 346, + 164 + ], + [ + 358, + 108 + ], + [ + 337, + 149 + ] + ], + [ + [ + 347, + 132 + ], + [ + 343, + 115 + ], + [ + 334, + 152 + ], + [ + 359, + 112 + ], + [ + 339, + 167 + ], + [ + 362, + 128 + ] + ] + ] + }, + { + "task_id": "0004272", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061330#252", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004272/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 359, + 152 + ], + [ + 338, + 146 + ], + [ + 364, + 128 + ], + [ + 355, + 175 + ], + [ + 387, + 134 + ], + [ + 377, + 182 + ] + ], + [ + [ + 371, + 145 + ], + [ + 367, + 129 + ], + [ + 360, + 166 + ], + [ + 381, + 124 + ], + [ + 365, + 179 + ], + [ + 384, + 139 + ] + ], + [ + [ + 389, + 144 + ], + [ + 401, + 134 + ], + [ + 372, + 162 + ], + [ + 404, + 126 + ], + [ + 362, + 171 + ], + [ + 392, + 136 + ] + ], + [ + [ + 365, + 136 + ], + [ + 353, + 145 + ], + [ + 388, + 144 + ], + [ + 341, + 128 + ], + [ + 402, + 134 + ], + [ + 352, + 116 + ] + ] + ] + }, + { + "task_id": "0004273", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061330#365", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004273/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 341, + 178 + ], + [ + 316, + 176 + ], + [ + 347, + 187 + ], + [ + 333, + 168 + ], + [ + 372, + 189 + ], + [ + 360, + 170 + ] + ], + [ + [ + 316, + 133 + ], + [ + 309, + 149 + ], + [ + 293, + 123 + ], + [ + 339, + 143 + ], + [ + 300, + 105 + ], + [ + 349, + 127 + ] + ], + [ + [ + 336, + 149 + ], + [ + 323, + 152 + ], + [ + 331, + 173 + ], + [ + 340, + 123 + ], + [ + 342, + 168 + ], + [ + 352, + 121 + ] + ], + [ + [ + 343, + 131 + ], + [ + 368, + 130 + ], + [ + 353, + 126 + ], + [ + 334, + 135 + ], + [ + 325, + 127 + ], + [ + 310, + 136 + ] + ] + ] + }, + { + "task_id": "0004274", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0003_cfg_0007_104122061330#492", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004274/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 358, + 106 + ], + [ + 357, + 121 + ], + [ + 334, + 97 + ], + [ + 382, + 116 + ], + [ + 334, + 80 + ], + [ + 383, + 99 + ] + ], + [ + [ + 322, + 121 + ], + [ + 346, + 118 + ], + [ + 315, + 136 + ], + [ + 329, + 106 + ], + [ + 293, + 139 + ], + [ + 307, + 109 + ] + ], + [ + [ + 339, + 123 + ], + [ + 338, + 101 + ], + [ + 326, + 135 + ], + [ + 351, + 112 + ], + [ + 327, + 157 + ], + [ + 352, + 134 + ] + ], + [ + [ + 341, + 95 + ], + [ + 325, + 106 + ], + [ + 353, + 117 + ], + [ + 329, + 73 + ], + [ + 368, + 106 + ], + [ + 345, + 64 + ] + ] + ] + }, + { + "task_id": "0004276", + "id": "RH20T_cfg1_task_0041_user_0001_scene_0006_cfg_0001_750612070853#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the tabletop with a sponge'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004276/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 312, + 262 + ], + [ + 284, + 221 + ], + [ + 335, + 250 + ], + [ + 284, + 277 + ], + [ + 361, + 289 + ], + [ + 315, + 325 + ] + ], + [ + [ + 350, + 261 + ], + [ + 347, + 223 + ], + [ + 319, + 240 + ], + [ + 385, + 284 + ], + [ + 323, + 273 + ], + [ + 386, + 319 + ] + ], + [ + [ + 338, + 262 + ], + [ + 374, + 246 + ], + [ + 305, + 226 + ], + [ + 375, + 303 + ], + [ + 263, + 242 + ], + [ + 334, + 328 + ] + ], + [ + [ + 345, + 308 + ], + [ + 381, + 282 + ], + [ + 377, + 333 + ], + [ + 317, + 288 + ], + [ + 335, + 360 + ], + [ + 279, + 314 + ] + ] + ] + }, + { + "task_id": "0004278", + "id": "RH20T_cfg4_task_0120_user_0007_scene_0004_cfg_0004_104122062823#297", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004278/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 292, + 69 + ], + [ + 289, + 35 + ], + [ + 322, + 57 + ], + [ + 260, + 83 + ], + [ + 323, + 87 + ], + [ + 264, + 114 + ] + ], + [ + [ + 286, + 34 + ], + [ + 289, + 72 + ], + [ + 253, + 43 + ], + [ + 322, + 24 + ], + [ + 250, + 8 + ], + [ + 319, + 0 + ] + ], + [ + [ + 286, + 92 + ], + [ + 282, + 88 + ], + [ + 252, + 77 + ], + [ + 319, + 105 + ], + [ + 254, + 81 + ], + [ + 326, + 111 + ] + ], + [ + [ + 268, + 69 + ], + [ + 303, + 78 + ], + [ + 263, + 95 + ], + [ + 274, + 39 + ], + [ + 229, + 87 + ], + [ + 237, + 30 + ] + ] + ] + }, + { + "task_id": "0004279", + "id": "RH20T_cfg4_task_0120_user_0007_scene_0004_cfg_0004_104122062823#538", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004279/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 327, + 83 + ], + [ + 363, + 72 + ], + [ + 314, + 46 + ], + [ + 339, + 119 + ], + [ + 280, + 59 + ], + [ + 306, + 128 + ] + ], + [ + [ + 304, + 59 + ], + [ + 302, + 24 + ], + [ + 332, + 46 + ], + [ + 274, + 73 + ], + [ + 333, + 78 + ], + [ + 278, + 105 + ] + ], + [ + [ + 284, + 41 + ], + [ + 265, + 71 + ], + [ + 257, + 17 + ], + [ + 308, + 63 + ], + [ + 276, + 0 + ], + [ + 324, + 34 + ] + ], + [ + [ + 313, + 26 + ], + [ + 290, + 9 + ], + [ + 289, + 58 + ], + [ + 338, + 0 + ], + [ + 313, + 78 + ], + [ + 365, + 11 + ] + ] + ] + }, + { + "task_id": "0004281", + "id": "RH20T_cfg4_task_0120_user_0007_scene_0004_cfg_0004_104122062823#681", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004281/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 295, + 75 + ], + [ + 292, + 41 + ], + [ + 327, + 64 + ], + [ + 261, + 87 + ], + [ + 328, + 95 + ], + [ + 266, + 118 + ] + ], + [ + [ + 314, + 67 + ], + [ + 349, + 79 + ], + [ + 309, + 83 + ], + [ + 317, + 54 + ], + [ + 270, + 70 + ], + [ + 283, + 43 + ] + ], + [ + [ + 308, + 31 + ], + [ + 311, + 52 + ], + [ + 274, + 46 + ], + [ + 344, + 15 + ], + [ + 274, + 27 + ], + [ + 340, + 0 + ] + ], + [ + [ + 273, + 69 + ], + [ + 267, + 56 + ], + [ + 260, + 37 + ], + [ + 284, + 99 + ], + [ + 266, + 50 + ], + [ + 291, + 115 + ] + ] + ] + }, + { + "task_id": "0004282", + "id": "RH20T_cfg4_task_0120_user_0007_scene_0004_cfg_0004_104122062823#741", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Install the light bulb by rotating it'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004282/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 265, + 41 + ], + [ + 246, + 25 + ], + [ + 256, + 6 + ], + [ + 272, + 72 + ], + [ + 277, + 22 + ], + [ + 292, + 90 + ] + ], + [ + [ + 285, + 27 + ], + [ + 264, + 0 + ], + [ + 254, + 45 + ], + [ + 318, + 8 + ], + [ + 275, + 68 + ], + [ + 336, + 34 + ] + ], + [ + [ + 325, + 111 + ], + [ + 351, + 136 + ], + [ + 297, + 133 + ], + [ + 352, + 91 + ], + [ + 273, + 109 + ], + [ + 328, + 69 + ] + ], + [ + [ + 303, + 76 + ], + [ + 300, + 41 + ], + [ + 333, + 63 + ], + [ + 271, + 90 + ], + [ + 335, + 94 + ], + [ + 275, + 121 + ] + ] + ] + }, + { + "task_id": "0004285", + "id": "RH20T_cfg3_task_0023_user_0010_scene_0004_cfg_0003_104422070011#302", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004285/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 231, + 155 + ], + [ + 236, + 144 + ], + [ + 227, + 180 + ], + [ + 234, + 130 + ], + [ + 222, + 189 + ], + [ + 229, + 141 + ] + ], + [ + [ + 252, + 164 + ], + [ + 277, + 160 + ], + [ + 248, + 149 + ], + [ + 257, + 181 + ], + [ + 223, + 152 + ], + [ + 230, + 186 + ] + ], + [ + [ + 247, + 170 + ], + [ + 224, + 169 + ], + [ + 263, + 164 + ], + [ + 232, + 175 + ], + [ + 284, + 165 + ], + [ + 252, + 175 + ] + ], + [ + [ + 239, + 163 + ], + [ + 247, + 186 + ], + [ + 260, + 148 + ], + [ + 220, + 177 + ], + [ + 251, + 125 + ], + [ + 213, + 154 + ] + ] + ] + }, + { + "task_id": "0004289", + "id": "RH20T_cfg3_task_0091_user_0010_scene_0004_cfg_0003_104122062295#557", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004289/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 166, + 178 + ], + [ + 146, + 175 + ], + [ + 196, + 155 + ], + [ + 132, + 203 + ], + [ + 218, + 157 + ], + [ + 153, + 209 + ] + ], + [ + [ + 146, + 229 + ], + [ + 141, + 195 + ], + [ + 101, + 246 + ], + [ + 186, + 214 + ], + [ + 109, + 278 + ], + [ + 189, + 244 + ] + ], + [ + [ + 107, + 236 + ], + [ + 124, + 265 + ], + [ + 67, + 231 + ], + [ + 141, + 241 + ], + [ + 52, + 202 + ], + [ + 124, + 217 + ] + ], + [ + [ + 96, + 243 + ], + [ + 99, + 248 + ], + [ + 135, + 264 + ], + [ + 56, + 221 + ], + [ + 134, + 260 + ], + [ + 48, + 213 + ] + ] + ] + }, + { + "task_id": "0004292", + "id": "RH20T_cfg5_task_0041_user_0007_scene_0006_cfg_0005_036422060215#262", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the tabletop with a sponge'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004292/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 220, + 183 + ], + [ + 203, + 191 + ], + [ + 230, + 202 + ], + [ + 210, + 163 + ], + [ + 246, + 194 + ], + [ + 227, + 156 + ] + ], + [ + [ + 208, + 174 + ], + [ + 209, + 156 + ], + [ + 230, + 171 + ], + [ + 185, + 176 + ], + [ + 229, + 188 + ], + [ + 185, + 193 + ] + ], + [ + [ + 222, + 178 + ], + [ + 225, + 160 + ], + [ + 246, + 182 + ], + [ + 199, + 174 + ], + [ + 243, + 200 + ], + [ + 195, + 192 + ] + ], + [ + [ + 218, + 190 + ], + [ + 216, + 181 + ], + [ + 234, + 206 + ], + [ + 203, + 175 + ], + [ + 235, + 213 + ], + [ + 206, + 184 + ] + ] + ] + }, + { + "task_id": "0004293", + "id": "RH20T_cfg3_task_0018_user_0011_scene_0002_cfg_0003_104122063550#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004293/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 289, + 199 + ], + [ + 273, + 179 + ], + [ + 267, + 217 + ], + [ + 310, + 181 + ], + [ + 283, + 238 + ], + [ + 327, + 202 + ] + ], + [ + [ + 293, + 205 + ], + [ + 314, + 223 + ], + [ + 312, + 184 + ], + [ + 275, + 226 + ], + [ + 291, + 165 + ], + [ + 254, + 207 + ] + ], + [ + [ + 320, + 199 + ], + [ + 317, + 175 + ], + [ + 294, + 202 + ], + [ + 346, + 197 + ], + [ + 297, + 228 + ], + [ + 350, + 223 + ] + ], + [ + [ + 309, + 225 + ], + [ + 311, + 216 + ], + [ + 334, + 218 + ], + [ + 282, + 233 + ], + [ + 331, + 226 + ], + [ + 282, + 240 + ] + ] + ] + }, + { + "task_id": "0004295", + "id": "RH20T_cfg4_task_0005_user_0010_scene_0004_cfg_0004_104122062823#106", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004295/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 356, + 95 + ], + [ + 337, + 84 + ], + [ + 333, + 115 + ], + [ + 381, + 74 + ], + [ + 351, + 124 + ], + [ + 398, + 86 + ] + ], + [ + [ + 351, + 83 + ], + [ + 352, + 52 + ], + [ + 384, + 92 + ], + [ + 319, + 73 + ], + [ + 381, + 120 + ], + [ + 320, + 100 + ] + ], + [ + [ + 384, + 73 + ], + [ + 396, + 38 + ], + [ + 357, + 55 + ], + [ + 408, + 90 + ], + [ + 343, + 92 + ], + [ + 396, + 124 + ] + ], + [ + [ + 358, + 102 + ], + [ + 336, + 96 + ], + [ + 384, + 98 + ], + [ + 330, + 107 + ], + [ + 410, + 104 + ], + [ + 354, + 114 + ] + ] + ] + }, + { + "task_id": "0004299", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122064161#229", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004299/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 426, + 220 + ], + [ + 430, + 201 + ], + [ + 431, + 237 + ], + [ + 421, + 204 + ], + [ + 426, + 255 + ], + [ + 417, + 221 + ] + ], + [ + [ + 402, + 232 + ], + [ + 393, + 211 + ], + [ + 404, + 222 + ], + [ + 399, + 241 + ], + [ + 413, + 245 + ], + [ + 407, + 261 + ] + ], + [ + [ + 439, + 212 + ], + [ + 462, + 213 + ], + [ + 438, + 189 + ], + [ + 440, + 235 + ], + [ + 416, + 188 + ], + [ + 417, + 234 + ] + ], + [ + [ + 425, + 215 + ], + [ + 412, + 212 + ], + [ + 439, + 232 + ], + [ + 412, + 199 + ], + [ + 451, + 234 + ], + [ + 425, + 202 + ] + ] + ] + }, + { + "task_id": "0004302", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122064161#565", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004302/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 424, + 186 + ], + [ + 401, + 184 + ], + [ + 423, + 196 + ], + [ + 425, + 177 + ], + [ + 447, + 198 + ], + [ + 447, + 179 + ] + ], + [ + [ + 422, + 152 + ], + [ + 404, + 165 + ], + [ + 436, + 170 + ], + [ + 408, + 134 + ], + [ + 455, + 155 + ], + [ + 426, + 120 + ] + ], + [ + [ + 420, + 171 + ], + [ + 425, + 156 + ], + [ + 422, + 189 + ], + [ + 418, + 155 + ], + [ + 417, + 203 + ], + [ + 413, + 169 + ] + ], + [ + [ + 406, + 198 + ], + [ + 393, + 215 + ], + [ + 402, + 198 + ], + [ + 411, + 197 + ], + [ + 414, + 181 + ], + [ + 425, + 179 + ] + ] + ] + }, + { + "task_id": "0004304", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522063145#150", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004304/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 440, + 114 + ], + [ + 436, + 106 + ], + [ + 417, + 114 + ], + [ + 463, + 114 + ], + [ + 419, + 122 + ], + [ + 467, + 122 + ] + ], + [ + [ + 414, + 119 + ], + [ + 419, + 101 + ], + [ + 434, + 128 + ], + [ + 393, + 111 + ], + [ + 429, + 145 + ], + [ + 389, + 128 + ] + ], + [ + [ + 415, + 138 + ], + [ + 418, + 160 + ], + [ + 396, + 141 + ], + [ + 434, + 136 + ], + [ + 393, + 119 + ], + [ + 431, + 114 + ] + ], + [ + [ + 430, + 122 + ], + [ + 420, + 127 + ], + [ + 451, + 119 + ], + [ + 410, + 125 + ], + [ + 461, + 114 + ], + [ + 420, + 121 + ] + ] + ] + }, + { + "task_id": "0004305", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522063145#292", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004305/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 431, + 143 + ], + [ + 417, + 128 + ], + [ + 418, + 149 + ], + [ + 443, + 138 + ], + [ + 433, + 165 + ], + [ + 458, + 154 + ] + ], + [ + [ + 441, + 131 + ], + [ + 432, + 118 + ], + [ + 462, + 120 + ], + [ + 420, + 141 + ], + [ + 473, + 133 + ], + [ + 428, + 155 + ] + ], + [ + [ + 429, + 149 + ], + [ + 434, + 133 + ], + [ + 453, + 151 + ], + [ + 406, + 148 + ], + [ + 448, + 166 + ], + [ + 403, + 163 + ] + ], + [ + [ + 434, + 144 + ], + [ + 426, + 138 + ], + [ + 429, + 123 + ], + [ + 439, + 164 + ], + [ + 439, + 130 + ], + [ + 448, + 172 + ] + ] + ] + }, + { + "task_id": "0004306", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522063145#430", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004306/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 382, + 136 + ], + [ + 387, + 119 + ], + [ + 404, + 145 + ], + [ + 360, + 127 + ], + [ + 398, + 160 + ], + [ + 357, + 142 + ] + ], + [ + [ + 400, + 112 + ], + [ + 385, + 127 + ], + [ + 382, + 97 + ], + [ + 419, + 127 + ], + [ + 397, + 84 + ], + [ + 434, + 112 + ] + ], + [ + [ + 377, + 148 + ], + [ + 365, + 162 + ], + [ + 379, + 131 + ], + [ + 376, + 165 + ], + [ + 391, + 118 + ], + [ + 388, + 152 + ] + ], + [ + [ + 356, + 142 + ], + [ + 362, + 123 + ], + [ + 335, + 132 + ], + [ + 376, + 152 + ], + [ + 329, + 152 + ], + [ + 370, + 171 + ] + ] + ] + }, + { + "task_id": "0004307", + "id": "RH20T_cfg1_task_0037_user_0014_scene_0001_cfg_0001_038522063145#640", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004307/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 431, + 171 + ], + [ + 432, + 155 + ], + [ + 421, + 188 + ], + [ + 442, + 152 + ], + [ + 420, + 206 + ], + [ + 442, + 169 + ] + ], + [ + [ + 420, + 183 + ], + [ + 431, + 204 + ], + [ + 441, + 172 + ], + [ + 400, + 194 + ], + [ + 430, + 151 + ], + [ + 390, + 174 + ] + ], + [ + [ + 407, + 188 + ], + [ + 384, + 185 + ], + [ + 405, + 165 + ], + [ + 409, + 210 + ], + [ + 429, + 167 + ], + [ + 433, + 213 + ] + ], + [ + [ + 426, + 174 + ], + [ + 431, + 157 + ], + [ + 447, + 168 + ], + [ + 404, + 180 + ], + [ + 442, + 184 + ], + [ + 400, + 196 + ] + ] + ] + }, + { + "task_id": "0004308", + "id": "RH20T_cfg1_task_0038_user_0015_scene_0010_cfg_0001_750612070853#150", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004308/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 405, + 164 + ], + [ + 405, + 115 + ], + [ + 357, + 155 + ], + [ + 459, + 175 + ], + [ + 359, + 198 + ], + [ + 456, + 221 + ] + ], + [ + [ + 437, + 173 + ], + [ + 411, + 158 + ], + [ + 468, + 131 + ], + [ + 408, + 214 + ], + [ + 504, + 145 + ], + [ + 436, + 236 + ] + ], + [ + [ + 373, + 159 + ], + [ + 348, + 122 + ], + [ + 357, + 145 + ], + [ + 394, + 177 + ], + [ + 378, + 176 + ], + [ + 418, + 213 + ] + ], + [ + [ + 356, + 206 + ], + [ + 375, + 256 + ], + [ + 323, + 221 + ], + [ + 385, + 193 + ], + [ + 303, + 167 + ], + [ + 368, + 147 + ] + ] + ] + }, + { + "task_id": "0004310", + "id": "RH20T_cfg1_task_0038_user_0014_scene_0004_cfg_0001_035622060973#79", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004310/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 207 + ], + [ + 297, + 203 + ], + [ + 309, + 229 + ], + [ + 321, + 185 + ], + [ + 326, + 233 + ], + [ + 338, + 190 + ] + ], + [ + [ + 291, + 207 + ], + [ + 291, + 220 + ], + [ + 305, + 220 + ], + [ + 276, + 192 + ], + [ + 305, + 208 + ], + [ + 276, + 181 + ] + ], + [ + [ + 325, + 186 + ], + [ + 313, + 183 + ], + [ + 338, + 202 + ], + [ + 311, + 171 + ], + [ + 351, + 206 + ], + [ + 323, + 173 + ] + ], + [ + [ + 310, + 203 + ], + [ + 309, + 192 + ], + [ + 292, + 192 + ], + [ + 330, + 216 + ], + [ + 294, + 203 + ], + [ + 330, + 226 + ] + ] + ] + }, + { + "task_id": "0004312", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0008_cfg_0004_104122062295#235", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004312/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 231, + 150 + ], + [ + 238, + 115 + ], + [ + 191, + 149 + ], + [ + 267, + 152 + ], + [ + 186, + 183 + ], + [ + 259, + 184 + ] + ], + [ + [ + 210, + 161 + ], + [ + 242, + 178 + ], + [ + 228, + 131 + ], + [ + 191, + 192 + ], + [ + 197, + 114 + ], + [ + 158, + 175 + ] + ], + [ + [ + 266, + 131 + ], + [ + 234, + 154 + ], + [ + 286, + 162 + ], + [ + 246, + 102 + ], + [ + 319, + 139 + ], + [ + 277, + 79 + ] + ], + [ + [ + 222, + 160 + ], + [ + 225, + 191 + ], + [ + 212, + 178 + ], + [ + 232, + 140 + ], + [ + 210, + 150 + ], + [ + 229, + 110 + ] + ] + ] + }, + { + "task_id": "0004313", + "id": "RH20T_cfg7_task_0055_user_0020_scene_0007_cfg_0007_104122061018#134", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004313/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 121 + ], + [ + 316, + 149 + ], + [ + 302, + 119 + ], + [ + 347, + 123 + ], + [ + 313, + 87 + ], + [ + 357, + 94 + ] + ], + [ + [ + 331, + 157 + ], + [ + 344, + 158 + ], + [ + 303, + 147 + ], + [ + 361, + 168 + ], + [ + 287, + 145 + ], + [ + 347, + 167 + ] + ], + [ + [ + 300, + 121 + ], + [ + 282, + 146 + ], + [ + 318, + 123 + ], + [ + 280, + 118 + ], + [ + 337, + 98 + ], + [ + 301, + 90 + ] + ], + [ + [ + 323, + 125 + ], + [ + 331, + 99 + ], + [ + 309, + 103 + ], + [ + 338, + 150 + ], + [ + 303, + 126 + ], + [ + 330, + 172 + ] + ] + ] + }, + { + "task_id": "0004316", + "id": "RH20T_cfg4_task_0047_user_0014_scene_0012_cfg_0004_104122062823#203", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004316/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 229, + 78 + ], + [ + 239, + 38 + ], + [ + 193, + 68 + ], + [ + 266, + 88 + ], + [ + 183, + 107 + ], + [ + 256, + 128 + ] + ], + [ + [ + 254, + 34 + ], + [ + 257, + 42 + ], + [ + 251, + 78 + ], + [ + 258, + 0 + ], + [ + 249, + 68 + ], + [ + 256, + 0 + ] + ], + [ + [ + 231, + 32 + ], + [ + 242, + 2 + ], + [ + 253, + 17 + ], + [ + 206, + 50 + ], + [ + 242, + 43 + ], + [ + 197, + 77 + ] + ], + [ + [ + 240, + 48 + ], + [ + 222, + 36 + ], + [ + 208, + 47 + ], + [ + 275, + 48 + ], + [ + 225, + 58 + ], + [ + 288, + 59 + ] + ] + ] + }, + { + "task_id": "0004319", + "id": "RH20T_cfg7_task_0065_user_0014_scene_0003_cfg_0007_104122063633#471", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004319/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 126, + 184 + ], + [ + 142, + 168 + ], + [ + 138, + 195 + ], + [ + 115, + 174 + ], + [ + 122, + 213 + ], + [ + 99, + 191 + ] + ], + [ + [ + 159, + 174 + ], + [ + 140, + 164 + ], + [ + 153, + 192 + ], + [ + 166, + 154 + ], + [ + 172, + 202 + ], + [ + 186, + 164 + ] + ], + [ + [ + 135, + 191 + ], + [ + 132, + 174 + ], + [ + 140, + 175 + ], + [ + 129, + 208 + ], + [ + 143, + 190 + ], + [ + 132, + 223 + ] + ], + [ + [ + 130, + 205 + ], + [ + 117, + 208 + ], + [ + 116, + 221 + ], + [ + 144, + 190 + ], + [ + 130, + 218 + ], + [ + 158, + 186 + ] + ] + ] + }, + { + "task_id": "0004322", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0009_cfg_0003_104122062295#299", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004322/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 234, + 312 + ], + [ + 239, + 270 + ], + [ + 209, + 274 + ], + [ + 264, + 358 + ], + [ + 206, + 310 + ], + [ + 257, + 360 + ] + ], + [ + [ + 230, + 336 + ], + [ + 237, + 360 + ], + [ + 232, + 323 + ], + [ + 228, + 352 + ], + [ + 227, + 273 + ], + [ + 221, + 288 + ] + ], + [ + [ + 229, + 355 + ], + [ + 250, + 350 + ], + [ + 197, + 360 + ], + [ + 257, + 306 + ], + [ + 182, + 360 + ], + [ + 239, + 314 + ] + ], + [ + [ + 230, + 360 + ], + [ + 215, + 360 + ], + [ + 207, + 312 + ], + [ + 256, + 360 + ], + [ + 221, + 319 + ], + [ + 266, + 360 + ] + ] + ] + }, + { + "task_id": "0004323", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0001_cfg_0006_104122061330#208", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004323/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 379, + 114 + ], + [ + 378, + 95 + ], + [ + 368, + 133 + ], + [ + 389, + 96 + ], + [ + 370, + 151 + ], + [ + 390, + 114 + ] + ], + [ + [ + 392, + 114 + ], + [ + 417, + 127 + ], + [ + 403, + 90 + ], + [ + 380, + 139 + ], + [ + 379, + 77 + ], + [ + 354, + 125 + ] + ], + [ + [ + 389, + 143 + ], + [ + 385, + 117 + ], + [ + 398, + 149 + ], + [ + 382, + 138 + ], + [ + 402, + 176 + ], + [ + 386, + 162 + ] + ], + [ + [ + 363, + 111 + ], + [ + 385, + 98 + ], + [ + 350, + 90 + ], + [ + 375, + 132 + ], + [ + 328, + 102 + ], + [ + 354, + 143 + ] + ] + ] + }, + { + "task_id": "0004326", + "id": "RH20T_cfg2_task_0066_user_0014_scene_0006_cfg_0002_105422061350#643", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004326/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 303, + 32 + ], + [ + 287, + 32 + ], + [ + 304, + 55 + ], + [ + 301, + 9 + ], + [ + 318, + 54 + ], + [ + 316, + 10 + ] + ], + [ + [ + 284, + 24 + ], + [ + 288, + 5 + ], + [ + 305, + 33 + ], + [ + 262, + 15 + ], + [ + 300, + 50 + ], + [ + 259, + 33 + ] + ], + [ + [ + 264, + 18 + ], + [ + 255, + 28 + ], + [ + 286, + 23 + ], + [ + 243, + 13 + ], + [ + 293, + 13 + ], + [ + 252, + 3 + ] + ], + [ + [ + 274, + 45 + ], + [ + 254, + 56 + ], + [ + 262, + 28 + ], + [ + 285, + 60 + ], + [ + 282, + 15 + ], + [ + 304, + 49 + ] + ] + ] + }, + { + "task_id": "0004328", + "id": "RH20T_cfg2_task_0066_user_0014_scene_0006_cfg_0002_105422061350#1188", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004328/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 343, + 48 + ], + [ + 348, + 33 + ], + [ + 345, + 30 + ], + [ + 341, + 67 + ], + [ + 341, + 45 + ], + [ + 336, + 81 + ] + ], + [ + [ + 355, + 63 + ], + [ + 346, + 42 + ], + [ + 365, + 53 + ], + [ + 344, + 74 + ], + [ + 373, + 72 + ], + [ + 353, + 94 + ] + ], + [ + [ + 337, + 34 + ], + [ + 352, + 29 + ], + [ + 349, + 53 + ], + [ + 324, + 15 + ], + [ + 334, + 59 + ], + [ + 308, + 20 + ] + ], + [ + [ + 333, + 52 + ], + [ + 335, + 28 + ], + [ + 350, + 51 + ], + [ + 317, + 53 + ], + [ + 348, + 76 + ], + [ + 315, + 76 + ] + ] + ] + }, + { + "task_id": "0004330", + "id": "RH20T_cfg2_task_0066_user_0014_scene_0006_cfg_0002_105422061350#1764", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004330/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 169, + 232 + ], + [ + 162, + 214 + ], + [ + 195, + 245 + ], + [ + 144, + 219 + ], + [ + 199, + 260 + ], + [ + 151, + 236 + ] + ], + [ + [ + 158, + 219 + ], + [ + 152, + 226 + ], + [ + 168, + 249 + ], + [ + 147, + 189 + ], + [ + 173, + 241 + ], + [ + 153, + 185 + ] + ], + [ + [ + 166, + 253 + ], + [ + 145, + 258 + ], + [ + 159, + 283 + ], + [ + 173, + 221 + ], + [ + 178, + 277 + ], + [ + 191, + 219 + ] + ], + [ + [ + 188, + 206 + ], + [ + 216, + 191 + ], + [ + 199, + 226 + ], + [ + 177, + 185 + ], + [ + 171, + 241 + ], + [ + 146, + 201 + ] + ] + ] + }, + { + "task_id": "0004331", + "id": "RH20T_cfg4_task_0009_user_0010_scene_0004_cfg_0004_038522062288#115", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004331/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 229, + 185 + ], + [ + 229, + 163 + ], + [ + 259, + 184 + ], + [ + 198, + 186 + ], + [ + 258, + 203 + ], + [ + 199, + 206 + ] + ], + [ + [ + 216, + 189 + ], + [ + 209, + 170 + ], + [ + 189, + 177 + ], + [ + 243, + 201 + ], + [ + 197, + 194 + ], + [ + 248, + 218 + ] + ], + [ + [ + 218, + 211 + ], + [ + 189, + 222 + ], + [ + 233, + 216 + ], + [ + 202, + 206 + ], + [ + 261, + 205 + ], + [ + 234, + 194 + ] + ], + [ + [ + 203, + 193 + ], + [ + 183, + 219 + ], + [ + 182, + 190 + ], + [ + 225, + 197 + ], + [ + 200, + 167 + ], + [ + 243, + 173 + ] + ] + ] + }, + { + "task_id": "0004332", + "id": "RH20T_cfg6_task_0038_user_0014_scene_0006_cfg_0006_104122061602#103", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004332/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 477, + 136 + ], + [ + 495, + 122 + ], + [ + 468, + 150 + ], + [ + 487, + 119 + ], + [ + 450, + 164 + ], + [ + 468, + 134 + ] + ], + [ + [ + 491, + 157 + ], + [ + 494, + 132 + ], + [ + 513, + 171 + ], + [ + 471, + 145 + ], + [ + 510, + 195 + ], + [ + 469, + 168 + ] + ], + [ + [ + 493, + 175 + ], + [ + 521, + 164 + ], + [ + 497, + 182 + ], + [ + 490, + 169 + ], + [ + 468, + 193 + ], + [ + 464, + 179 + ] + ], + [ + [ + 500, + 171 + ], + [ + 491, + 198 + ], + [ + 521, + 178 + ], + [ + 477, + 165 + ], + [ + 530, + 151 + ], + [ + 485, + 138 + ] + ] + ] + }, + { + "task_id": "0004335", + "id": "RH20T_cfg1_task_0010_user_0001_scene_0005_cfg_0001_819112070044#356", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004335/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 90 + ], + [ + 309, + 75 + ], + [ + 294, + 78 + ], + [ + 317, + 103 + ], + [ + 290, + 93 + ], + [ + 313, + 117 + ] + ], + [ + [ + 300, + 94 + ], + [ + 295, + 76 + ], + [ + 289, + 106 + ], + [ + 310, + 83 + ], + [ + 294, + 123 + ], + [ + 314, + 99 + ] + ], + [ + [ + 296, + 86 + ], + [ + 316, + 84 + ], + [ + 293, + 80 + ], + [ + 300, + 93 + ], + [ + 272, + 82 + ], + [ + 281, + 95 + ] + ], + [ + [ + 302, + 100 + ], + [ + 302, + 84 + ], + [ + 284, + 106 + ], + [ + 320, + 95 + ], + [ + 284, + 121 + ], + [ + 320, + 110 + ] + ] + ] + }, + { + "task_id": "0004336", + "id": "RH20T_cfg5_task_0001_user_0020_scene_0001_cfg_0005_037522062165#299", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004336/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 256, + 202 + ], + [ + 259, + 177 + ], + [ + 276, + 214 + ], + [ + 234, + 190 + ], + [ + 273, + 240 + ], + [ + 229, + 218 + ] + ], + [ + [ + 220, + 205 + ], + [ + 245, + 200 + ], + [ + 239, + 196 + ], + [ + 202, + 213 + ], + [ + 210, + 201 + ], + [ + 174, + 218 + ] + ], + [ + [ + 272, + 209 + ], + [ + 284, + 204 + ], + [ + 285, + 183 + ], + [ + 258, + 236 + ], + [ + 274, + 189 + ], + [ + 248, + 239 + ] + ], + [ + [ + 242, + 219 + ], + [ + 242, + 202 + ], + [ + 216, + 205 + ], + [ + 268, + 234 + ], + [ + 217, + 221 + ], + [ + 267, + 249 + ] + ] + ] + }, + { + "task_id": "0004337", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0002_cfg_0006_104122061330#211", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004337/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 324, + 191 + ], + [ + 345, + 210 + ], + [ + 338, + 170 + ], + [ + 310, + 210 + ], + [ + 317, + 151 + ], + [ + 291, + 191 + ] + ], + [ + [ + 284, + 208 + ], + [ + 309, + 218 + ], + [ + 272, + 232 + ], + [ + 296, + 186 + ], + [ + 248, + 221 + ], + [ + 272, + 177 + ] + ], + [ + [ + 282, + 188 + ], + [ + 304, + 204 + ], + [ + 280, + 174 + ], + [ + 283, + 200 + ], + [ + 258, + 157 + ], + [ + 263, + 185 + ] + ], + [ + [ + 303, + 210 + ], + [ + 295, + 209 + ], + [ + 292, + 236 + ], + [ + 313, + 185 + ], + [ + 300, + 236 + ], + [ + 319, + 187 + ] + ] + ] + }, + { + "task_id": "0004338", + "id": "RH20T_cfg5_task_0091_user_0007_scene_0004_cfg_0005_104122063678#249", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004338/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 242 + ], + [ + 380, + 260 + ], + [ + 371, + 228 + ], + [ + 416, + 256 + ], + [ + 383, + 211 + ], + [ + 426, + 238 + ] + ], + [ + [ + 410, + 248 + ], + [ + 420, + 271 + ], + [ + 396, + 266 + ], + [ + 425, + 230 + ], + [ + 387, + 245 + ], + [ + 415, + 209 + ] + ], + [ + [ + 390, + 245 + ], + [ + 396, + 231 + ], + [ + 376, + 263 + ], + [ + 404, + 229 + ], + [ + 371, + 276 + ], + [ + 398, + 242 + ] + ], + [ + [ + 377, + 226 + ], + [ + 377, + 210 + ], + [ + 372, + 245 + ], + [ + 382, + 207 + ], + [ + 372, + 263 + ], + [ + 383, + 224 + ] + ] + ] + }, + { + "task_id": "0004339", + "id": "RH20T_cfg5_task_0091_user_0007_scene_0004_cfg_0005_104122063678#436", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004339/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 331, + 181 + ], + [ + 330, + 205 + ], + [ + 307, + 179 + ], + [ + 356, + 184 + ], + [ + 308, + 156 + ], + [ + 356, + 161 + ] + ], + [ + [ + 318, + 194 + ], + [ + 320, + 178 + ], + [ + 298, + 203 + ], + [ + 338, + 185 + ], + [ + 297, + 217 + ], + [ + 336, + 199 + ] + ], + [ + [ + 341, + 214 + ], + [ + 359, + 226 + ], + [ + 325, + 224 + ], + [ + 357, + 205 + ], + [ + 308, + 211 + ], + [ + 340, + 194 + ] + ], + [ + [ + 317, + 210 + ], + [ + 317, + 234 + ], + [ + 294, + 208 + ], + [ + 340, + 213 + ], + [ + 295, + 185 + ], + [ + 341, + 190 + ] + ] + ] + }, + { + "task_id": "0004341", + "id": "RH20T_cfg4_task_0046_user_0014_scene_0015_cfg_0004_104422070011#107", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004341/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 169 + ], + [ + 300, + 153 + ], + [ + 293, + 196 + ], + [ + 277, + 144 + ], + [ + 279, + 210 + ], + [ + 264, + 159 + ] + ], + [ + [ + 278, + 173 + ], + [ + 270, + 197 + ], + [ + 261, + 153 + ], + [ + 293, + 190 + ], + [ + 269, + 130 + ], + [ + 300, + 168 + ] + ], + [ + [ + 298, + 190 + ], + [ + 303, + 193 + ], + [ + 291, + 160 + ], + [ + 305, + 221 + ], + [ + 285, + 156 + ], + [ + 299, + 221 + ] + ], + [ + [ + 318, + 156 + ], + [ + 301, + 148 + ], + [ + 345, + 145 + ], + [ + 293, + 167 + ], + [ + 359, + 153 + ], + [ + 309, + 174 + ] + ] + ] + }, + { + "task_id": "0004344", + "id": "RH20T_cfg3_task_0032_user_0011_scene_0002_cfg_0003_104422070011#193", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004344/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 396, + 160 + ], + [ + 417, + 151 + ], + [ + 384, + 158 + ], + [ + 408, + 162 + ], + [ + 364, + 167 + ], + [ + 389, + 170 + ] + ], + [ + [ + 392, + 167 + ], + [ + 405, + 157 + ], + [ + 371, + 159 + ], + [ + 412, + 175 + ], + [ + 359, + 168 + ], + [ + 399, + 184 + ] + ], + [ + [ + 390, + 147 + ], + [ + 366, + 153 + ], + [ + 384, + 153 + ], + [ + 397, + 140 + ], + [ + 406, + 147 + ], + [ + 422, + 134 + ] + ], + [ + [ + 395, + 167 + ], + [ + 417, + 172 + ], + [ + 389, + 181 + ], + [ + 402, + 153 + ], + [ + 368, + 176 + ], + [ + 380, + 148 + ] + ] + ] + }, + { + "task_id": "0004345", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104422070011#122", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004345/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 302, + 144 + ], + [ + 297, + 115 + ], + [ + 283, + 136 + ], + [ + 323, + 153 + ], + [ + 288, + 162 + ], + [ + 327, + 181 + ] + ], + [ + [ + 336, + 126 + ], + [ + 348, + 106 + ], + [ + 311, + 136 + ], + [ + 359, + 117 + ], + [ + 301, + 154 + ], + [ + 348, + 135 + ] + ], + [ + [ + 352, + 139 + ], + [ + 324, + 131 + ], + [ + 363, + 111 + ], + [ + 341, + 168 + ], + [ + 393, + 119 + ], + [ + 370, + 178 + ] + ], + [ + [ + 319, + 130 + ], + [ + 304, + 118 + ], + [ + 292, + 126 + ], + [ + 345, + 135 + ], + [ + 307, + 140 + ], + [ + 362, + 148 + ] + ] + ] + }, + { + "task_id": "0004346", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104422070011#182", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004346/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 161 + ], + [ + 387, + 176 + ], + [ + 345, + 174 + ], + [ + 380, + 149 + ], + [ + 321, + 158 + ], + [ + 357, + 135 + ] + ], + [ + [ + 377, + 152 + ], + [ + 395, + 136 + ], + [ + 356, + 163 + ], + [ + 396, + 142 + ], + [ + 341, + 176 + ], + [ + 380, + 156 + ] + ], + [ + [ + 373, + 183 + ], + [ + 363, + 211 + ], + [ + 402, + 193 + ], + [ + 346, + 173 + ], + [ + 412, + 164 + ], + [ + 356, + 145 + ] + ], + [ + [ + 388, + 163 + ], + [ + 392, + 145 + ], + [ + 413, + 154 + ], + [ + 362, + 173 + ], + [ + 409, + 171 + ], + [ + 360, + 189 + ] + ] + ] + }, + { + "task_id": "0004347", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104422070011#222", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004347/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 114 + ], + [ + 315, + 86 + ], + [ + 286, + 113 + ], + [ + 342, + 116 + ], + [ + 284, + 141 + ], + [ + 341, + 144 + ] + ], + [ + [ + 339, + 100 + ], + [ + 328, + 73 + ], + [ + 324, + 96 + ], + [ + 357, + 104 + ], + [ + 335, + 120 + ], + [ + 368, + 129 + ] + ], + [ + [ + 340, + 142 + ], + [ + 353, + 170 + ], + [ + 311, + 156 + ], + [ + 368, + 129 + ], + [ + 298, + 128 + ], + [ + 354, + 101 + ] + ], + [ + [ + 327, + 123 + ], + [ + 342, + 114 + ], + [ + 304, + 136 + ], + [ + 348, + 112 + ], + [ + 291, + 144 + ], + [ + 333, + 120 + ] + ] + ] + }, + { + "task_id": "0004348", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104422070011#265", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004348/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 158 + ], + [ + 321, + 133 + ], + [ + 314, + 151 + ], + [ + 356, + 165 + ], + [ + 326, + 172 + ], + [ + 368, + 188 + ] + ], + [ + [ + 308, + 155 + ], + [ + 327, + 140 + ], + [ + 290, + 170 + ], + [ + 325, + 141 + ], + [ + 273, + 183 + ], + [ + 308, + 154 + ] + ], + [ + [ + 316, + 189 + ], + [ + 348, + 177 + ], + [ + 306, + 182 + ], + [ + 325, + 196 + ], + [ + 273, + 195 + ], + [ + 296, + 207 + ] + ], + [ + [ + 320, + 171 + ], + [ + 341, + 199 + ], + [ + 297, + 187 + ], + [ + 345, + 154 + ], + [ + 277, + 160 + ], + [ + 323, + 124 + ] + ] + ] + }, + { + "task_id": "0004351", + "id": "RH20T_cfg1_task_0065_user_0015_scene_0001_cfg_0001_819112070044#319", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004351/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 242, + 288 + ], + [ + 226, + 303 + ], + [ + 245, + 306 + ], + [ + 239, + 269 + ], + [ + 259, + 293 + ], + [ + 254, + 256 + ] + ], + [ + [ + 228, + 275 + ], + [ + 203, + 273 + ], + [ + 223, + 296 + ], + [ + 232, + 256 + ], + [ + 249, + 297 + ], + [ + 255, + 257 + ] + ], + [ + [ + 263, + 237 + ], + [ + 242, + 225 + ], + [ + 254, + 254 + ], + [ + 271, + 220 + ], + [ + 275, + 266 + ], + [ + 291, + 231 + ] + ], + [ + [ + 254, + 263 + ], + [ + 254, + 247 + ], + [ + 258, + 245 + ], + [ + 250, + 282 + ], + [ + 258, + 259 + ], + [ + 250, + 296 + ] + ] + ] + }, + { + "task_id": "0004352", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0005_cfg_0004_104122062823#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004352/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 290, + 0 + ], + [ + 301, + 0 + ], + [ + 267, + 0 + ], + [ + 316, + 0 + ], + [ + 258, + 30 + ], + [ + 305, + 30 + ] + ], + [ + [ + 345, + 0 + ], + [ + 347, + 0 + ], + [ + 383, + 0 + ], + [ + 309, + 0 + ], + [ + 382, + 0 + ], + [ + 304, + 0 + ] + ], + [ + [ + 305, + 9 + ], + [ + 310, + 0 + ], + [ + 330, + 27 + ], + [ + 282, + 0 + ], + [ + 324, + 57 + ], + [ + 279, + 22 + ] + ], + [ + [ + 297, + 16 + ], + [ + 329, + 0 + ], + [ + 294, + 0 + ], + [ + 301, + 42 + ], + [ + 264, + 9 + ], + [ + 269, + 58 + ] + ] + ] + }, + { + "task_id": "0004356", + "id": "RH20T_cfg1_task_0072_user_0010_scene_0010_cfg_0001_750612070853#284", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drop coins into a piggy bank'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004356/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 119, + 178 + ], + [ + 98, + 210 + ], + [ + 97, + 176 + ], + [ + 144, + 181 + ], + [ + 116, + 148 + ], + [ + 164, + 151 + ] + ], + [ + [ + 130, + 191 + ], + [ + 158, + 164 + ], + [ + 101, + 156 + ], + [ + 157, + 223 + ], + [ + 65, + 186 + ], + [ + 128, + 255 + ] + ], + [ + [ + 112, + 177 + ], + [ + 134, + 156 + ], + [ + 106, + 206 + ], + [ + 118, + 144 + ], + [ + 82, + 231 + ], + [ + 92, + 166 + ] + ], + [ + [ + 116, + 180 + ], + [ + 101, + 145 + ], + [ + 81, + 178 + ], + [ + 152, + 183 + ], + [ + 96, + 209 + ], + [ + 164, + 215 + ] + ] + ] + }, + { + "task_id": "0004357", + "id": "RH20T_cfg1_task_0072_user_0010_scene_0010_cfg_0001_750612070853#515", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drop coins into a piggy bank'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004357/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 138, + 178 + ], + [ + 165, + 162 + ], + [ + 105, + 152 + ], + [ + 168, + 201 + ], + [ + 80, + 168 + ], + [ + 143, + 214 + ] + ], + [ + [ + 133, + 189 + ], + [ + 142, + 210 + ], + [ + 170, + 195 + ], + [ + 91, + 183 + ], + [ + 160, + 178 + ], + [ + 84, + 164 + ] + ], + [ + [ + 124, + 201 + ], + [ + 151, + 178 + ], + [ + 100, + 195 + ], + [ + 150, + 208 + ], + [ + 70, + 220 + ], + [ + 120, + 236 + ] + ], + [ + [ + 126, + 147 + ], + [ + 164, + 131 + ], + [ + 120, + 155 + ], + [ + 131, + 140 + ], + [ + 76, + 173 + ], + [ + 94, + 155 + ] + ] + ] + }, + { + "task_id": "0004358", + "id": "RH20T_cfg1_task_0072_user_0010_scene_0010_cfg_0001_750612070853#570", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drop coins into a piggy bank'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004358/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 82, + 185 + ], + [ + 116, + 210 + ], + [ + 116, + 160 + ], + [ + 48, + 209 + ], + [ + 80, + 129 + ], + [ + 8, + 182 + ] + ], + [ + [ + 141, + 159 + ], + [ + 163, + 137 + ], + [ + 151, + 130 + ], + [ + 131, + 184 + ], + [ + 125, + 154 + ], + [ + 106, + 210 + ] + ], + [ + [ + 125, + 164 + ], + [ + 154, + 144 + ], + [ + 88, + 138 + ], + [ + 158, + 187 + ], + [ + 60, + 158 + ], + [ + 130, + 204 + ] + ], + [ + [ + 168, + 161 + ], + [ + 178, + 200 + ], + [ + 174, + 159 + ], + [ + 161, + 164 + ], + [ + 165, + 122 + ], + [ + 150, + 120 + ] + ] + ] + }, + { + "task_id": "0004362", + "id": "RH20T_cfg4_task_0047_user_0014_scene_0005_cfg_0004_104122062295#254", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004362/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 294, + 282 + ], + [ + 264, + 324 + ], + [ + 322, + 289 + ], + [ + 260, + 273 + ], + [ + 350, + 249 + ], + [ + 292, + 226 + ] + ], + [ + [ + 256, + 295 + ], + [ + 265, + 247 + ], + [ + 196, + 286 + ], + [ + 314, + 303 + ], + [ + 190, + 329 + ], + [ + 302, + 344 + ] + ], + [ + [ + 288, + 360 + ], + [ + 236, + 360 + ], + [ + 330, + 360 + ], + [ + 254, + 352 + ], + [ + 375, + 360 + ], + [ + 297, + 322 + ] + ], + [ + [ + 204, + 332 + ], + [ + 231, + 272 + ], + [ + 263, + 354 + ], + [ + 148, + 310 + ], + [ + 236, + 360 + ], + [ + 115, + 360 + ] + ] + ] + }, + { + "task_id": "0004363", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_105422061350#608", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004363/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 365, + 0 + ], + [ + 392, + 0 + ], + [ + 354, + 0 + ], + [ + 377, + 0 + ], + [ + 326, + 0 + ], + [ + 347, + 0 + ] + ], + [ + [ + 356, + 0 + ], + [ + 369, + 0 + ], + [ + 327, + 0 + ], + [ + 388, + 0 + ], + [ + 316, + 0 + ], + [ + 374, + 0 + ] + ], + [ + [ + 363, + 0 + ], + [ + 335, + 0 + ], + [ + 364, + 0 + ], + [ + 361, + 0 + ], + [ + 395, + 0 + ], + [ + 388, + 0 + ] + ], + [ + [ + 331, + 0 + ], + [ + 341, + 0 + ], + [ + 346, + 0 + ], + [ + 314, + 0 + ], + [ + 336, + 0 + ], + [ + 306, + 0 + ] + ] + ] + }, + { + "task_id": "0004364", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_105422061350#824", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004364/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 362, + 9 + ], + [ + 355, + 0 + ], + [ + 369, + 0 + ], + [ + 356, + 29 + ], + [ + 375, + 14 + ], + [ + 363, + 53 + ] + ], + [ + [ + 375, + 40 + ], + [ + 405, + 37 + ], + [ + 371, + 52 + ], + [ + 378, + 29 + ], + [ + 340, + 56 + ], + [ + 350, + 33 + ] + ], + [ + [ + 356, + 0 + ], + [ + 362, + 0 + ], + [ + 356, + 0 + ], + [ + 355, + 31 + ], + [ + 351, + 0 + ], + [ + 350, + 42 + ] + ], + [ + [ + 329, + 0 + ], + [ + 302, + 0 + ], + [ + 342, + 0 + ], + [ + 315, + 10 + ], + [ + 369, + 0 + ], + [ + 344, + 16 + ] + ] + ] + }, + { + "task_id": "0004365", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_105422061350#1474", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004365/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 255, + 55 + ], + [ + 258, + 69 + ], + [ + 229, + 54 + ], + [ + 280, + 56 + ], + [ + 227, + 41 + ], + [ + 276, + 43 + ] + ], + [ + [ + 264, + 51 + ], + [ + 276, + 75 + ], + [ + 255, + 51 + ], + [ + 273, + 50 + ], + [ + 243, + 27 + ], + [ + 260, + 23 + ] + ], + [ + [ + 240, + 33 + ], + [ + 235, + 13 + ], + [ + 266, + 35 + ], + [ + 214, + 30 + ], + [ + 269, + 52 + ], + [ + 220, + 48 + ] + ], + [ + [ + 249, + 20 + ], + [ + 224, + 13 + ], + [ + 242, + 10 + ], + [ + 256, + 29 + ], + [ + 270, + 18 + ], + [ + 280, + 36 + ] + ] + ] + }, + { + "task_id": "0004366", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_105422061350#1811", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004366/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 370, + 22 + ], + [ + 359, + 2 + ], + [ + 373, + 6 + ], + [ + 366, + 38 + ], + [ + 383, + 24 + ], + [ + 377, + 56 + ] + ], + [ + [ + 387, + 19 + ], + [ + 399, + 9 + ], + [ + 364, + 17 + ], + [ + 410, + 22 + ], + [ + 355, + 27 + ], + [ + 399, + 32 + ] + ], + [ + [ + 374, + 31 + ], + [ + 353, + 39 + ], + [ + 361, + 28 + ], + [ + 387, + 34 + ], + [ + 381, + 21 + ], + [ + 409, + 26 + ] + ], + [ + [ + 348, + 19 + ], + [ + 353, + 39 + ], + [ + 344, + 35 + ], + [ + 352, + 1 + ], + [ + 340, + 16 + ], + [ + 347, + 0 + ] + ] + ] + }, + { + "task_id": "0004367", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_105422061350#2071", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004367/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 304, + 0 + ], + [ + 284, + 0 + ], + [ + 324, + 0 + ], + [ + 282, + 0 + ], + [ + 345, + 0 + ], + [ + 304, + 0 + ] + ], + [ + [ + 315, + 0 + ], + [ + 320, + 0 + ], + [ + 338, + 0 + ], + [ + 290, + 0 + ], + [ + 333, + 0 + ], + [ + 287, + 0 + ] + ], + [ + [ + 328, + 0 + ], + [ + 357, + 0 + ], + [ + 339, + 0 + ], + [ + 318, + 0 + ], + [ + 310, + 0 + ], + [ + 290, + 0 + ] + ], + [ + [ + 297, + 0 + ], + [ + 283, + 0 + ], + [ + 283, + 0 + ], + [ + 312, + 0 + ], + [ + 295, + 0 + ], + [ + 325, + 0 + ] + ] + ] + }, + { + "task_id": "0004370", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522062547#955", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004370/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 430, + 184 + ], + [ + 436, + 173 + ], + [ + 432, + 210 + ], + [ + 428, + 160 + ], + [ + 426, + 218 + ], + [ + 423, + 171 + ] + ], + [ + [ + 417, + 202 + ], + [ + 442, + 202 + ], + [ + 425, + 216 + ], + [ + 410, + 190 + ], + [ + 398, + 215 + ], + [ + 385, + 190 + ] + ], + [ + [ + 421, + 179 + ], + [ + 447, + 186 + ], + [ + 428, + 153 + ], + [ + 413, + 204 + ], + [ + 402, + 146 + ], + [ + 387, + 197 + ] + ], + [ + [ + 439, + 156 + ], + [ + 422, + 142 + ], + [ + 460, + 137 + ], + [ + 418, + 174 + ], + [ + 480, + 151 + ], + [ + 435, + 190 + ] + ] + ] + }, + { + "task_id": "0004373", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522062547#1757", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004373/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 449, + 114 + ], + [ + 442, + 133 + ], + [ + 470, + 107 + ], + [ + 431, + 121 + ], + [ + 476, + 89 + ], + [ + 437, + 104 + ] + ], + [ + [ + 466, + 147 + ], + [ + 491, + 150 + ], + [ + 474, + 153 + ], + [ + 460, + 141 + ], + [ + 448, + 149 + ], + [ + 437, + 137 + ] + ], + [ + [ + 453, + 130 + ], + [ + 460, + 117 + ], + [ + 453, + 153 + ], + [ + 453, + 109 + ], + [ + 447, + 164 + ], + [ + 447, + 122 + ] + ], + [ + [ + 466, + 151 + ], + [ + 485, + 156 + ], + [ + 449, + 157 + ], + [ + 485, + 146 + ], + [ + 431, + 152 + ], + [ + 466, + 140 + ] + ] + ] + }, + { + "task_id": "0004374", + "id": "RH20T_cfg6_task_0126_user_0014_scene_0006_cfg_0006_104122060811#114", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord to the socket'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004374/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 333, + 107 + ], + [ + 328, + 108 + ], + [ + 322, + 116 + ], + [ + 343, + 98 + ], + [ + 327, + 116 + ], + [ + 349, + 97 + ] + ], + [ + [ + 336, + 94 + ], + [ + 329, + 88 + ], + [ + 323, + 93 + ], + [ + 348, + 95 + ], + [ + 329, + 99 + ], + [ + 355, + 101 + ] + ], + [ + [ + 343, + 111 + ], + [ + 342, + 104 + ], + [ + 357, + 106 + ], + [ + 330, + 116 + ], + [ + 359, + 113 + ], + [ + 332, + 123 + ] + ], + [ + [ + 333, + 108 + ], + [ + 332, + 112 + ], + [ + 333, + 94 + ], + [ + 333, + 123 + ], + [ + 334, + 89 + ], + [ + 334, + 120 + ] + ] + ] + }, + { + "task_id": "0004375", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122064161#129", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004375/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 262 + ], + [ + 376, + 242 + ], + [ + 384, + 277 + ], + [ + 341, + 247 + ], + [ + 371, + 298 + ], + [ + 327, + 267 + ] + ], + [ + [ + 358, + 229 + ], + [ + 365, + 208 + ], + [ + 362, + 245 + ], + [ + 355, + 214 + ], + [ + 356, + 265 + ], + [ + 350, + 233 + ] + ], + [ + [ + 352, + 223 + ], + [ + 338, + 207 + ], + [ + 363, + 230 + ], + [ + 341, + 215 + ], + [ + 378, + 246 + ], + [ + 356, + 232 + ] + ], + [ + [ + 341, + 220 + ], + [ + 315, + 217 + ], + [ + 347, + 203 + ], + [ + 335, + 240 + ], + [ + 373, + 205 + ], + [ + 363, + 243 + ] + ] + ] + }, + { + "task_id": "0004377", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0001_cfg_0007_104122064161#265", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004377/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 400, + 233 + ], + [ + 375, + 233 + ], + [ + 399, + 208 + ], + [ + 401, + 259 + ], + [ + 425, + 209 + ], + [ + 428, + 259 + ] + ], + [ + [ + 383, + 224 + ], + [ + 388, + 205 + ], + [ + 386, + 241 + ], + [ + 381, + 209 + ], + [ + 381, + 260 + ], + [ + 376, + 228 + ] + ], + [ + [ + 390, + 204 + ], + [ + 375, + 220 + ], + [ + 406, + 204 + ], + [ + 376, + 205 + ], + [ + 421, + 188 + ], + [ + 390, + 191 + ] + ], + [ + [ + 394, + 216 + ], + [ + 377, + 228 + ], + [ + 391, + 235 + ], + [ + 397, + 196 + ], + [ + 407, + 224 + ], + [ + 414, + 185 + ] + ] + ] + }, + { + "task_id": "0004383", + "id": "RH20T_cfg5_task_0020_user_0010_scene_0007_cfg_0005_104122061850#271", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004383/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 115 + ], + [ + 303, + 111 + ], + [ + 328, + 94 + ], + [ + 324, + 134 + ], + [ + 352, + 98 + ], + [ + 346, + 138 + ] + ], + [ + [ + 358, + 101 + ], + [ + 364, + 90 + ], + [ + 347, + 120 + ], + [ + 369, + 83 + ], + [ + 343, + 129 + ], + [ + 364, + 94 + ] + ], + [ + [ + 336, + 91 + ], + [ + 325, + 107 + ], + [ + 313, + 82 + ], + [ + 357, + 100 + ], + [ + 325, + 65 + ], + [ + 370, + 84 + ] + ], + [ + [ + 379, + 81 + ], + [ + 391, + 81 + ], + [ + 362, + 68 + ], + [ + 396, + 96 + ], + [ + 348, + 68 + ], + [ + 384, + 97 + ] + ] + ] + }, + { + "task_id": "0004384", + "id": "RH20T_cfg5_task_0042_user_0007_scene_0008_cfg_0005_037522062165#272", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004384/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 199, + 253 + ], + [ + 170, + 265 + ], + [ + 193, + 239 + ], + [ + 205, + 268 + ], + [ + 220, + 228 + ], + [ + 234, + 256 + ] + ], + [ + [ + 211, + 242 + ], + [ + 215, + 228 + ], + [ + 225, + 261 + ], + [ + 197, + 223 + ], + [ + 222, + 278 + ], + [ + 193, + 238 + ] + ], + [ + [ + 199, + 262 + ], + [ + 196, + 247 + ], + [ + 186, + 241 + ], + [ + 212, + 283 + ], + [ + 189, + 255 + ], + [ + 214, + 296 + ] + ], + [ + [ + 217, + 255 + ], + [ + 216, + 227 + ], + [ + 188, + 256 + ], + [ + 245, + 254 + ], + [ + 189, + 284 + ], + [ + 246, + 281 + ] + ] + ] + }, + { + "task_id": "0004388", + "id": "RH20T_cfg1_task_0038_user_0014_scene_0004_cfg_0001_819112070044#107", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004388/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 321, + 208 + ], + [ + 341, + 220 + ], + [ + 320, + 226 + ], + [ + 323, + 189 + ], + [ + 301, + 215 + ], + [ + 302, + 178 + ] + ], + [ + [ + 355, + 226 + ], + [ + 376, + 237 + ], + [ + 361, + 241 + ], + [ + 349, + 210 + ], + [ + 342, + 231 + ], + [ + 329, + 201 + ] + ], + [ + [ + 340, + 219 + ], + [ + 327, + 202 + ], + [ + 361, + 218 + ], + [ + 321, + 219 + ], + [ + 374, + 234 + ], + [ + 334, + 234 + ] + ], + [ + [ + 337, + 219 + ], + [ + 339, + 200 + ], + [ + 352, + 206 + ], + [ + 321, + 232 + ], + [ + 349, + 224 + ], + [ + 320, + 250 + ] + ] + ] + }, + { + "task_id": "0004393", + "id": "RH20T_cfg5_task_0038_user_0007_scene_0009_cfg_0005_036422060215#281", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004393/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 313, + 236 + ], + [ + 303, + 225 + ], + [ + 287, + 242 + ], + [ + 338, + 231 + ], + [ + 296, + 255 + ], + [ + 350, + 243 + ] + ], + [ + [ + 311, + 210 + ], + [ + 318, + 192 + ], + [ + 333, + 206 + ], + [ + 286, + 215 + ], + [ + 326, + 222 + ], + [ + 280, + 231 + ] + ], + [ + [ + 307, + 219 + ], + [ + 306, + 206 + ], + [ + 321, + 237 + ], + [ + 293, + 199 + ], + [ + 323, + 252 + ], + [ + 294, + 212 + ] + ], + [ + [ + 330, + 203 + ], + [ + 355, + 215 + ], + [ + 337, + 190 + ], + [ + 323, + 214 + ], + [ + 310, + 178 + ], + [ + 299, + 203 + ] + ] + ] + }, + { + "task_id": "0004394", + "id": "RH20T_cfg7_task_0042_user_0020_scene_0003_cfg_0007_104122060811#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004394/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 397, + 127 + ], + [ + 393, + 119 + ], + [ + 387, + 122 + ], + [ + 408, + 133 + ], + [ + 392, + 129 + ], + [ + 412, + 140 + ] + ], + [ + [ + 372, + 146 + ], + [ + 367, + 153 + ], + [ + 367, + 134 + ], + [ + 376, + 158 + ], + [ + 371, + 128 + ], + [ + 380, + 152 + ] + ], + [ + [ + 382, + 130 + ], + [ + 388, + 123 + ], + [ + 394, + 128 + ], + [ + 370, + 132 + ], + [ + 389, + 135 + ], + [ + 365, + 139 + ] + ], + [ + [ + 364, + 132 + ], + [ + 361, + 120 + ], + [ + 363, + 129 + ], + [ + 366, + 136 + ], + [ + 366, + 142 + ], + [ + 369, + 148 + ] + ] + ] + }, + { + "task_id": "0004398", + "id": "RH20T_cfg1_task_0016_user_0014_scene_0001_cfg_0001_038522063145#262", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004398/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 333, + 73 + ], + [ + 333, + 57 + ], + [ + 351, + 67 + ], + [ + 315, + 79 + ], + [ + 350, + 82 + ], + [ + 316, + 94 + ] + ], + [ + [ + 324, + 77 + ], + [ + 320, + 93 + ], + [ + 342, + 82 + ], + [ + 305, + 73 + ], + [ + 346, + 66 + ], + [ + 308, + 56 + ] + ], + [ + [ + 319, + 93 + ], + [ + 326, + 91 + ], + [ + 328, + 110 + ], + [ + 310, + 75 + ], + [ + 321, + 113 + ], + [ + 302, + 76 + ] + ], + [ + [ + 318, + 75 + ], + [ + 318, + 56 + ], + [ + 300, + 76 + ], + [ + 336, + 75 + ], + [ + 300, + 95 + ], + [ + 336, + 95 + ] + ] + ] + }, + { + "task_id": "0004406", + "id": "RH20T_cfg7_task_0047_user_0020_scene_0004_cfg_0007_104122064161#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004406/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 403, + 145 + ], + [ + 403, + 158 + ], + [ + 410, + 167 + ], + [ + 395, + 123 + ], + [ + 410, + 155 + ], + [ + 395, + 112 + ] + ], + [ + [ + 398, + 160 + ], + [ + 378, + 144 + ], + [ + 414, + 140 + ], + [ + 381, + 179 + ], + [ + 435, + 155 + ], + [ + 401, + 196 + ] + ], + [ + [ + 430, + 153 + ], + [ + 456, + 146 + ], + [ + 424, + 129 + ], + [ + 437, + 177 + ], + [ + 399, + 136 + ], + [ + 412, + 184 + ] + ], + [ + [ + 395, + 132 + ], + [ + 369, + 142 + ], + [ + 399, + 144 + ], + [ + 391, + 118 + ], + [ + 424, + 134 + ], + [ + 419, + 107 + ] + ] + ] + }, + { + "task_id": "0004412", + "id": "RH20T_cfg1_task_0001_user_0014_scene_0006_cfg_0001_038522062547#237", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004412/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 394, + 64 + ], + [ + 377, + 67 + ], + [ + 410, + 60 + ], + [ + 376, + 68 + ], + [ + 429, + 57 + ], + [ + 393, + 65 + ] + ], + [ + [ + 409, + 24 + ], + [ + 389, + 30 + ], + [ + 407, + 0 + ], + [ + 410, + 50 + ], + [ + 426, + 0 + ], + [ + 427, + 43 + ] + ], + [ + [ + 397, + 68 + ], + [ + 371, + 71 + ], + [ + 401, + 94 + ], + [ + 394, + 42 + ], + [ + 427, + 90 + ], + [ + 420, + 39 + ] + ], + [ + [ + 394, + 63 + ], + [ + 396, + 46 + ], + [ + 400, + 84 + ], + [ + 389, + 43 + ], + [ + 398, + 98 + ], + [ + 388, + 59 + ] + ] + ] + }, + { + "task_id": "0004415", + "id": "RH20T_cfg1_task_0038_user_0016_scene_0009_cfg_0001_750612070853#122", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004415/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 317, + 149 + ], + [ + 318, + 104 + ], + [ + 282, + 133 + ], + [ + 356, + 167 + ], + [ + 283, + 172 + ], + [ + 353, + 208 + ] + ], + [ + [ + 312, + 193 + ], + [ + 324, + 241 + ], + [ + 360, + 181 + ], + [ + 265, + 204 + ], + [ + 348, + 132 + ], + [ + 252, + 157 + ] + ], + [ + [ + 310, + 204 + ], + [ + 301, + 248 + ], + [ + 274, + 216 + ], + [ + 351, + 191 + ], + [ + 283, + 178 + ], + [ + 357, + 150 + ] + ], + [ + [ + 310, + 133 + ], + [ + 347, + 126 + ], + [ + 277, + 107 + ], + [ + 347, + 162 + ], + [ + 234, + 113 + ], + [ + 305, + 175 + ] + ] + ] + }, + { + "task_id": "0004418", + "id": "RH20T_cfg5_task_0047_user_0007_scene_0009_cfg_0005_104122063678#300", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004418/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 367, + 120 + ], + [ + 374, + 104 + ], + [ + 361, + 136 + ], + [ + 372, + 105 + ], + [ + 355, + 150 + ], + [ + 366, + 119 + ] + ], + [ + [ + 359, + 89 + ], + [ + 382, + 92 + ], + [ + 356, + 98 + ], + [ + 363, + 81 + ], + [ + 332, + 94 + ], + [ + 341, + 78 + ] + ], + [ + [ + 359, + 129 + ], + [ + 382, + 133 + ], + [ + 354, + 149 + ], + [ + 365, + 108 + ], + [ + 331, + 145 + ], + [ + 341, + 104 + ] + ], + [ + [ + 368, + 134 + ], + [ + 362, + 155 + ], + [ + 347, + 127 + ], + [ + 389, + 140 + ], + [ + 353, + 107 + ], + [ + 395, + 120 + ] + ] + ] + }, + { + "task_id": "0004419", + "id": "RH20T_cfg4_task_0111_user_0007_scene_0001_cfg_0004_036422060909#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object with the corresponding shape into the corresponding hole'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004419/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 310, + 77 + ], + [ + 311, + 60 + ], + [ + 323, + 66 + ], + [ + 295, + 89 + ], + [ + 321, + 82 + ], + [ + 294, + 106 + ] + ], + [ + [ + 313, + 82 + ], + [ + 330, + 97 + ], + [ + 310, + 100 + ], + [ + 317, + 64 + ], + [ + 294, + 86 + ], + [ + 300, + 49 + ] + ], + [ + [ + 314, + 63 + ], + [ + 331, + 71 + ], + [ + 302, + 80 + ], + [ + 326, + 46 + ], + [ + 284, + 73 + ], + [ + 307, + 38 + ] + ], + [ + [ + 319, + 86 + ], + [ + 332, + 95 + ], + [ + 301, + 89 + ], + [ + 338, + 83 + ], + [ + 286, + 79 + ], + [ + 324, + 73 + ] + ] + ] + }, + { + "task_id": "0004421", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0006_cfg_0004_104122062823#326", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004421/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 230, + 112 + ], + [ + 194, + 127 + ], + [ + 248, + 136 + ], + [ + 213, + 89 + ], + [ + 283, + 121 + ], + [ + 246, + 76 + ] + ], + [ + [ + 225, + 149 + ], + [ + 236, + 121 + ], + [ + 191, + 138 + ], + [ + 258, + 159 + ], + [ + 177, + 167 + ], + [ + 247, + 190 + ] + ], + [ + [ + 218, + 128 + ], + [ + 187, + 106 + ], + [ + 224, + 110 + ], + [ + 212, + 149 + ], + [ + 251, + 130 + ], + [ + 243, + 170 + ] + ], + [ + [ + 194, + 107 + ], + [ + 180, + 88 + ], + [ + 158, + 121 + ], + [ + 228, + 94 + ], + [ + 172, + 143 + ], + [ + 245, + 113 + ] + ] + ] + }, + { + "task_id": "0004423", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0006_cfg_0004_104122062823#1144", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004423/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 30 + ], + [ + 357, + 0 + ], + [ + 345, + 25 + ], + [ + 400, + 36 + ], + [ + 359, + 55 + ], + [ + 411, + 69 + ] + ], + [ + [ + 372, + 82 + ], + [ + 400, + 100 + ], + [ + 388, + 87 + ], + [ + 359, + 77 + ], + [ + 357, + 67 + ], + [ + 331, + 60 + ] + ], + [ + [ + 364, + 66 + ], + [ + 367, + 104 + ], + [ + 348, + 53 + ], + [ + 378, + 76 + ], + [ + 344, + 13 + ], + [ + 375, + 41 + ] + ], + [ + [ + 357, + 47 + ], + [ + 356, + 10 + ], + [ + 377, + 30 + ], + [ + 334, + 64 + ], + [ + 377, + 62 + ], + [ + 336, + 98 + ] + ] + ] + }, + { + "task_id": "0004425", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_104122063678#258", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004425/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 386, + 242 + ], + [ + 406, + 230 + ], + [ + 377, + 240 + ], + [ + 394, + 243 + ], + [ + 357, + 252 + ], + [ + 375, + 254 + ] + ], + [ + [ + 361, + 246 + ], + [ + 362, + 255 + ], + [ + 365, + 270 + ], + [ + 357, + 223 + ], + [ + 363, + 261 + ], + [ + 355, + 213 + ] + ], + [ + [ + 372, + 263 + ], + [ + 373, + 251 + ], + [ + 381, + 285 + ], + [ + 364, + 242 + ], + [ + 379, + 296 + ], + [ + 363, + 254 + ] + ], + [ + [ + 399, + 265 + ], + [ + 412, + 271 + ], + [ + 422, + 265 + ], + [ + 378, + 265 + ], + [ + 408, + 257 + ], + [ + 364, + 258 + ] + ] + ] + }, + { + "task_id": "0004428", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_104122063678#854", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004428/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 320, + 201 + ], + [ + 310, + 209 + ], + [ + 340, + 197 + ], + [ + 302, + 205 + ], + [ + 349, + 190 + ], + [ + 311, + 197 + ] + ], + [ + [ + 330, + 195 + ], + [ + 329, + 180 + ], + [ + 330, + 212 + ], + [ + 330, + 180 + ], + [ + 330, + 226 + ], + [ + 331, + 194 + ] + ], + [ + [ + 341, + 200 + ], + [ + 350, + 198 + ], + [ + 342, + 179 + ], + [ + 340, + 221 + ], + [ + 334, + 182 + ], + [ + 332, + 222 + ] + ], + [ + [ + 325, + 216 + ], + [ + 337, + 204 + ], + [ + 321, + 230 + ], + [ + 328, + 202 + ], + [ + 309, + 243 + ], + [ + 316, + 216 + ] + ] + ] + }, + { + "task_id": "0004430", + "id": "RH20T_cfg2_task_0073_user_0014_scene_0003_cfg_0002_036422060215#108", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004430/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 264, + 206 + ], + [ + 245, + 205 + ], + [ + 258, + 223 + ], + [ + 270, + 190 + ], + [ + 279, + 225 + ], + [ + 290, + 191 + ] + ], + [ + [ + 275, + 213 + ], + [ + 262, + 201 + ], + [ + 259, + 211 + ], + [ + 290, + 215 + ], + [ + 273, + 224 + ], + [ + 303, + 227 + ] + ], + [ + [ + 283, + 206 + ], + [ + 284, + 192 + ], + [ + 301, + 200 + ], + [ + 263, + 213 + ], + [ + 300, + 213 + ], + [ + 263, + 226 + ] + ], + [ + [ + 292, + 220 + ], + [ + 283, + 206 + ], + [ + 274, + 227 + ], + [ + 310, + 213 + ], + [ + 284, + 241 + ], + [ + 321, + 226 + ] + ] + ] + }, + { + "task_id": "0004431", + "id": "RH20T_cfg2_task_0073_user_0014_scene_0003_cfg_0002_036422060215#209", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004431/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 147 + ], + [ + 307, + 159 + ], + [ + 323, + 161 + ], + [ + 288, + 132 + ], + [ + 322, + 149 + ], + [ + 288, + 122 + ] + ], + [ + [ + 311, + 168 + ], + [ + 310, + 148 + ], + [ + 335, + 167 + ], + [ + 288, + 170 + ], + [ + 336, + 186 + ], + [ + 290, + 189 + ] + ], + [ + [ + 336, + 178 + ], + [ + 348, + 159 + ], + [ + 352, + 192 + ], + [ + 319, + 164 + ], + [ + 340, + 211 + ], + [ + 308, + 183 + ] + ], + [ + [ + 318, + 156 + ], + [ + 308, + 148 + ], + [ + 328, + 135 + ], + [ + 309, + 177 + ], + [ + 340, + 143 + ], + [ + 320, + 186 + ] + ] + ] + }, + { + "task_id": "0004432", + "id": "RH20T_cfg5_task_0007_user_0010_scene_0007_cfg_0005_036422060215#252", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004432/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 172 + ], + [ + 364, + 182 + ], + [ + 339, + 166 + ], + [ + 355, + 178 + ], + [ + 323, + 157 + ], + [ + 340, + 169 + ] + ], + [ + [ + 331, + 179 + ], + [ + 326, + 175 + ], + [ + 320, + 162 + ], + [ + 343, + 196 + ], + [ + 325, + 166 + ], + [ + 348, + 201 + ] + ], + [ + [ + 333, + 171 + ], + [ + 337, + 155 + ], + [ + 348, + 166 + ], + [ + 318, + 177 + ], + [ + 345, + 181 + ], + [ + 314, + 193 + ] + ], + [ + [ + 326, + 166 + ], + [ + 309, + 161 + ], + [ + 337, + 149 + ], + [ + 315, + 183 + ], + [ + 354, + 154 + ], + [ + 332, + 189 + ] + ] + ] + }, + { + "task_id": "0004433", + "id": "RH20T_cfg6_task_0039_user_0014_scene_0002_cfg_0006_104122061330#129", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004433/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 278, + 162 + ], + [ + 277, + 181 + ], + [ + 289, + 144 + ], + [ + 266, + 181 + ], + [ + 290, + 124 + ], + [ + 267, + 160 + ] + ], + [ + [ + 261, + 136 + ], + [ + 284, + 130 + ], + [ + 270, + 114 + ], + [ + 253, + 160 + ], + [ + 249, + 121 + ], + [ + 232, + 165 + ] + ], + [ + [ + 241, + 150 + ], + [ + 236, + 159 + ], + [ + 219, + 167 + ], + [ + 264, + 133 + ], + [ + 225, + 159 + ], + [ + 268, + 127 + ] + ], + [ + [ + 274, + 154 + ], + [ + 267, + 133 + ], + [ + 260, + 175 + ], + [ + 286, + 133 + ], + [ + 267, + 193 + ], + [ + 291, + 152 + ] + ] + ] + }, + { + "task_id": "0004434", + "id": "RH20T_cfg4_task_0068_user_0015_scene_0006_cfg_0004_104122062823#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button horizontally with obstacles'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004434/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 529, + 135 + ], + [ + 527, + 101 + ], + [ + 569, + 124 + ], + [ + 489, + 145 + ], + [ + 568, + 155 + ], + [ + 492, + 175 + ] + ], + [ + [ + 495, + 109 + ], + [ + 510, + 122 + ], + [ + 532, + 88 + ], + [ + 460, + 129 + ], + [ + 516, + 78 + ], + [ + 450, + 117 + ] + ], + [ + [ + 501, + 158 + ], + [ + 488, + 134 + ], + [ + 468, + 146 + ], + [ + 532, + 170 + ], + [ + 479, + 173 + ], + [ + 547, + 197 + ] + ], + [ + [ + 536, + 169 + ], + [ + 521, + 150 + ], + [ + 535, + 137 + ], + [ + 537, + 199 + ], + [ + 553, + 158 + ], + [ + 553, + 223 + ] + ] + ] + }, + { + "task_id": "0004436", + "id": "RH20T_cfg4_task_0103_user_0007_scene_0003_cfg_0004_038522062288#75", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004436/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 239, + 183 + ], + [ + 225, + 173 + ], + [ + 260, + 185 + ], + [ + 216, + 181 + ], + [ + 275, + 196 + ], + [ + 230, + 193 + ] + ], + [ + [ + 254, + 184 + ], + [ + 224, + 191 + ], + [ + 259, + 189 + ], + [ + 247, + 178 + ], + [ + 287, + 183 + ], + [ + 279, + 171 + ] + ], + [ + [ + 230, + 157 + ], + [ + 231, + 184 + ], + [ + 201, + 159 + ], + [ + 257, + 155 + ], + [ + 199, + 130 + ], + [ + 256, + 128 + ] + ], + [ + [ + 246, + 176 + ], + [ + 243, + 159 + ], + [ + 224, + 163 + ], + [ + 269, + 190 + ], + [ + 228, + 179 + ], + [ + 271, + 205 + ] + ] + ] + }, + { + "task_id": "0004437", + "id": "RH20T_cfg4_task_0103_user_0007_scene_0003_cfg_0004_038522062288#109", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004437/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 219, + 154 + ], + [ + 234, + 174 + ], + [ + 199, + 156 + ], + [ + 237, + 152 + ], + [ + 185, + 135 + ], + [ + 224, + 133 + ] + ], + [ + [ + 249, + 150 + ], + [ + 275, + 155 + ], + [ + 241, + 176 + ], + [ + 258, + 125 + ], + [ + 216, + 170 + ], + [ + 233, + 121 + ] + ], + [ + [ + 230, + 154 + ], + [ + 226, + 137 + ], + [ + 208, + 142 + ], + [ + 252, + 166 + ], + [ + 213, + 158 + ], + [ + 255, + 181 + ] + ], + [ + [ + 231, + 140 + ], + [ + 236, + 167 + ], + [ + 249, + 135 + ], + [ + 214, + 144 + ], + [ + 244, + 108 + ], + [ + 210, + 119 + ] + ] + ] + }, + { + "task_id": "0004439", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122061330#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004439/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 110 + ], + [ + 298, + 99 + ], + [ + 320, + 90 + ], + [ + 312, + 128 + ], + [ + 339, + 102 + ], + [ + 330, + 140 + ] + ], + [ + [ + 301, + 102 + ], + [ + 323, + 104 + ], + [ + 307, + 81 + ], + [ + 296, + 124 + ], + [ + 286, + 80 + ], + [ + 275, + 122 + ] + ], + [ + [ + 292, + 105 + ], + [ + 291, + 81 + ], + [ + 315, + 104 + ], + [ + 269, + 105 + ], + [ + 316, + 127 + ], + [ + 270, + 129 + ] + ], + [ + [ + 301, + 94 + ], + [ + 289, + 93 + ], + [ + 297, + 117 + ], + [ + 306, + 70 + ], + [ + 308, + 117 + ], + [ + 317, + 72 + ] + ] + ] + }, + { + "task_id": "0004440", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_104122061330#325", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004440/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 331, + 120 + ], + [ + 318, + 142 + ], + [ + 338, + 137 + ], + [ + 324, + 102 + ], + [ + 351, + 117 + ], + [ + 338, + 80 + ] + ], + [ + [ + 338, + 150 + ], + [ + 318, + 167 + ], + [ + 356, + 171 + ], + [ + 321, + 129 + ], + [ + 377, + 153 + ], + [ + 342, + 111 + ] + ], + [ + [ + 315, + 114 + ], + [ + 333, + 91 + ], + [ + 292, + 96 + ], + [ + 338, + 132 + ], + [ + 275, + 118 + ], + [ + 320, + 153 + ] + ], + [ + [ + 335, + 133 + ], + [ + 357, + 144 + ], + [ + 317, + 138 + ], + [ + 352, + 128 + ], + [ + 296, + 127 + ], + [ + 332, + 118 + ] + ] + ] + }, + { + "task_id": "0004441", + "id": "RH20T_cfg1_task_0108_user_0014_scene_0004_cfg_0001_819112070044#125", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Sharpen the pencil with a pencil sharpener'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004441/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 357, + 225 + ], + [ + 377, + 216 + ], + [ + 348, + 204 + ], + [ + 365, + 246 + ], + [ + 327, + 213 + ], + [ + 344, + 256 + ] + ], + [ + [ + 330, + 208 + ], + [ + 350, + 211 + ], + [ + 332, + 223 + ], + [ + 327, + 193 + ], + [ + 310, + 220 + ], + [ + 307, + 189 + ] + ], + [ + [ + 361, + 221 + ], + [ + 368, + 206 + ], + [ + 368, + 241 + ], + [ + 355, + 202 + ], + [ + 361, + 255 + ], + [ + 349, + 217 + ] + ], + [ + [ + 344, + 208 + ], + [ + 347, + 191 + ], + [ + 348, + 193 + ], + [ + 341, + 225 + ], + [ + 345, + 209 + ], + [ + 338, + 241 + ] + ] + ] + }, + { + "task_id": "0004443", + "id": "RH20T_cfg1_task_0012_user_0014_scene_0004_cfg_0001_038522062547#117", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push the soccer ball into the goal'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004443/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 437, + 142 + ], + [ + 447, + 133 + ], + [ + 436, + 166 + ], + [ + 439, + 119 + ], + [ + 427, + 172 + ], + [ + 430, + 127 + ] + ], + [ + [ + 428, + 122 + ], + [ + 410, + 120 + ], + [ + 433, + 145 + ], + [ + 423, + 100 + ], + [ + 450, + 146 + ], + [ + 439, + 102 + ] + ], + [ + [ + 425, + 163 + ], + [ + 405, + 173 + ], + [ + 419, + 139 + ], + [ + 432, + 185 + ], + [ + 439, + 129 + ], + [ + 451, + 175 + ] + ], + [ + [ + 441, + 125 + ], + [ + 435, + 124 + ], + [ + 418, + 133 + ], + [ + 465, + 117 + ], + [ + 424, + 134 + ], + [ + 469, + 119 + ] + ] + ] + }, + { + "task_id": "0004445", + "id": "RH20T_cfg1_task_0064_user_0015_scene_0008_cfg_0001_750612070853#184", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004445/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 609, + 141 + ], + [ + 609, + 101 + ], + [ + 568, + 172 + ], + [ + 640, + 110 + ], + [ + 571, + 204 + ], + [ + 640, + 146 + ] + ], + [ + [ + 589, + 174 + ], + [ + 572, + 218 + ], + [ + 632, + 160 + ], + [ + 554, + 185 + ], + [ + 640, + 114 + ], + [ + 569, + 148 + ] + ], + [ + [ + 640, + 166 + ], + [ + 640, + 183 + ], + [ + 582, + 177 + ], + [ + 640, + 153 + ], + [ + 572, + 159 + ], + [ + 640, + 130 + ] + ], + [ + [ + 583, + 83 + ], + [ + 603, + 46 + ], + [ + 563, + 45 + ], + [ + 606, + 125 + ], + [ + 549, + 78 + ], + [ + 588, + 153 + ] + ] + ] + }, + { + "task_id": "0004449", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0001_cfg_0006_037522061512#198", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004449/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 163, + 181 + ], + [ + 150, + 167 + ], + [ + 146, + 200 + ], + [ + 180, + 162 + ], + [ + 158, + 212 + ], + [ + 191, + 175 + ] + ], + [ + [ + 171, + 140 + ], + [ + 155, + 129 + ], + [ + 178, + 157 + ], + [ + 163, + 122 + ], + [ + 195, + 168 + ], + [ + 181, + 135 + ] + ], + [ + [ + 179, + 167 + ], + [ + 170, + 150 + ], + [ + 193, + 149 + ], + [ + 163, + 185 + ], + [ + 200, + 164 + ], + [ + 172, + 200 + ] + ], + [ + [ + 176, + 154 + ], + [ + 196, + 167 + ], + [ + 173, + 141 + ], + [ + 178, + 165 + ], + [ + 153, + 126 + ], + [ + 160, + 152 + ] + ] + ] + }, + { + "task_id": "0004450", + "id": "RH20T_cfg7_task_0020_user_0014_scene_0010_cfg_0007_104122061602#149", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004450/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 472, + 162 + ], + [ + 481, + 162 + ], + [ + 472, + 187 + ], + [ + 473, + 137 + ], + [ + 464, + 186 + ], + [ + 466, + 139 + ] + ], + [ + [ + 442, + 175 + ], + [ + 429, + 171 + ], + [ + 462, + 177 + ], + [ + 423, + 174 + ], + [ + 474, + 181 + ], + [ + 436, + 179 + ] + ], + [ + [ + 440, + 175 + ], + [ + 456, + 186 + ], + [ + 438, + 195 + ], + [ + 442, + 154 + ], + [ + 424, + 185 + ], + [ + 427, + 145 + ] + ], + [ + [ + 427, + 165 + ], + [ + 430, + 143 + ], + [ + 427, + 168 + ], + [ + 427, + 162 + ], + [ + 425, + 190 + ], + [ + 424, + 186 + ] + ] + ] + }, + { + "task_id": "0004452", + "id": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_104122064161#255", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004452/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 389, + 149 + ], + [ + 368, + 150 + ], + [ + 388, + 155 + ], + [ + 389, + 142 + ], + [ + 408, + 154 + ], + [ + 410, + 140 + ] + ], + [ + [ + 386, + 150 + ], + [ + 366, + 149 + ], + [ + 387, + 167 + ], + [ + 386, + 134 + ], + [ + 408, + 168 + ], + [ + 406, + 135 + ] + ], + [ + [ + 378, + 131 + ], + [ + 379, + 123 + ], + [ + 378, + 150 + ], + [ + 377, + 113 + ], + [ + 376, + 158 + ], + [ + 376, + 121 + ] + ], + [ + [ + 376, + 137 + ], + [ + 358, + 137 + ], + [ + 371, + 120 + ], + [ + 381, + 154 + ], + [ + 388, + 120 + ], + [ + 399, + 154 + ] + ] + ] + }, + { + "task_id": "0004456", + "id": "RH20T_cfg3_task_0007_user_0016_scene_0008_cfg_0003_036422060909#181", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004456/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 322, + 121 + ], + [ + 341, + 116 + ], + [ + 324, + 104 + ], + [ + 320, + 137 + ], + [ + 303, + 109 + ], + [ + 301, + 142 + ] + ], + [ + [ + 290, + 118 + ], + [ + 277, + 103 + ], + [ + 274, + 124 + ], + [ + 305, + 113 + ], + [ + 287, + 140 + ], + [ + 318, + 128 + ] + ], + [ + [ + 302, + 111 + ], + [ + 301, + 94 + ], + [ + 319, + 101 + ], + [ + 284, + 122 + ], + [ + 319, + 117 + ], + [ + 285, + 138 + ] + ], + [ + [ + 283, + 120 + ], + [ + 291, + 99 + ], + [ + 300, + 129 + ], + [ + 264, + 110 + ], + [ + 292, + 149 + ], + [ + 255, + 131 + ] + ] + ] + }, + { + "task_id": "0004458", + "id": "RH20T_cfg3_task_0001_user_0016_scene_0007_cfg_0003_036422060909#239", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004458/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 227, + 127 + ], + [ + 206, + 117 + ], + [ + 208, + 133 + ], + [ + 244, + 122 + ], + [ + 231, + 144 + ], + [ + 265, + 131 + ] + ], + [ + [ + 268, + 149 + ], + [ + 253, + 168 + ], + [ + 267, + 139 + ], + [ + 270, + 158 + ], + [ + 283, + 119 + ], + [ + 284, + 140 + ] + ], + [ + [ + 259, + 148 + ], + [ + 258, + 130 + ], + [ + 281, + 137 + ], + [ + 235, + 159 + ], + [ + 281, + 154 + ], + [ + 237, + 175 + ] + ], + [ + [ + 254, + 125 + ], + [ + 249, + 117 + ], + [ + 249, + 98 + ], + [ + 259, + 151 + ], + [ + 253, + 107 + ], + [ + 263, + 157 + ] + ] + ] + }, + { + "task_id": "0004460", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0001_cfg_0003_036422060909#228", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004460/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 327, + 125 + ], + [ + 333, + 108 + ], + [ + 346, + 117 + ], + [ + 306, + 134 + ], + [ + 340, + 132 + ], + [ + 301, + 150 + ] + ], + [ + [ + 314, + 128 + ], + [ + 333, + 134 + ], + [ + 329, + 112 + ], + [ + 299, + 144 + ], + [ + 312, + 107 + ], + [ + 281, + 138 + ] + ], + [ + [ + 346, + 126 + ], + [ + 372, + 129 + ], + [ + 346, + 148 + ], + [ + 347, + 103 + ], + [ + 322, + 146 + ], + [ + 321, + 101 + ] + ], + [ + [ + 309, + 121 + ], + [ + 294, + 106 + ], + [ + 327, + 102 + ], + [ + 292, + 139 + ], + [ + 344, + 117 + ], + [ + 307, + 154 + ] + ] + ] + }, + { + "task_id": "0004461", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0001_cfg_0003_036422060909#281", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004461/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 262, + 185 + ], + [ + 277, + 212 + ], + [ + 272, + 187 + ], + [ + 252, + 184 + ], + [ + 258, + 162 + ], + [ + 236, + 157 + ] + ], + [ + [ + 255, + 190 + ], + [ + 231, + 178 + ], + [ + 251, + 183 + ], + [ + 258, + 196 + ], + [ + 276, + 197 + ], + [ + 281, + 208 + ] + ], + [ + [ + 219, + 158 + ], + [ + 239, + 176 + ], + [ + 241, + 147 + ], + [ + 199, + 169 + ], + [ + 219, + 127 + ], + [ + 177, + 152 + ] + ], + [ + [ + 248, + 178 + ], + [ + 250, + 161 + ], + [ + 272, + 166 + ], + [ + 223, + 190 + ], + [ + 269, + 181 + ], + [ + 223, + 206 + ] + ] + ] + }, + { + "task_id": "0004464", + "id": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_036422060909#86", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004464/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 262, + 151 + ], + [ + 238, + 147 + ], + [ + 271, + 125 + ], + [ + 254, + 177 + ], + [ + 294, + 130 + ], + [ + 277, + 180 + ] + ], + [ + [ + 267, + 104 + ], + [ + 260, + 107 + ], + [ + 265, + 76 + ], + [ + 269, + 132 + ], + [ + 274, + 71 + ], + [ + 278, + 130 + ] + ], + [ + [ + 281, + 134 + ], + [ + 285, + 118 + ], + [ + 303, + 125 + ], + [ + 257, + 144 + ], + [ + 298, + 140 + ], + [ + 254, + 158 + ] + ], + [ + [ + 278, + 129 + ], + [ + 272, + 126 + ], + [ + 281, + 103 + ], + [ + 275, + 156 + ], + [ + 286, + 107 + ], + [ + 280, + 157 + ] + ] + ] + }, + { + "task_id": "0004468", + "id": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_036422060909#317", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004468/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 336, + 117 + ], + [ + 355, + 108 + ], + [ + 322, + 105 + ], + [ + 352, + 130 + ], + [ + 303, + 115 + ], + [ + 332, + 140 + ] + ], + [ + [ + 335, + 93 + ], + [ + 325, + 78 + ], + [ + 357, + 90 + ], + [ + 314, + 97 + ], + [ + 366, + 104 + ], + [ + 324, + 111 + ] + ], + [ + [ + 321, + 98 + ], + [ + 326, + 80 + ], + [ + 319, + 81 + ], + [ + 323, + 117 + ], + [ + 315, + 98 + ], + [ + 318, + 133 + ] + ], + [ + [ + 327, + 103 + ], + [ + 338, + 107 + ], + [ + 348, + 97 + ], + [ + 305, + 109 + ], + [ + 337, + 94 + ], + [ + 296, + 106 + ] + ] + ] + }, + { + "task_id": "0004470", + "id": "RH20T_cfg4_task_0217_user_0010_scene_0003_cfg_0004_036422060909#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004470/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 247, + 150 + ], + [ + 244, + 155 + ], + [ + 258, + 174 + ], + [ + 236, + 125 + ], + [ + 260, + 168 + ], + [ + 239, + 122 + ] + ], + [ + [ + 240, + 114 + ], + [ + 239, + 95 + ], + [ + 255, + 98 + ], + [ + 224, + 131 + ], + [ + 255, + 115 + ], + [ + 226, + 148 + ] + ], + [ + [ + 237, + 126 + ], + [ + 212, + 130 + ], + [ + 243, + 128 + ], + [ + 231, + 123 + ], + [ + 267, + 124 + ], + [ + 259, + 118 + ] + ], + [ + [ + 229, + 136 + ], + [ + 206, + 131 + ], + [ + 237, + 113 + ], + [ + 221, + 160 + ], + [ + 260, + 117 + ], + [ + 245, + 166 + ] + ] + ] + }, + { + "task_id": "0004473", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0003_cfg_0006_104122060811#184", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004473/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 324, + 144 + ], + [ + 331, + 130 + ], + [ + 335, + 146 + ], + [ + 313, + 142 + ], + [ + 328, + 162 + ], + [ + 306, + 157 + ] + ], + [ + [ + 340, + 147 + ], + [ + 341, + 163 + ], + [ + 351, + 146 + ], + [ + 329, + 147 + ], + [ + 349, + 129 + ], + [ + 327, + 131 + ] + ], + [ + [ + 334, + 142 + ], + [ + 330, + 134 + ], + [ + 350, + 136 + ], + [ + 319, + 147 + ], + [ + 353, + 144 + ], + [ + 323, + 154 + ] + ], + [ + [ + 339, + 127 + ], + [ + 346, + 114 + ], + [ + 328, + 128 + ], + [ + 352, + 125 + ], + [ + 321, + 140 + ], + [ + 345, + 138 + ] + ] + ] + }, + { + "task_id": "0004474", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0003_cfg_0006_104122060811#532", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004474/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 143 + ], + [ + 396, + 149 + ], + [ + 397, + 145 + ], + [ + 415, + 142 + ], + [ + 407, + 138 + ], + [ + 425, + 137 + ] + ], + [ + [ + 389, + 122 + ], + [ + 379, + 123 + ], + [ + 380, + 112 + ], + [ + 398, + 131 + ], + [ + 391, + 110 + ], + [ + 408, + 129 + ] + ], + [ + [ + 393, + 123 + ], + [ + 391, + 110 + ], + [ + 402, + 127 + ], + [ + 385, + 119 + ], + [ + 404, + 139 + ], + [ + 387, + 132 + ] + ], + [ + [ + 399, + 131 + ], + [ + 403, + 123 + ], + [ + 412, + 127 + ], + [ + 386, + 134 + ], + [ + 408, + 135 + ], + [ + 383, + 141 + ] + ] + ] + }, + { + "task_id": "0004476", + "id": "RH20T_cfg5_task_0058_user_0007_scene_0002_cfg_0005_105422061350#222", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004476/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 210, + 118 + ], + [ + 233, + 131 + ], + [ + 221, + 94 + ], + [ + 200, + 140 + ], + [ + 198, + 81 + ], + [ + 178, + 127 + ] + ], + [ + [ + 212, + 102 + ], + [ + 219, + 87 + ], + [ + 227, + 84 + ], + [ + 196, + 122 + ], + [ + 221, + 98 + ], + [ + 191, + 135 + ] + ], + [ + [ + 234, + 106 + ], + [ + 235, + 82 + ], + [ + 244, + 93 + ], + [ + 222, + 121 + ], + [ + 243, + 114 + ], + [ + 222, + 143 + ] + ], + [ + [ + 231, + 114 + ], + [ + 238, + 130 + ], + [ + 228, + 136 + ], + [ + 233, + 89 + ], + [ + 221, + 121 + ], + [ + 226, + 75 + ] + ] + ] + }, + { + "task_id": "0004477", + "id": "RH20T_cfg5_task_0058_user_0007_scene_0002_cfg_0005_105422061350#313", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004477/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 263, + 121 + ], + [ + 238, + 128 + ], + [ + 269, + 139 + ], + [ + 258, + 104 + ], + [ + 294, + 133 + ], + [ + 282, + 98 + ] + ], + [ + [ + 225, + 128 + ], + [ + 209, + 126 + ], + [ + 242, + 146 + ], + [ + 206, + 110 + ], + [ + 259, + 149 + ], + [ + 223, + 112 + ] + ], + [ + [ + 258, + 92 + ], + [ + 235, + 107 + ], + [ + 262, + 90 + ], + [ + 254, + 94 + ], + [ + 285, + 74 + ], + [ + 275, + 79 + ] + ], + [ + [ + 256, + 111 + ], + [ + 258, + 86 + ], + [ + 266, + 99 + ], + [ + 246, + 124 + ], + [ + 263, + 122 + ], + [ + 244, + 148 + ] + ] + ] + }, + { + "task_id": "0004482", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0008_cfg_0004_036422060909#97", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004482/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 354, + 107 + ], + [ + 347, + 132 + ], + [ + 380, + 114 + ], + [ + 330, + 100 + ], + [ + 387, + 89 + ], + [ + 337, + 75 + ] + ], + [ + [ + 355, + 125 + ], + [ + 361, + 108 + ], + [ + 347, + 106 + ], + [ + 364, + 145 + ], + [ + 342, + 122 + ], + [ + 358, + 160 + ] + ], + [ + [ + 353, + 109 + ], + [ + 332, + 97 + ], + [ + 340, + 111 + ], + [ + 367, + 107 + ], + [ + 359, + 122 + ], + [ + 387, + 119 + ] + ], + [ + [ + 373, + 149 + ], + [ + 369, + 155 + ], + [ + 380, + 173 + ], + [ + 365, + 127 + ], + [ + 385, + 168 + ], + [ + 368, + 120 + ] + ] + ] + }, + { + "task_id": "0004483", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0008_cfg_0004_036422060909#244", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004483/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 349, + 140 + ], + [ + 357, + 123 + ], + [ + 341, + 121 + ], + [ + 358, + 160 + ], + [ + 334, + 137 + ], + [ + 351, + 175 + ] + ], + [ + [ + 375, + 143 + ], + [ + 358, + 137 + ], + [ + 374, + 118 + ], + [ + 377, + 169 + ], + [ + 390, + 125 + ], + [ + 394, + 173 + ] + ], + [ + [ + 333, + 144 + ], + [ + 358, + 136 + ], + [ + 340, + 164 + ], + [ + 326, + 124 + ], + [ + 316, + 171 + ], + [ + 300, + 132 + ] + ], + [ + [ + 366, + 108 + ], + [ + 342, + 107 + ], + [ + 355, + 104 + ], + [ + 378, + 113 + ], + [ + 377, + 106 + ], + [ + 401, + 114 + ] + ] + ] + }, + { + "task_id": "0004484", + "id": "RH20T_cfg5_task_0001_user_0010_scene_0001_cfg_0005_036422060215#141", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004484/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 307, + 168 + ], + [ + 290, + 169 + ], + [ + 300, + 147 + ], + [ + 315, + 188 + ], + [ + 318, + 146 + ], + [ + 333, + 188 + ] + ], + [ + [ + 300, + 160 + ], + [ + 304, + 142 + ], + [ + 320, + 157 + ], + [ + 280, + 164 + ], + [ + 316, + 174 + ], + [ + 277, + 182 + ] + ], + [ + [ + 276, + 150 + ], + [ + 288, + 148 + ], + [ + 269, + 171 + ], + [ + 283, + 129 + ], + [ + 256, + 174 + ], + [ + 270, + 131 + ] + ], + [ + [ + 302, + 163 + ], + [ + 296, + 152 + ], + [ + 303, + 182 + ], + [ + 302, + 144 + ], + [ + 309, + 191 + ], + [ + 308, + 154 + ] + ] + ] + }, + { + "task_id": "0004485", + "id": "RH20T_cfg4_task_0068_user_0015_scene_0006_cfg_0004_104422070011#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button horizontally with obstacles'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004485/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 159, + 256 + ], + [ + 183, + 231 + ], + [ + 173, + 261 + ], + [ + 143, + 250 + ], + [ + 151, + 285 + ], + [ + 116, + 278 + ] + ], + [ + [ + 154, + 242 + ], + [ + 165, + 227 + ], + [ + 131, + 267 + ], + [ + 175, + 219 + ], + [ + 122, + 279 + ], + [ + 165, + 233 + ] + ], + [ + [ + 176, + 253 + ], + [ + 205, + 266 + ], + [ + 159, + 283 + ], + [ + 192, + 224 + ], + [ + 128, + 269 + ], + [ + 163, + 211 + ] + ], + [ + [ + 155, + 257 + ], + [ + 158, + 281 + ], + [ + 138, + 241 + ], + [ + 173, + 276 + ], + [ + 133, + 216 + ], + [ + 170, + 250 + ] + ] + ] + }, + { + "task_id": "0004487", + "id": "RH20T_cfg6_task_0082_user_0020_scene_0001_cfg_0006_104122061602#285", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004487/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 476, + 131 + ], + [ + 448, + 127 + ], + [ + 466, + 115 + ], + [ + 486, + 149 + ], + [ + 492, + 120 + ], + [ + 515, + 153 + ] + ], + [ + [ + 481, + 156 + ], + [ + 502, + 154 + ], + [ + 468, + 174 + ], + [ + 496, + 136 + ], + [ + 447, + 176 + ], + [ + 474, + 137 + ] + ], + [ + [ + 454, + 148 + ], + [ + 470, + 159 + ], + [ + 479, + 139 + ], + [ + 431, + 157 + ], + [ + 463, + 126 + ], + [ + 415, + 146 + ] + ], + [ + [ + 497, + 151 + ], + [ + 510, + 176 + ], + [ + 482, + 160 + ], + [ + 513, + 140 + ], + [ + 470, + 137 + ], + [ + 499, + 114 + ] + ] + ] + }, + { + "task_id": "0004488", + "id": "RH20T_cfg6_task_0082_user_0020_scene_0001_cfg_0006_104122061602#541", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004488/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 497, + 199 + ], + [ + 475, + 179 + ], + [ + 494, + 191 + ], + [ + 500, + 207 + ], + [ + 519, + 213 + ], + [ + 523, + 227 + ] + ], + [ + [ + 478, + 190 + ], + [ + 497, + 211 + ], + [ + 472, + 209 + ], + [ + 486, + 169 + ], + [ + 456, + 189 + ], + [ + 468, + 149 + ] + ], + [ + [ + 470, + 151 + ], + [ + 446, + 157 + ], + [ + 473, + 180 + ], + [ + 467, + 122 + ], + [ + 500, + 176 + ], + [ + 493, + 115 + ] + ], + [ + [ + 460, + 176 + ], + [ + 484, + 173 + ], + [ + 462, + 148 + ], + [ + 458, + 203 + ], + [ + 437, + 150 + ], + [ + 434, + 206 + ] + ] + ] + }, + { + "task_id": "0004491", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_819112070044#105", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004491/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 499, + 202 + ], + [ + 507, + 189 + ], + [ + 495, + 188 + ], + [ + 502, + 218 + ], + [ + 488, + 201 + ], + [ + 494, + 230 + ] + ], + [ + [ + 487, + 202 + ], + [ + 489, + 223 + ], + [ + 466, + 202 + ], + [ + 510, + 203 + ], + [ + 465, + 182 + ], + [ + 508, + 183 + ] + ], + [ + [ + 503, + 207 + ], + [ + 515, + 210 + ], + [ + 507, + 226 + ], + [ + 499, + 188 + ], + [ + 495, + 222 + ], + [ + 487, + 184 + ] + ], + [ + [ + 515, + 205 + ], + [ + 524, + 195 + ], + [ + 534, + 206 + ], + [ + 497, + 204 + ], + [ + 525, + 217 + ], + [ + 487, + 213 + ] + ] + ] + }, + { + "task_id": "0004493", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_819112070044#202", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004493/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 489, + 186 + ], + [ + 471, + 187 + ], + [ + 487, + 168 + ], + [ + 492, + 206 + ], + [ + 505, + 167 + ], + [ + 510, + 205 + ] + ], + [ + [ + 484, + 186 + ], + [ + 469, + 176 + ], + [ + 498, + 181 + ], + [ + 471, + 189 + ], + [ + 513, + 191 + ], + [ + 485, + 198 + ] + ], + [ + [ + 481, + 197 + ], + [ + 489, + 184 + ], + [ + 479, + 183 + ], + [ + 483, + 213 + ], + [ + 473, + 195 + ], + [ + 476, + 225 + ] + ], + [ + [ + 458, + 215 + ], + [ + 444, + 209 + ], + [ + 470, + 222 + ], + [ + 447, + 209 + ], + [ + 484, + 229 + ], + [ + 461, + 215 + ] + ] + ] + }, + { + "task_id": "0004495", + "id": "RH20T_cfg1_task_0057_user_0015_scene_0009_cfg_0001_819112070044#292", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004495/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 457, + 188 + ], + [ + 462, + 174 + ], + [ + 446, + 176 + ], + [ + 468, + 200 + ], + [ + 442, + 189 + ], + [ + 463, + 214 + ] + ], + [ + [ + 438, + 201 + ], + [ + 427, + 214 + ], + [ + 437, + 188 + ], + [ + 439, + 214 + ], + [ + 448, + 175 + ], + [ + 450, + 200 + ] + ], + [ + [ + 450, + 190 + ], + [ + 456, + 177 + ], + [ + 449, + 176 + ], + [ + 452, + 205 + ], + [ + 444, + 188 + ], + [ + 446, + 218 + ] + ], + [ + [ + 440, + 193 + ], + [ + 425, + 205 + ], + [ + 450, + 198 + ], + [ + 429, + 187 + ], + [ + 465, + 186 + ], + [ + 444, + 174 + ] + ] + ] + }, + { + "task_id": "0004502", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0003_cfg_0005_036422060215#216", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004502/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 250, + 165 + ], + [ + 236, + 186 + ], + [ + 252, + 162 + ], + [ + 248, + 168 + ], + [ + 266, + 140 + ], + [ + 260, + 149 + ] + ], + [ + [ + 264, + 218 + ], + [ + 249, + 222 + ], + [ + 258, + 195 + ], + [ + 271, + 242 + ], + [ + 272, + 192 + ], + [ + 284, + 237 + ] + ], + [ + [ + 243, + 188 + ], + [ + 243, + 171 + ], + [ + 220, + 183 + ], + [ + 266, + 193 + ], + [ + 220, + 202 + ], + [ + 267, + 211 + ] + ], + [ + [ + 242, + 186 + ], + [ + 247, + 169 + ], + [ + 265, + 186 + ], + [ + 217, + 185 + ], + [ + 259, + 201 + ], + [ + 213, + 201 + ] + ] + ] + }, + { + "task_id": "0004503", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0003_cfg_0005_036422060215#406", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004503/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 235, + 191 + ], + [ + 238, + 176 + ], + [ + 258, + 198 + ], + [ + 211, + 184 + ], + [ + 254, + 212 + ], + [ + 208, + 199 + ] + ], + [ + [ + 220, + 172 + ], + [ + 229, + 162 + ], + [ + 235, + 159 + ], + [ + 204, + 184 + ], + [ + 225, + 170 + ], + [ + 193, + 196 + ] + ], + [ + [ + 212, + 192 + ], + [ + 197, + 192 + ], + [ + 224, + 214 + ], + [ + 201, + 171 + ], + [ + 237, + 213 + ], + [ + 215, + 172 + ] + ], + [ + [ + 255, + 198 + ], + [ + 250, + 221 + ], + [ + 232, + 192 + ], + [ + 279, + 204 + ], + [ + 237, + 169 + ], + [ + 285, + 181 + ] + ] + ] + }, + { + "task_id": "0004504", + "id": "RH20T_cfg2_task_0005_user_0013_scene_0002_cfg_0002_037522062165#185", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004504/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 302, + 289 + ], + [ + 273, + 300 + ], + [ + 303, + 302 + ], + [ + 301, + 277 + ], + [ + 333, + 291 + ], + [ + 327, + 266 + ] + ], + [ + [ + 266, + 296 + ], + [ + 265, + 284 + ], + [ + 251, + 274 + ], + [ + 282, + 319 + ], + [ + 252, + 286 + ], + [ + 282, + 329 + ] + ], + [ + [ + 260, + 320 + ], + [ + 267, + 348 + ], + [ + 231, + 328 + ], + [ + 288, + 312 + ], + [ + 225, + 300 + ], + [ + 280, + 286 + ] + ], + [ + [ + 260, + 328 + ], + [ + 256, + 298 + ], + [ + 237, + 338 + ], + [ + 281, + 319 + ], + [ + 242, + 360 + ], + [ + 285, + 348 + ] + ] + ] + }, + { + "task_id": "0004507", + "id": "RH20T_cfg2_task_0101_user_0007_scene_0002_cfg_0002_037522062165#192", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the table with a cloth'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004507/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 173, + 248 + ], + [ + 158, + 274 + ], + [ + 148, + 235 + ], + [ + 199, + 261 + ], + [ + 163, + 211 + ], + [ + 213, + 236 + ] + ], + [ + [ + 174, + 246 + ], + [ + 148, + 247 + ], + [ + 170, + 222 + ], + [ + 179, + 271 + ], + [ + 194, + 222 + ], + [ + 203, + 269 + ] + ], + [ + [ + 183, + 244 + ], + [ + 172, + 265 + ], + [ + 194, + 264 + ], + [ + 174, + 226 + ], + [ + 206, + 242 + ], + [ + 185, + 205 + ] + ], + [ + [ + 176, + 251 + ], + [ + 172, + 238 + ], + [ + 155, + 237 + ], + [ + 197, + 266 + ], + [ + 159, + 249 + ], + [ + 199, + 278 + ] + ] + ] + }, + { + "task_id": "0004508", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_105422061350#453", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004508/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 273, + 130 + ], + [ + 281, + 111 + ], + [ + 253, + 138 + ], + [ + 294, + 121 + ], + [ + 243, + 159 + ], + [ + 286, + 142 + ] + ], + [ + [ + 268, + 185 + ], + [ + 245, + 170 + ], + [ + 273, + 165 + ], + [ + 263, + 206 + ], + [ + 294, + 179 + ], + [ + 286, + 220 + ] + ], + [ + [ + 273, + 154 + ], + [ + 280, + 155 + ], + [ + 278, + 124 + ], + [ + 268, + 183 + ], + [ + 271, + 124 + ], + [ + 262, + 180 + ] + ], + [ + [ + 266, + 160 + ], + [ + 268, + 143 + ], + [ + 294, + 167 + ], + [ + 237, + 152 + ], + [ + 291, + 182 + ], + [ + 237, + 168 + ] + ] + ] + }, + { + "task_id": "0004510", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_105422061350#648", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004510/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 127 + ], + [ + 277, + 109 + ], + [ + 299, + 139 + ], + [ + 249, + 116 + ], + [ + 296, + 155 + ], + [ + 248, + 133 + ] + ], + [ + [ + 252, + 148 + ], + [ + 237, + 162 + ], + [ + 246, + 124 + ], + [ + 258, + 170 + ], + [ + 261, + 112 + ], + [ + 272, + 157 + ] + ], + [ + [ + 261, + 126 + ], + [ + 259, + 152 + ], + [ + 244, + 117 + ], + [ + 277, + 133 + ], + [ + 246, + 89 + ], + [ + 279, + 108 + ] + ], + [ + [ + 287, + 145 + ], + [ + 269, + 138 + ], + [ + 268, + 140 + ], + [ + 305, + 148 + ], + [ + 288, + 148 + ], + [ + 324, + 155 + ] + ] + ] + }, + { + "task_id": "0004511", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0002_cfg_0005_105422061350#700", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004511/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 97 + ], + [ + 282, + 111 + ], + [ + 262, + 73 + ], + [ + 270, + 119 + ], + [ + 247, + 60 + ], + [ + 256, + 106 + ] + ], + [ + [ + 288, + 113 + ], + [ + 292, + 94 + ], + [ + 312, + 124 + ], + [ + 264, + 101 + ], + [ + 308, + 140 + ], + [ + 261, + 119 + ] + ], + [ + [ + 307, + 119 + ], + [ + 317, + 101 + ], + [ + 328, + 134 + ], + [ + 285, + 103 + ], + [ + 318, + 153 + ], + [ + 273, + 121 + ] + ], + [ + [ + 306, + 118 + ], + [ + 320, + 108 + ], + [ + 287, + 103 + ], + [ + 325, + 134 + ], + [ + 272, + 113 + ], + [ + 311, + 147 + ] + ] + ] + }, + { + "task_id": "0004514", + "id": "RH20T_cfg5_task_0039_user_0007_scene_0007_cfg_0005_105422061350#279", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004514/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 200, + 222 + ], + [ + 173, + 226 + ], + [ + 200, + 191 + ], + [ + 199, + 253 + ], + [ + 228, + 186 + ], + [ + 228, + 250 + ] + ], + [ + [ + 195, + 227 + ], + [ + 223, + 242 + ], + [ + 179, + 248 + ], + [ + 209, + 209 + ], + [ + 149, + 232 + ], + [ + 182, + 194 + ] + ], + [ + [ + 193, + 163 + ], + [ + 216, + 186 + ], + [ + 209, + 151 + ], + [ + 177, + 175 + ], + [ + 183, + 126 + ], + [ + 153, + 153 + ] + ], + [ + [ + 193, + 191 + ], + [ + 190, + 174 + ], + [ + 213, + 169 + ], + [ + 171, + 216 + ], + [ + 214, + 185 + ], + [ + 175, + 230 + ] + ] + ] + }, + { + "task_id": "0004515", + "id": "RH20T_cfg1_task_0002_user_0001_scene_0006_cfg_0001_035622060973#241", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004515/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 268, + 205 + ], + [ + 249, + 224 + ], + [ + 264, + 224 + ], + [ + 272, + 184 + ], + [ + 280, + 207 + ], + [ + 289, + 168 + ] + ], + [ + [ + 267, + 216 + ], + [ + 275, + 206 + ], + [ + 265, + 189 + ], + [ + 269, + 245 + ], + [ + 258, + 199 + ], + [ + 262, + 252 + ] + ], + [ + [ + 261, + 200 + ], + [ + 240, + 222 + ], + [ + 280, + 222 + ], + [ + 242, + 179 + ], + [ + 301, + 200 + ], + [ + 262, + 157 + ] + ], + [ + [ + 274, + 208 + ], + [ + 272, + 189 + ], + [ + 245, + 217 + ], + [ + 302, + 200 + ], + [ + 249, + 234 + ], + [ + 303, + 218 + ] + ] + ] + }, + { + "task_id": "0004518", + "id": "RH20T_cfg4_task_0066_user_0015_scene_0001_cfg_0004_104422070011#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004518/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 374, + 171 + ], + [ + 365, + 196 + ], + [ + 400, + 177 + ], + [ + 349, + 166 + ], + [ + 408, + 152 + ], + [ + 358, + 143 + ] + ], + [ + [ + 407, + 169 + ], + [ + 424, + 182 + ], + [ + 412, + 146 + ], + [ + 402, + 190 + ], + [ + 395, + 134 + ], + [ + 387, + 178 + ] + ], + [ + [ + 396, + 147 + ], + [ + 390, + 127 + ], + [ + 370, + 151 + ], + [ + 422, + 143 + ], + [ + 375, + 172 + ], + [ + 429, + 165 + ] + ], + [ + [ + 385, + 158 + ], + [ + 395, + 143 + ], + [ + 362, + 163 + ], + [ + 407, + 153 + ], + [ + 354, + 175 + ], + [ + 397, + 165 + ] + ] + ] + }, + { + "task_id": "0004520", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_819112070044#891", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004520/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 373, + 185 + ], + [ + 376, + 171 + ], + [ + 390, + 181 + ], + [ + 356, + 189 + ], + [ + 387, + 194 + ], + [ + 354, + 203 + ] + ], + [ + [ + 362, + 168 + ], + [ + 372, + 158 + ], + [ + 378, + 167 + ], + [ + 348, + 168 + ], + [ + 368, + 178 + ], + [ + 337, + 178 + ] + ], + [ + [ + 361, + 201 + ], + [ + 349, + 185 + ], + [ + 352, + 203 + ], + [ + 370, + 198 + ], + [ + 363, + 219 + ], + [ + 381, + 214 + ] + ], + [ + [ + 373, + 210 + ], + [ + 372, + 208 + ], + [ + 367, + 191 + ], + [ + 379, + 228 + ], + [ + 367, + 193 + ], + [ + 380, + 231 + ] + ] + ] + }, + { + "task_id": "0004525", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_039422060546#104", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004525/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 172, + 360 + ], + [ + 207, + 360 + ], + [ + 202, + 350 + ], + [ + 136, + 360 + ], + [ + 171, + 352 + ], + [ + 102, + 360 + ] + ], + [ + [ + 214, + 308 + ], + [ + 203, + 289 + ], + [ + 210, + 353 + ], + [ + 218, + 268 + ], + [ + 220, + 360 + ], + [ + 226, + 286 + ] + ], + [ + [ + 169, + 342 + ], + [ + 150, + 360 + ], + [ + 148, + 352 + ], + [ + 194, + 331 + ], + [ + 165, + 310 + ], + [ + 210, + 285 + ] + ], + [ + [ + 236, + 326 + ], + [ + 230, + 360 + ], + [ + 194, + 316 + ], + [ + 274, + 334 + ], + [ + 203, + 268 + ], + [ + 278, + 290 + ] + ] + ] + }, + { + "task_id": "0004526", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_039422060546#241", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004526/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 296, + 128 + ], + [ + 268, + 120 + ], + [ + 289, + 160 + ], + [ + 304, + 95 + ], + [ + 319, + 170 + ], + [ + 335, + 103 + ] + ], + [ + [ + 320, + 200 + ], + [ + 287, + 195 + ], + [ + 311, + 234 + ], + [ + 328, + 169 + ], + [ + 346, + 240 + ], + [ + 361, + 173 + ] + ], + [ + [ + 301, + 159 + ], + [ + 296, + 136 + ], + [ + 286, + 186 + ], + [ + 315, + 133 + ], + [ + 292, + 205 + ], + [ + 319, + 154 + ] + ], + [ + [ + 312, + 154 + ], + [ + 343, + 134 + ], + [ + 291, + 125 + ], + [ + 335, + 184 + ], + [ + 259, + 145 + ], + [ + 302, + 206 + ] + ] + ] + }, + { + "task_id": "0004527", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_039422060546#412", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004527/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 89 + ], + [ + 388, + 112 + ], + [ + 355, + 105 + ], + [ + 385, + 74 + ], + [ + 337, + 81 + ], + [ + 368, + 52 + ] + ], + [ + [ + 355, + 81 + ], + [ + 354, + 51 + ], + [ + 339, + 87 + ], + [ + 370, + 75 + ], + [ + 341, + 116 + ], + [ + 371, + 103 + ] + ], + [ + [ + 346, + 64 + ], + [ + 344, + 39 + ], + [ + 340, + 83 + ], + [ + 352, + 46 + ], + [ + 342, + 106 + ], + [ + 354, + 68 + ] + ], + [ + [ + 331, + 60 + ], + [ + 361, + 64 + ], + [ + 336, + 28 + ], + [ + 327, + 90 + ], + [ + 307, + 25 + ], + [ + 299, + 86 + ] + ] + ] + }, + { + "task_id": "0004528", + "id": "RH20T_cfg4_task_0011_user_0016_scene_0002_cfg_0004_036422060909#164", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004528/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 241, + 180 + ], + [ + 209, + 169 + ], + [ + 247, + 156 + ], + [ + 234, + 206 + ], + [ + 276, + 166 + ], + [ + 266, + 216 + ] + ], + [ + [ + 244, + 154 + ], + [ + 238, + 135 + ], + [ + 228, + 131 + ], + [ + 261, + 177 + ], + [ + 234, + 149 + ], + [ + 265, + 192 + ] + ], + [ + [ + 255, + 164 + ], + [ + 254, + 157 + ], + [ + 244, + 134 + ], + [ + 265, + 193 + ], + [ + 243, + 139 + ], + [ + 266, + 202 + ] + ], + [ + [ + 237, + 148 + ], + [ + 255, + 166 + ], + [ + 253, + 164 + ], + [ + 222, + 134 + ], + [ + 233, + 144 + ], + [ + 202, + 115 + ] + ] + ] + }, + { + "task_id": "0004529", + "id": "RH20T_cfg6_task_0013_user_0014_scene_0007_cfg_0006_104122061602#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004529/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 395, + 215 + ], + [ + 390, + 194 + ], + [ + 402, + 198 + ], + [ + 388, + 233 + ], + [ + 406, + 218 + ], + [ + 393, + 253 + ] + ], + [ + [ + 361, + 203 + ], + [ + 338, + 215 + ], + [ + 349, + 180 + ], + [ + 374, + 225 + ], + [ + 373, + 167 + ], + [ + 397, + 212 + ] + ], + [ + [ + 389, + 200 + ], + [ + 403, + 187 + ], + [ + 402, + 225 + ], + [ + 377, + 178 + ], + [ + 388, + 236 + ], + [ + 365, + 191 + ] + ], + [ + [ + 383, + 202 + ], + [ + 369, + 203 + ], + [ + 399, + 182 + ], + [ + 368, + 221 + ], + [ + 411, + 181 + ], + [ + 381, + 219 + ] + ] + ] + }, + { + "task_id": "0004530", + "id": "RH20T_cfg1_task_0014_user_0014_scene_0001_cfg_0001_038522062547#136", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the object from the scale'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004530/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 424, + 123 + ], + [ + 411, + 127 + ], + [ + 415, + 146 + ], + [ + 434, + 100 + ], + [ + 430, + 143 + ], + [ + 450, + 94 + ] + ], + [ + [ + 421, + 121 + ], + [ + 436, + 135 + ], + [ + 433, + 100 + ], + [ + 409, + 143 + ], + [ + 417, + 84 + ], + [ + 392, + 128 + ] + ], + [ + [ + 403, + 103 + ], + [ + 408, + 90 + ], + [ + 407, + 126 + ], + [ + 399, + 80 + ], + [ + 402, + 137 + ], + [ + 395, + 93 + ] + ], + [ + [ + 397, + 126 + ], + [ + 422, + 118 + ], + [ + 400, + 125 + ], + [ + 394, + 128 + ], + [ + 376, + 133 + ], + [ + 368, + 137 + ] + ] + ] + }, + { + "task_id": "0004534", + "id": "RH20T_cfg2_task_0065_user_0007_scene_0010_cfg_0002_104122061850#1022", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004534/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 457, + 274 + ], + [ + 473, + 260 + ], + [ + 482, + 277 + ], + [ + 434, + 271 + ], + [ + 463, + 293 + ], + [ + 417, + 284 + ] + ], + [ + [ + 463, + 267 + ], + [ + 461, + 255 + ], + [ + 459, + 294 + ], + [ + 467, + 241 + ], + [ + 460, + 303 + ], + [ + 468, + 252 + ] + ], + [ + [ + 452, + 255 + ], + [ + 482, + 261 + ], + [ + 448, + 283 + ], + [ + 456, + 227 + ], + [ + 420, + 276 + ], + [ + 428, + 221 + ] + ], + [ + [ + 473, + 272 + ], + [ + 477, + 282 + ], + [ + 482, + 247 + ], + [ + 464, + 299 + ], + [ + 478, + 235 + ], + [ + 459, + 289 + ] + ] + ] + }, + { + "task_id": "0004538", + "id": "RH20T_cfg2_task_0029_user_0003_scene_0005_cfg_0002_037522062165#172", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004538/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 137, + 151 + ], + [ + 111, + 157 + ], + [ + 124, + 159 + ], + [ + 148, + 144 + ], + [ + 152, + 152 + ], + [ + 173, + 138 + ] + ], + [ + [ + 152, + 152 + ], + [ + 146, + 133 + ], + [ + 175, + 136 + ], + [ + 127, + 169 + ], + [ + 179, + 153 + ], + [ + 134, + 185 + ] + ], + [ + [ + 148, + 185 + ], + [ + 143, + 209 + ], + [ + 143, + 166 + ], + [ + 152, + 203 + ], + [ + 148, + 144 + ], + [ + 156, + 182 + ] + ], + [ + [ + 153, + 139 + ], + [ + 149, + 146 + ], + [ + 124, + 142 + ], + [ + 181, + 135 + ], + [ + 127, + 135 + ], + [ + 187, + 128 + ] + ] + ] + }, + { + "task_id": "0004541", + "id": "RH20T_cfg2_task_0029_user_0003_scene_0005_cfg_0002_037522062165#307", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004541/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 184, + 181 + ], + [ + 180, + 205 + ], + [ + 162, + 187 + ], + [ + 208, + 174 + ], + [ + 166, + 164 + ], + [ + 211, + 151 + ] + ], + [ + [ + 160, + 167 + ], + [ + 130, + 169 + ], + [ + 158, + 139 + ], + [ + 161, + 195 + ], + [ + 186, + 138 + ], + [ + 188, + 192 + ] + ], + [ + [ + 163, + 169 + ], + [ + 155, + 151 + ], + [ + 183, + 150 + ], + [ + 141, + 189 + ], + [ + 189, + 166 + ], + [ + 150, + 204 + ] + ], + [ + [ + 170, + 145 + ], + [ + 189, + 158 + ], + [ + 149, + 163 + ], + [ + 190, + 128 + ], + [ + 130, + 151 + ], + [ + 171, + 117 + ] + ] + ] + }, + { + "task_id": "0004544", + "id": "RH20T_cfg6_task_0104_user_0020_scene_0006_cfg_0006_104122064161#184", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate back after holding it down'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004544/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 367, + 122 + ], + [ + 370, + 132 + ], + [ + 356, + 138 + ], + [ + 378, + 106 + ], + [ + 353, + 128 + ], + [ + 374, + 97 + ] + ], + [ + [ + 369, + 125 + ], + [ + 384, + 135 + ], + [ + 361, + 123 + ], + [ + 377, + 126 + ], + [ + 346, + 114 + ], + [ + 361, + 116 + ] + ], + [ + [ + 365, + 134 + ], + [ + 368, + 116 + ], + [ + 346, + 133 + ], + [ + 384, + 135 + ], + [ + 343, + 152 + ], + [ + 382, + 155 + ] + ], + [ + [ + 360, + 128 + ], + [ + 371, + 131 + ], + [ + 345, + 127 + ], + [ + 377, + 129 + ], + [ + 332, + 123 + ], + [ + 365, + 126 + ] + ] + ] + }, + { + "task_id": "0004547", + "id": "RH20T_cfg4_task_0059_user_0016_scene_0003_cfg_0004_104122062295#306", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004547/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 0, + 218 + ], + [ + 0, + 232 + ], + [ + 0, + 168 + ], + [ + 0, + 274 + ], + [ + 35, + 162 + ], + [ + 63, + 257 + ] + ], + [ + [ + 8, + 199 + ], + [ + 6, + 147 + ], + [ + 0, + 168 + ], + [ + 22, + 235 + ], + [ + 0, + 212 + ], + [ + 22, + 282 + ] + ], + [ + [ + 35, + 174 + ], + [ + 4, + 150 + ], + [ + 0, + 225 + ], + [ + 81, + 129 + ], + [ + 15, + 239 + ], + [ + 99, + 152 + ] + ], + [ + [ + 15, + 254 + ], + [ + 84, + 249 + ], + [ + 30, + 248 + ], + [ + 0, + 261 + ], + [ + 0, + 252 + ], + [ + 0, + 268 + ] + ] + ] + }, + { + "task_id": "0004548", + "id": "RH20T_cfg5_task_0011_user_0007_scene_0004_cfg_0005_105422061350#246", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004548/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 174, + 130 + ], + [ + 188, + 136 + ], + [ + 205, + 139 + ], + [ + 140, + 121 + ], + [ + 192, + 134 + ], + [ + 128, + 116 + ] + ], + [ + [ + 179, + 133 + ], + [ + 178, + 112 + ], + [ + 214, + 132 + ], + [ + 143, + 135 + ], + [ + 213, + 150 + ], + [ + 147, + 153 + ] + ], + [ + [ + 203, + 105 + ], + [ + 225, + 81 + ], + [ + 230, + 121 + ], + [ + 174, + 88 + ], + [ + 209, + 142 + ], + [ + 152, + 113 + ] + ], + [ + [ + 179, + 134 + ], + [ + 198, + 125 + ], + [ + 199, + 117 + ], + [ + 161, + 150 + ], + [ + 179, + 126 + ], + [ + 140, + 161 + ] + ] + ] + }, + { + "task_id": "0004551", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0006_cfg_0003_104122062823#282", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004551/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 166, + 48 + ], + [ + 181, + 20 + ], + [ + 148, + 30 + ], + [ + 186, + 66 + ], + [ + 134, + 56 + ], + [ + 171, + 92 + ] + ], + [ + [ + 177, + 61 + ], + [ + 156, + 63 + ], + [ + 156, + 39 + ], + [ + 197, + 81 + ], + [ + 179, + 36 + ], + [ + 219, + 80 + ] + ], + [ + [ + 190, + 82 + ], + [ + 215, + 67 + ], + [ + 209, + 104 + ], + [ + 171, + 60 + ], + [ + 182, + 121 + ], + [ + 144, + 76 + ] + ], + [ + [ + 188, + 53 + ], + [ + 182, + 24 + ], + [ + 213, + 65 + ], + [ + 163, + 42 + ], + [ + 217, + 90 + ], + [ + 169, + 67 + ] + ] + ] + }, + { + "task_id": "0004554", + "id": "RH20T_cfg5_task_0001_user_0010_scene_0003_cfg_0005_104422070044#166", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004554/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 419, + 151 + ], + [ + 407, + 156 + ], + [ + 407, + 140 + ], + [ + 432, + 163 + ], + [ + 420, + 134 + ], + [ + 445, + 158 + ] + ], + [ + [ + 435, + 153 + ], + [ + 446, + 147 + ], + [ + 420, + 147 + ], + [ + 450, + 160 + ], + [ + 409, + 153 + ], + [ + 439, + 167 + ] + ], + [ + [ + 416, + 153 + ], + [ + 417, + 139 + ], + [ + 433, + 159 + ], + [ + 399, + 146 + ], + [ + 432, + 173 + ], + [ + 399, + 160 + ] + ], + [ + [ + 417, + 152 + ], + [ + 429, + 166 + ], + [ + 402, + 162 + ], + [ + 432, + 141 + ], + [ + 391, + 148 + ], + [ + 421, + 126 + ] + ] + ] + }, + { + "task_id": "0004556", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0006_cfg_0001_819112070044#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004556/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 396, + 193 + ], + [ + 406, + 208 + ], + [ + 408, + 196 + ], + [ + 382, + 191 + ], + [ + 398, + 183 + ], + [ + 372, + 177 + ] + ], + [ + [ + 406, + 175 + ], + [ + 424, + 188 + ], + [ + 394, + 190 + ], + [ + 418, + 160 + ], + [ + 376, + 177 + ], + [ + 400, + 148 + ] + ], + [ + [ + 389, + 183 + ], + [ + 385, + 171 + ], + [ + 406, + 172 + ], + [ + 372, + 194 + ], + [ + 409, + 183 + ], + [ + 376, + 205 + ] + ], + [ + [ + 384, + 199 + ], + [ + 379, + 195 + ], + [ + 405, + 200 + ], + [ + 363, + 198 + ], + [ + 408, + 204 + ], + [ + 368, + 201 + ] + ] + ] + }, + { + "task_id": "0004559", + "id": "RH20T_cfg5_task_0033_user_0010_scene_0010_cfg_0005_105422061350#280", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004559/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 286, + 86 + ], + [ + 306, + 90 + ], + [ + 297, + 61 + ], + [ + 275, + 112 + ], + [ + 279, + 58 + ], + [ + 258, + 108 + ] + ], + [ + [ + 287, + 105 + ], + [ + 287, + 86 + ], + [ + 301, + 88 + ], + [ + 271, + 122 + ], + [ + 300, + 105 + ], + [ + 272, + 139 + ] + ], + [ + [ + 298, + 96 + ], + [ + 274, + 84 + ], + [ + 287, + 98 + ], + [ + 309, + 94 + ], + [ + 312, + 111 + ], + [ + 332, + 106 + ] + ], + [ + [ + 303, + 88 + ], + [ + 278, + 100 + ], + [ + 296, + 73 + ], + [ + 311, + 105 + ], + [ + 320, + 61 + ], + [ + 338, + 93 + ] + ] + ] + }, + { + "task_id": "0004561", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_036422060215#465", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004561/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 265, + 149 + ], + [ + 252, + 133 + ], + [ + 249, + 163 + ], + [ + 281, + 136 + ], + [ + 262, + 179 + ], + [ + 294, + 152 + ] + ], + [ + [ + 276, + 135 + ], + [ + 281, + 120 + ], + [ + 293, + 133 + ], + [ + 258, + 137 + ], + [ + 288, + 147 + ], + [ + 253, + 151 + ] + ], + [ + [ + 257, + 133 + ], + [ + 263, + 152 + ], + [ + 247, + 135 + ], + [ + 269, + 131 + ], + [ + 242, + 115 + ], + [ + 263, + 110 + ] + ], + [ + [ + 270, + 122 + ], + [ + 250, + 123 + ], + [ + 271, + 108 + ], + [ + 268, + 137 + ], + [ + 290, + 106 + ], + [ + 288, + 136 + ] + ] + ] + }, + { + "task_id": "0004562", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_036422060215#714", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004562/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 238, + 146 + ], + [ + 241, + 130 + ], + [ + 257, + 142 + ], + [ + 218, + 150 + ], + [ + 254, + 157 + ], + [ + 216, + 165 + ] + ], + [ + [ + 246, + 141 + ], + [ + 229, + 150 + ], + [ + 255, + 160 + ], + [ + 238, + 121 + ], + [ + 271, + 150 + ], + [ + 255, + 112 + ] + ], + [ + [ + 217, + 153 + ], + [ + 202, + 139 + ], + [ + 209, + 153 + ], + [ + 224, + 152 + ], + [ + 225, + 167 + ], + [ + 239, + 165 + ] + ], + [ + [ + 243, + 137 + ], + [ + 228, + 133 + ], + [ + 232, + 153 + ], + [ + 254, + 120 + ], + [ + 246, + 155 + ], + [ + 268, + 124 + ] + ] + ] + }, + { + "task_id": "0004563", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0010_cfg_0005_036422060215#1018", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004563/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 283, + 210 + ], + [ + 276, + 191 + ], + [ + 297, + 213 + ], + [ + 268, + 206 + ], + [ + 304, + 232 + ], + [ + 275, + 226 + ] + ], + [ + [ + 266, + 184 + ], + [ + 270, + 168 + ], + [ + 287, + 181 + ], + [ + 244, + 187 + ], + [ + 283, + 196 + ], + [ + 242, + 203 + ] + ], + [ + [ + 262, + 211 + ], + [ + 270, + 228 + ], + [ + 255, + 198 + ], + [ + 269, + 224 + ], + [ + 248, + 182 + ], + [ + 263, + 208 + ] + ], + [ + [ + 275, + 160 + ], + [ + 281, + 177 + ], + [ + 289, + 170 + ], + [ + 261, + 151 + ], + [ + 284, + 154 + ], + [ + 256, + 135 + ] + ] + ] + }, + { + "task_id": "0004564", + "id": "RH20T_cfg6_task_0082_user_0020_scene_0001_cfg_0006_104122061018#261", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004564/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 291, + 100 + ], + [ + 278, + 127 + ], + [ + 263, + 82 + ], + [ + 317, + 117 + ], + [ + 276, + 55 + ], + [ + 328, + 91 + ] + ], + [ + [ + 282, + 78 + ], + [ + 270, + 72 + ], + [ + 281, + 110 + ], + [ + 282, + 44 + ], + [ + 294, + 119 + ], + [ + 296, + 49 + ] + ], + [ + [ + 281, + 47 + ], + [ + 306, + 51 + ], + [ + 288, + 77 + ], + [ + 275, + 18 + ], + [ + 260, + 73 + ], + [ + 248, + 11 + ] + ], + [ + [ + 255, + 76 + ], + [ + 236, + 51 + ], + [ + 235, + 75 + ], + [ + 277, + 76 + ], + [ + 253, + 96 + ], + [ + 294, + 99 + ] + ] + ] + }, + { + "task_id": "0004566", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0005_cfg_0004_036422060909#290", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004566/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 94 + ], + [ + 329, + 115 + ], + [ + 331, + 112 + ], + [ + 346, + 75 + ], + [ + 339, + 93 + ], + [ + 354, + 55 + ] + ], + [ + [ + 332, + 80 + ], + [ + 302, + 89 + ], + [ + 330, + 68 + ], + [ + 333, + 91 + ], + [ + 363, + 59 + ], + [ + 361, + 82 + ] + ], + [ + [ + 349, + 75 + ], + [ + 323, + 81 + ], + [ + 352, + 48 + ], + [ + 346, + 99 + ], + [ + 378, + 43 + ], + [ + 370, + 93 + ] + ], + [ + [ + 350, + 79 + ], + [ + 354, + 54 + ], + [ + 342, + 61 + ], + [ + 359, + 98 + ], + [ + 339, + 84 + ], + [ + 355, + 120 + ] + ] + ] + }, + { + "task_id": "0004567", + "id": "RH20T_cfg5_task_0002_user_0014_scene_0010_cfg_0005_104122063678#197", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004567/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 374, + 218 + ], + [ + 388, + 238 + ], + [ + 356, + 231 + ], + [ + 394, + 205 + ], + [ + 343, + 211 + ], + [ + 380, + 185 + ] + ], + [ + [ + 359, + 226 + ], + [ + 361, + 247 + ], + [ + 357, + 240 + ], + [ + 362, + 210 + ], + [ + 356, + 221 + ], + [ + 360, + 191 + ] + ], + [ + [ + 381, + 221 + ], + [ + 389, + 205 + ], + [ + 387, + 241 + ], + [ + 375, + 202 + ], + [ + 379, + 256 + ], + [ + 368, + 217 + ] + ], + [ + [ + 399, + 221 + ], + [ + 412, + 213 + ], + [ + 413, + 206 + ], + [ + 385, + 236 + ], + [ + 401, + 214 + ], + [ + 374, + 243 + ] + ] + ] + }, + { + "task_id": "0004569", + "id": "RH20T_cfg5_task_0008_user_0007_scene_0001_cfg_0005_036422060215#320", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004569/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 391, + 192 + ], + [ + 407, + 188 + ], + [ + 386, + 208 + ], + [ + 395, + 175 + ], + [ + 370, + 212 + ], + [ + 380, + 179 + ] + ], + [ + [ + 388, + 167 + ], + [ + 408, + 176 + ], + [ + 390, + 156 + ], + [ + 386, + 179 + ], + [ + 371, + 148 + ], + [ + 366, + 170 + ] + ], + [ + [ + 398, + 157 + ], + [ + 386, + 143 + ], + [ + 411, + 158 + ], + [ + 385, + 156 + ], + [ + 423, + 172 + ], + [ + 397, + 170 + ] + ], + [ + [ + 391, + 168 + ], + [ + 399, + 154 + ], + [ + 407, + 165 + ], + [ + 374, + 171 + ], + [ + 398, + 178 + ], + [ + 367, + 184 + ] + ] + ] + }, + { + "task_id": "0004576", + "id": "RH20T_cfg1_task_0081_user_0010_scene_0001_cfg_0001_819112070044#598", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004576/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 351, + 223 + ], + [ + 373, + 219 + ], + [ + 354, + 244 + ], + [ + 347, + 204 + ], + [ + 331, + 248 + ], + [ + 326, + 208 + ] + ], + [ + [ + 327, + 230 + ], + [ + 334, + 213 + ], + [ + 335, + 220 + ], + [ + 319, + 241 + ], + [ + 328, + 236 + ], + [ + 312, + 258 + ] + ], + [ + [ + 326, + 229 + ], + [ + 314, + 211 + ], + [ + 316, + 235 + ], + [ + 336, + 223 + ], + [ + 329, + 255 + ], + [ + 348, + 241 + ] + ], + [ + [ + 341, + 220 + ], + [ + 361, + 231 + ], + [ + 331, + 240 + ], + [ + 351, + 201 + ], + [ + 312, + 229 + ], + [ + 332, + 191 + ] + ] + ] + }, + { + "task_id": "0004577", + "id": "RH20T_cfg7_task_0042_user_0020_scene_0003_cfg_0007_037522061512#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004577/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 253, + 129 + ], + [ + 253, + 111 + ], + [ + 265, + 119 + ], + [ + 241, + 140 + ], + [ + 265, + 135 + ], + [ + 242, + 156 + ] + ], + [ + [ + 239, + 133 + ], + [ + 259, + 138 + ], + [ + 239, + 121 + ], + [ + 239, + 144 + ], + [ + 219, + 115 + ], + [ + 220, + 139 + ] + ], + [ + [ + 251, + 105 + ], + [ + 264, + 115 + ], + [ + 265, + 90 + ], + [ + 237, + 121 + ], + [ + 252, + 81 + ], + [ + 225, + 112 + ] + ], + [ + [ + 240, + 150 + ], + [ + 236, + 167 + ], + [ + 220, + 151 + ], + [ + 260, + 150 + ], + [ + 224, + 135 + ], + [ + 263, + 134 + ] + ] + ] + }, + { + "task_id": "0004579", + "id": "RH20T_cfg7_task_0042_user_0020_scene_0003_cfg_0007_037522061512#454", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004579/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 179, + 163 + ], + [ + 199, + 170 + ], + [ + 180, + 185 + ], + [ + 179, + 141 + ], + [ + 162, + 178 + ], + [ + 159, + 135 + ] + ], + [ + [ + 206, + 152 + ], + [ + 202, + 133 + ], + [ + 220, + 138 + ], + [ + 191, + 166 + ], + [ + 223, + 155 + ], + [ + 195, + 183 + ] + ], + [ + [ + 188, + 128 + ], + [ + 190, + 129 + ], + [ + 179, + 151 + ], + [ + 197, + 105 + ], + [ + 178, + 149 + ], + [ + 195, + 105 + ] + ], + [ + [ + 235, + 156 + ], + [ + 248, + 143 + ], + [ + 252, + 157 + ], + [ + 218, + 156 + ], + [ + 238, + 171 + ], + [ + 204, + 169 + ] + ] + ] + }, + { + "task_id": "0004580", + "id": "RH20T_cfg7_task_0042_user_0020_scene_0003_cfg_0007_037522061512#675", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004580/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 239, + 140 + ], + [ + 249, + 161 + ], + [ + 246, + 146 + ], + [ + 232, + 133 + ], + [ + 237, + 127 + ], + [ + 222, + 112 + ] + ], + [ + [ + 203, + 139 + ], + [ + 218, + 125 + ], + [ + 203, + 123 + ], + [ + 203, + 153 + ], + [ + 186, + 137 + ], + [ + 188, + 167 + ] + ], + [ + [ + 178, + 148 + ], + [ + 184, + 159 + ], + [ + 203, + 146 + ], + [ + 151, + 150 + ], + [ + 196, + 136 + ], + [ + 146, + 139 + ] + ], + [ + [ + 210, + 151 + ], + [ + 204, + 135 + ], + [ + 224, + 135 + ], + [ + 196, + 168 + ], + [ + 229, + 150 + ], + [ + 202, + 183 + ] + ] + ] + }, + { + "task_id": "0004582", + "id": "RH20T_cfg7_task_0042_user_0020_scene_0003_cfg_0007_037522061512#850", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004582/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 200, + 176 + ], + [ + 218, + 169 + ], + [ + 210, + 164 + ], + [ + 191, + 187 + ], + [ + 190, + 170 + ], + [ + 171, + 194 + ] + ], + [ + [ + 213, + 135 + ], + [ + 199, + 131 + ], + [ + 231, + 121 + ], + [ + 194, + 150 + ], + [ + 246, + 125 + ], + [ + 209, + 155 + ] + ], + [ + [ + 209, + 152 + ], + [ + 223, + 135 + ], + [ + 217, + 165 + ], + [ + 201, + 138 + ], + [ + 203, + 183 + ], + [ + 185, + 157 + ] + ], + [ + [ + 210, + 166 + ], + [ + 206, + 153 + ], + [ + 192, + 152 + ], + [ + 227, + 179 + ], + [ + 196, + 167 + ], + [ + 232, + 194 + ] + ] + ] + }, + { + "task_id": "0004584", + "id": "RH20T_cfg7_task_0047_user_0020_scene_0004_cfg_0007_104122063633#143", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004584/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 145, + 161 + ], + [ + 168, + 175 + ], + [ + 150, + 150 + ], + [ + 138, + 173 + ], + [ + 127, + 136 + ], + [ + 112, + 158 + ] + ], + [ + [ + 144, + 144 + ], + [ + 114, + 139 + ], + [ + 140, + 156 + ], + [ + 148, + 130 + ], + [ + 167, + 161 + ], + [ + 178, + 135 + ] + ], + [ + [ + 161, + 138 + ], + [ + 167, + 115 + ], + [ + 185, + 143 + ], + [ + 136, + 132 + ], + [ + 179, + 165 + ], + [ + 130, + 155 + ] + ], + [ + [ + 164, + 145 + ], + [ + 172, + 125 + ], + [ + 157, + 155 + ], + [ + 172, + 133 + ], + [ + 148, + 176 + ], + [ + 164, + 155 + ] + ] + ] + }, + { + "task_id": "0004585", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_039422060546#112", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004585/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 357, + 68 + ], + [ + 381, + 62 + ], + [ + 360, + 98 + ], + [ + 354, + 37 + ], + [ + 336, + 105 + ], + [ + 328, + 42 + ] + ], + [ + [ + 372, + 80 + ], + [ + 372, + 52 + ], + [ + 370, + 92 + ], + [ + 373, + 69 + ], + [ + 370, + 119 + ], + [ + 372, + 95 + ] + ], + [ + [ + 389, + 62 + ], + [ + 400, + 87 + ], + [ + 363, + 71 + ], + [ + 414, + 53 + ], + [ + 352, + 45 + ], + [ + 403, + 27 + ] + ], + [ + [ + 379, + 106 + ], + [ + 362, + 96 + ], + [ + 361, + 129 + ], + [ + 397, + 84 + ], + [ + 379, + 141 + ], + [ + 417, + 94 + ] + ] + ] + }, + { + "task_id": "0004586", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_039422060546#321", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004586/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 404, + 180 + ], + [ + 421, + 158 + ], + [ + 378, + 179 + ], + [ + 433, + 181 + ], + [ + 359, + 202 + ], + [ + 414, + 206 + ] + ], + [ + [ + 380, + 190 + ], + [ + 379, + 192 + ], + [ + 361, + 216 + ], + [ + 399, + 164 + ], + [ + 363, + 212 + ], + [ + 398, + 164 + ] + ], + [ + [ + 401, + 189 + ], + [ + 389, + 175 + ], + [ + 432, + 199 + ], + [ + 372, + 180 + ], + [ + 441, + 211 + ], + [ + 383, + 192 + ] + ], + [ + [ + 405, + 184 + ], + [ + 392, + 171 + ], + [ + 413, + 154 + ], + [ + 397, + 215 + ], + [ + 425, + 167 + ], + [ + 410, + 225 + ] + ] + ] + }, + { + "task_id": "0004590", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_039422060546#467", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004590/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 377, + 109 + ], + [ + 409, + 122 + ], + [ + 389, + 80 + ], + [ + 365, + 139 + ], + [ + 358, + 68 + ], + [ + 334, + 127 + ] + ], + [ + [ + 416, + 128 + ], + [ + 427, + 138 + ], + [ + 411, + 102 + ], + [ + 422, + 156 + ], + [ + 399, + 89 + ], + [ + 410, + 146 + ] + ], + [ + [ + 399, + 153 + ], + [ + 382, + 179 + ], + [ + 381, + 142 + ], + [ + 415, + 163 + ], + [ + 399, + 115 + ], + [ + 432, + 139 + ] + ], + [ + [ + 397, + 121 + ], + [ + 406, + 103 + ], + [ + 398, + 148 + ], + [ + 396, + 97 + ], + [ + 390, + 164 + ], + [ + 389, + 114 + ] + ] + ] + }, + { + "task_id": "0004591", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_039422060546#543", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004591/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 333, + 87 + ], + [ + 313, + 80 + ], + [ + 347, + 60 + ], + [ + 321, + 113 + ], + [ + 365, + 69 + ], + [ + 339, + 119 + ] + ], + [ + [ + 358, + 63 + ], + [ + 380, + 83 + ], + [ + 338, + 82 + ], + [ + 378, + 45 + ], + [ + 317, + 62 + ], + [ + 357, + 26 + ] + ], + [ + [ + 344, + 62 + ], + [ + 347, + 38 + ], + [ + 335, + 81 + ], + [ + 352, + 45 + ], + [ + 333, + 102 + ], + [ + 350, + 66 + ] + ], + [ + [ + 323, + 45 + ], + [ + 304, + 38 + ], + [ + 347, + 38 + ], + [ + 302, + 51 + ], + [ + 364, + 45 + ], + [ + 320, + 57 + ] + ] + ] + }, + { + "task_id": "0004593", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_039422060546#834", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004593/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 475, + 175 + ], + [ + 462, + 177 + ], + [ + 454, + 199 + ], + [ + 497, + 148 + ], + [ + 466, + 196 + ], + [ + 507, + 147 + ] + ], + [ + [ + 470, + 202 + ], + [ + 503, + 219 + ], + [ + 454, + 232 + ], + [ + 486, + 172 + ], + [ + 423, + 216 + ], + [ + 453, + 156 + ] + ], + [ + [ + 449, + 168 + ], + [ + 457, + 148 + ], + [ + 438, + 194 + ], + [ + 459, + 144 + ], + [ + 431, + 211 + ], + [ + 452, + 163 + ] + ], + [ + [ + 484, + 168 + ], + [ + 476, + 198 + ], + [ + 510, + 177 + ], + [ + 458, + 158 + ], + [ + 518, + 147 + ], + [ + 467, + 127 + ] + ] + ] + }, + { + "task_id": "0004594", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_039422060546#841", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004594/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 448, + 167 + ], + [ + 457, + 147 + ], + [ + 437, + 192 + ], + [ + 458, + 142 + ], + [ + 429, + 209 + ], + [ + 450, + 161 + ] + ], + [ + [ + 435, + 154 + ], + [ + 432, + 128 + ], + [ + 410, + 140 + ], + [ + 463, + 169 + ], + [ + 414, + 164 + ], + [ + 465, + 194 + ] + ], + [ + [ + 460, + 177 + ], + [ + 465, + 186 + ], + [ + 466, + 214 + ], + [ + 454, + 143 + ], + [ + 460, + 206 + ], + [ + 448, + 130 + ] + ], + [ + [ + 479, + 172 + ], + [ + 446, + 154 + ], + [ + 496, + 146 + ], + [ + 460, + 198 + ], + [ + 530, + 163 + ], + [ + 495, + 218 + ] + ] + ] + }, + { + "task_id": "0004596", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_039422060546#886", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004596/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 414, + 137 + ], + [ + 437, + 159 + ], + [ + 439, + 117 + ], + [ + 390, + 156 + ], + [ + 415, + 93 + ], + [ + 367, + 134 + ] + ], + [ + [ + 438, + 193 + ], + [ + 461, + 217 + ], + [ + 464, + 171 + ], + [ + 414, + 215 + ], + [ + 441, + 147 + ], + [ + 391, + 192 + ] + ], + [ + [ + 409, + 135 + ], + [ + 428, + 164 + ], + [ + 382, + 150 + ], + [ + 439, + 119 + ], + [ + 363, + 123 + ], + [ + 419, + 89 + ] + ], + [ + [ + 428, + 151 + ], + [ + 437, + 131 + ], + [ + 417, + 176 + ], + [ + 439, + 128 + ], + [ + 410, + 192 + ], + [ + 431, + 146 + ] + ] + ] + }, + { + "task_id": "0004597", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0006_cfg_0001_039422060546#934", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004597/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 435, + 157 + ], + [ + 460, + 145 + ], + [ + 447, + 188 + ], + [ + 423, + 125 + ], + [ + 424, + 197 + ], + [ + 401, + 137 + ] + ], + [ + [ + 450, + 182 + ], + [ + 480, + 189 + ], + [ + 438, + 197 + ], + [ + 464, + 164 + ], + [ + 410, + 190 + ], + [ + 431, + 156 + ] + ], + [ + [ + 420, + 173 + ], + [ + 404, + 190 + ], + [ + 447, + 190 + ], + [ + 394, + 156 + ], + [ + 461, + 173 + ], + [ + 409, + 141 + ] + ], + [ + [ + 427, + 160 + ], + [ + 436, + 142 + ], + [ + 416, + 184 + ], + [ + 438, + 137 + ], + [ + 409, + 199 + ], + [ + 430, + 154 + ] + ] + ] + }, + { + "task_id": "0004599", + "id": "RH20T_cfg1_task_0005_user_0008_scene_0006_cfg_0001_819112070044#152", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004599/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 432, + 176 + ], + [ + 415, + 168 + ], + [ + 423, + 192 + ], + [ + 441, + 160 + ], + [ + 441, + 201 + ], + [ + 459, + 168 + ] + ], + [ + [ + 415, + 170 + ], + [ + 424, + 154 + ], + [ + 427, + 167 + ], + [ + 403, + 173 + ], + [ + 419, + 182 + ], + [ + 396, + 188 + ] + ], + [ + [ + 424, + 185 + ], + [ + 414, + 184 + ], + [ + 435, + 199 + ], + [ + 412, + 171 + ], + [ + 445, + 201 + ], + [ + 422, + 172 + ] + ], + [ + [ + 399, + 179 + ], + [ + 415, + 187 + ], + [ + 396, + 194 + ], + [ + 402, + 162 + ], + [ + 381, + 186 + ], + [ + 387, + 154 + ] + ] + ] + }, + { + "task_id": "0004601", + "id": "RH20T_cfg1_task_0096_user_0015_scene_0005_cfg_0001_035622060973#295", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004601/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 221, + 190 + ], + [ + 216, + 177 + ], + [ + 205, + 176 + ], + [ + 237, + 205 + ], + [ + 210, + 189 + ], + [ + 240, + 217 + ] + ], + [ + [ + 208, + 216 + ], + [ + 212, + 240 + ], + [ + 190, + 219 + ], + [ + 225, + 214 + ], + [ + 186, + 194 + ], + [ + 221, + 191 + ] + ], + [ + [ + 217, + 189 + ], + [ + 199, + 206 + ], + [ + 233, + 198 + ], + [ + 200, + 180 + ], + [ + 250, + 182 + ], + [ + 219, + 162 + ] + ], + [ + [ + 210, + 210 + ], + [ + 190, + 226 + ], + [ + 215, + 222 + ], + [ + 204, + 198 + ], + [ + 232, + 207 + ], + [ + 224, + 182 + ] + ] + ] + }, + { + "task_id": "0004603", + "id": "RH20T_cfg4_task_0035_user_0017_scene_0005_cfg_0004_036422060909#114", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up one small block'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004603/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 310, + 201 + ], + [ + 328, + 189 + ], + [ + 307, + 178 + ], + [ + 314, + 225 + ], + [ + 291, + 189 + ], + [ + 297, + 236 + ] + ], + [ + [ + 319, + 171 + ], + [ + 319, + 154 + ], + [ + 328, + 152 + ], + [ + 310, + 192 + ], + [ + 327, + 168 + ], + [ + 310, + 208 + ] + ], + [ + [ + 337, + 183 + ], + [ + 365, + 175 + ], + [ + 341, + 178 + ], + [ + 333, + 188 + ], + [ + 310, + 186 + ], + [ + 306, + 196 + ] + ], + [ + [ + 350, + 166 + ], + [ + 359, + 165 + ], + [ + 362, + 190 + ], + [ + 338, + 142 + ], + [ + 353, + 189 + ], + [ + 330, + 144 + ] + ] + ] + }, + { + "task_id": "0004605", + "id": "RH20T_cfg7_task_0013_user_0020_scene_0010_cfg_0007_104122064161#168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004605/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 404, + 220 + ], + [ + 395, + 199 + ], + [ + 394, + 232 + ], + [ + 415, + 206 + ], + [ + 403, + 254 + ], + [ + 426, + 228 + ] + ], + [ + [ + 436, + 251 + ], + [ + 464, + 259 + ], + [ + 431, + 274 + ], + [ + 442, + 226 + ], + [ + 405, + 267 + ], + [ + 415, + 219 + ] + ], + [ + [ + 419, + 238 + ], + [ + 395, + 234 + ], + [ + 425, + 230 + ], + [ + 412, + 246 + ], + [ + 448, + 233 + ], + [ + 437, + 250 + ] + ], + [ + [ + 409, + 239 + ], + [ + 413, + 223 + ], + [ + 411, + 260 + ], + [ + 407, + 219 + ], + [ + 406, + 274 + ], + [ + 403, + 234 + ] + ] + ] + }, + { + "task_id": "0004608", + "id": "RH20T_cfg3_task_0073_user_0010_scene_0010_cfg_0003_038522062288#225", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004608/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 245, + 208 + ], + [ + 276, + 204 + ], + [ + 252, + 235 + ], + [ + 239, + 179 + ], + [ + 222, + 238 + ], + [ + 207, + 183 + ] + ], + [ + [ + 242, + 229 + ], + [ + 235, + 255 + ], + [ + 253, + 246 + ], + [ + 229, + 211 + ], + [ + 258, + 222 + ], + [ + 235, + 187 + ] + ], + [ + [ + 235, + 170 + ], + [ + 218, + 195 + ], + [ + 209, + 152 + ], + [ + 259, + 187 + ], + [ + 227, + 127 + ], + [ + 275, + 163 + ] + ], + [ + [ + 224, + 211 + ], + [ + 211, + 191 + ], + [ + 196, + 213 + ], + [ + 252, + 208 + ], + [ + 208, + 230 + ], + [ + 262, + 226 + ] + ] + ] + }, + { + "task_id": "0004609", + "id": "RH20T_cfg3_task_0073_user_0010_scene_0010_cfg_0003_038522062288#255", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004609/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 236, + 222 + ], + [ + 225, + 204 + ], + [ + 208, + 222 + ], + [ + 265, + 222 + ], + [ + 220, + 238 + ], + [ + 274, + 238 + ] + ], + [ + [ + 233, + 229 + ], + [ + 223, + 205 + ], + [ + 235, + 247 + ], + [ + 232, + 212 + ], + [ + 245, + 269 + ], + [ + 241, + 233 + ] + ], + [ + [ + 231, + 222 + ], + [ + 234, + 243 + ], + [ + 241, + 202 + ], + [ + 222, + 240 + ], + [ + 237, + 182 + ], + [ + 220, + 220 + ] + ], + [ + [ + 227, + 215 + ], + [ + 215, + 234 + ], + [ + 251, + 202 + ], + [ + 204, + 228 + ], + [ + 260, + 187 + ], + [ + 215, + 211 + ] + ] + ] + }, + { + "task_id": "0004613", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_105422061350#163", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004613/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 254, + 18 + ], + [ + 266, + 28 + ], + [ + 229, + 7 + ], + [ + 279, + 30 + ], + [ + 214, + 0 + ], + [ + 266, + 20 + ] + ], + [ + [ + 240, + 38 + ], + [ + 239, + 13 + ], + [ + 263, + 24 + ], + [ + 217, + 52 + ], + [ + 263, + 46 + ], + [ + 219, + 74 + ] + ], + [ + [ + 271, + 48 + ], + [ + 244, + 42 + ], + [ + 260, + 55 + ], + [ + 281, + 41 + ], + [ + 290, + 61 + ], + [ + 308, + 46 + ] + ], + [ + [ + 238, + 35 + ], + [ + 229, + 38 + ], + [ + 216, + 14 + ], + [ + 258, + 55 + ], + [ + 224, + 9 + ], + [ + 269, + 53 + ] + ] + ] + }, + { + "task_id": "0004616", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0009_cfg_0005_105422061350#513", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004616/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 222, + 53 + ], + [ + 220, + 29 + ], + [ + 241, + 36 + ], + [ + 202, + 71 + ], + [ + 243, + 57 + ], + [ + 205, + 91 + ] + ], + [ + [ + 230, + 81 + ], + [ + 213, + 60 + ], + [ + 212, + 101 + ], + [ + 249, + 60 + ], + [ + 229, + 123 + ], + [ + 267, + 82 + ] + ], + [ + [ + 231, + 58 + ], + [ + 207, + 41 + ], + [ + 214, + 64 + ], + [ + 249, + 52 + ], + [ + 235, + 78 + ], + [ + 271, + 68 + ] + ], + [ + [ + 223, + 24 + ], + [ + 233, + 0 + ], + [ + 249, + 28 + ], + [ + 195, + 20 + ], + [ + 239, + 53 + ], + [ + 186, + 47 + ] + ] + ] + }, + { + "task_id": "0004617", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061330#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004617/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 256 + ], + [ + 282, + 243 + ], + [ + 318, + 262 + ], + [ + 293, + 250 + ], + [ + 342, + 276 + ], + [ + 319, + 265 + ] + ], + [ + [ + 334, + 252 + ], + [ + 332, + 237 + ], + [ + 321, + 278 + ], + [ + 345, + 228 + ], + [ + 323, + 291 + ], + [ + 346, + 243 + ] + ], + [ + [ + 327, + 252 + ], + [ + 298, + 239 + ], + [ + 314, + 280 + ], + [ + 340, + 224 + ], + [ + 342, + 293 + ], + [ + 368, + 237 + ] + ], + [ + [ + 328, + 248 + ], + [ + 346, + 270 + ], + [ + 308, + 261 + ], + [ + 346, + 236 + ], + [ + 289, + 238 + ], + [ + 329, + 215 + ] + ] + ] + }, + { + "task_id": "0004618", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061330#256", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004618/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 230 + ], + [ + 363, + 213 + ], + [ + 352, + 253 + ], + [ + 373, + 208 + ], + [ + 352, + 268 + ], + [ + 372, + 224 + ] + ], + [ + [ + 352, + 222 + ], + [ + 356, + 242 + ], + [ + 376, + 230 + ], + [ + 329, + 215 + ], + [ + 372, + 208 + ], + [ + 324, + 194 + ] + ], + [ + [ + 342, + 219 + ], + [ + 317, + 205 + ], + [ + 352, + 200 + ], + [ + 333, + 235 + ], + [ + 378, + 214 + ], + [ + 357, + 248 + ] + ], + [ + [ + 383, + 247 + ], + [ + 354, + 236 + ], + [ + 394, + 224 + ], + [ + 374, + 267 + ], + [ + 425, + 235 + ], + [ + 403, + 278 + ] + ] + ] + }, + { + "task_id": "0004621", + "id": "RH20T_cfg7_task_0066_user_0014_scene_0005_cfg_0007_104122061330#645", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004621/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 375, + 196 + ], + [ + 379, + 221 + ], + [ + 401, + 192 + ], + [ + 350, + 199 + ], + [ + 396, + 168 + ], + [ + 346, + 174 + ] + ], + [ + [ + 399, + 175 + ], + [ + 398, + 162 + ], + [ + 389, + 197 + ], + [ + 409, + 153 + ], + [ + 390, + 208 + ], + [ + 409, + 165 + ] + ], + [ + [ + 395, + 184 + ], + [ + 414, + 203 + ], + [ + 390, + 176 + ], + [ + 400, + 193 + ], + [ + 371, + 158 + ], + [ + 379, + 172 + ] + ], + [ + [ + 392, + 161 + ], + [ + 408, + 179 + ], + [ + 374, + 173 + ], + [ + 412, + 149 + ], + [ + 358, + 156 + ], + [ + 395, + 131 + ] + ] + ] + }, + { + "task_id": "0004622", + "id": "RH20T_cfg6_task_0015_user_0014_scene_0005_cfg_0006_104122061018#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004622/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 248, + 163 + ], + [ + 219, + 169 + ], + [ + 247, + 158 + ], + [ + 249, + 168 + ], + [ + 273, + 152 + ], + [ + 279, + 161 + ] + ], + [ + [ + 245, + 133 + ], + [ + 256, + 107 + ], + [ + 243, + 128 + ], + [ + 246, + 140 + ], + [ + 232, + 152 + ], + [ + 234, + 167 + ] + ], + [ + [ + 285, + 146 + ], + [ + 306, + 135 + ], + [ + 266, + 125 + ], + [ + 304, + 168 + ], + [ + 243, + 136 + ], + [ + 282, + 181 + ] + ], + [ + [ + 259, + 132 + ], + [ + 248, + 105 + ], + [ + 255, + 125 + ], + [ + 263, + 140 + ], + [ + 264, + 150 + ], + [ + 274, + 167 + ] + ] + ] + }, + { + "task_id": "0004626", + "id": "RH20T_cfg7_task_0063_user_0020_scene_0008_cfg_0007_104122060811#161", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a shovel to scoop up an object'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004626/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 337, + 159 + ], + [ + 337, + 173 + ], + [ + 330, + 163 + ], + [ + 344, + 154 + ], + [ + 330, + 148 + ], + [ + 343, + 140 + ] + ], + [ + [ + 329, + 140 + ], + [ + 319, + 131 + ], + [ + 325, + 132 + ], + [ + 333, + 148 + ], + [ + 334, + 140 + ], + [ + 342, + 156 + ] + ], + [ + [ + 326, + 154 + ], + [ + 341, + 149 + ], + [ + 330, + 158 + ], + [ + 323, + 149 + ], + [ + 316, + 162 + ], + [ + 308, + 154 + ] + ], + [ + [ + 328, + 141 + ], + [ + 328, + 151 + ], + [ + 323, + 130 + ], + [ + 333, + 152 + ], + [ + 323, + 121 + ], + [ + 333, + 142 + ] + ] + ] + }, + { + "task_id": "0004627", + "id": "RH20T_cfg6_task_0006_user_0014_scene_0005_cfg_0006_037522061512#171", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004627/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 273, + 113 + ], + [ + 252, + 118 + ], + [ + 269, + 93 + ], + [ + 277, + 133 + ], + [ + 289, + 87 + ], + [ + 298, + 127 + ] + ], + [ + [ + 289, + 103 + ], + [ + 307, + 98 + ], + [ + 279, + 101 + ], + [ + 298, + 105 + ], + [ + 260, + 106 + ], + [ + 278, + 110 + ] + ], + [ + [ + 270, + 104 + ], + [ + 282, + 112 + ], + [ + 268, + 123 + ], + [ + 271, + 85 + ], + [ + 254, + 115 + ], + [ + 258, + 75 + ] + ], + [ + [ + 275, + 113 + ], + [ + 273, + 94 + ], + [ + 288, + 103 + ], + [ + 261, + 123 + ], + [ + 290, + 121 + ], + [ + 263, + 141 + ] + ] + ] + }, + { + "task_id": "0004628", + "id": "RH20T_cfg6_task_0039_user_0014_scene_0003_cfg_0006_037522061512#168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004628/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 247, + 154 + ], + [ + 248, + 163 + ], + [ + 219, + 148 + ], + [ + 274, + 161 + ], + [ + 217, + 139 + ], + [ + 275, + 152 + ] + ], + [ + [ + 219, + 153 + ], + [ + 200, + 171 + ], + [ + 208, + 140 + ], + [ + 231, + 166 + ], + [ + 226, + 122 + ], + [ + 250, + 147 + ] + ], + [ + [ + 252, + 169 + ], + [ + 234, + 180 + ], + [ + 243, + 146 + ], + [ + 261, + 193 + ], + [ + 261, + 135 + ], + [ + 280, + 183 + ] + ], + [ + [ + 233, + 157 + ], + [ + 229, + 138 + ], + [ + 248, + 141 + ], + [ + 217, + 175 + ], + [ + 251, + 158 + ], + [ + 222, + 193 + ] + ] + ] + }, + { + "task_id": "0004629", + "id": "RH20T_cfg6_task_0125_user_0020_scene_0004_cfg_0006_104122060811#519", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull up the lever'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004629/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 287, + 89 + ], + [ + 294, + 77 + ], + [ + 290, + 97 + ], + [ + 284, + 81 + ], + [ + 282, + 109 + ], + [ + 277, + 92 + ] + ], + [ + [ + 302, + 100 + ], + [ + 305, + 92 + ], + [ + 315, + 97 + ], + [ + 288, + 104 + ], + [ + 312, + 105 + ], + [ + 286, + 112 + ] + ], + [ + [ + 297, + 90 + ], + [ + 293, + 78 + ], + [ + 300, + 81 + ], + [ + 293, + 99 + ], + [ + 304, + 94 + ], + [ + 297, + 111 + ] + ], + [ + [ + 299, + 111 + ], + [ + 299, + 99 + ], + [ + 311, + 105 + ], + [ + 288, + 118 + ], + [ + 310, + 116 + ], + [ + 288, + 129 + ] + ] + ] + }, + { + "task_id": "0004630", + "id": "RH20T_cfg6_task_0127_user_0020_scene_0006_cfg_0006_104122060811#262", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004630/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 308, + 120 + ], + [ + 320, + 121 + ], + [ + 303, + 110 + ], + [ + 313, + 129 + ], + [ + 291, + 110 + ], + [ + 302, + 128 + ] + ], + [ + [ + 321, + 117 + ], + [ + 319, + 111 + ], + [ + 334, + 110 + ], + [ + 308, + 123 + ], + [ + 336, + 116 + ], + [ + 311, + 128 + ] + ], + [ + [ + 322, + 125 + ], + [ + 331, + 137 + ], + [ + 334, + 116 + ], + [ + 311, + 134 + ], + [ + 325, + 104 + ], + [ + 302, + 123 + ] + ], + [ + [ + 330, + 115 + ], + [ + 336, + 103 + ], + [ + 332, + 111 + ], + [ + 327, + 119 + ], + [ + 327, + 122 + ], + [ + 321, + 131 + ] + ] + ] + }, + { + "task_id": "0004631", + "id": "RH20T_cfg4_task_0110_user_0007_scene_0009_cfg_0004_038522062288#104", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the pencil out from the pencil sharpener'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004631/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 273, + 213 + ], + [ + 270, + 196 + ], + [ + 241, + 209 + ], + [ + 305, + 218 + ], + [ + 246, + 225 + ], + [ + 306, + 234 + ] + ], + [ + [ + 280, + 252 + ], + [ + 283, + 222 + ], + [ + 247, + 250 + ], + [ + 312, + 255 + ], + [ + 243, + 282 + ], + [ + 309, + 287 + ] + ], + [ + [ + 294, + 198 + ], + [ + 310, + 214 + ], + [ + 313, + 208 + ], + [ + 273, + 187 + ], + [ + 298, + 194 + ], + [ + 258, + 173 + ] + ], + [ + [ + 266, + 202 + ], + [ + 281, + 227 + ], + [ + 240, + 218 + ], + [ + 292, + 186 + ], + [ + 227, + 194 + ], + [ + 278, + 162 + ] + ] + ] + }, + { + "task_id": "0004632", + "id": "RH20T_cfg6_task_0046_user_0020_scene_0004_cfg_0006_104122064161#395", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004632/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 365, + 135 + ], + [ + 349, + 134 + ], + [ + 384, + 139 + ], + [ + 345, + 130 + ], + [ + 402, + 140 + ], + [ + 363, + 130 + ] + ], + [ + [ + 371, + 151 + ], + [ + 369, + 176 + ], + [ + 365, + 156 + ], + [ + 378, + 145 + ], + [ + 367, + 132 + ], + [ + 380, + 119 + ] + ], + [ + [ + 363, + 122 + ], + [ + 360, + 146 + ], + [ + 336, + 120 + ], + [ + 389, + 124 + ], + [ + 339, + 94 + ], + [ + 393, + 98 + ] + ], + [ + [ + 393, + 166 + ], + [ + 416, + 155 + ], + [ + 382, + 142 + ], + [ + 403, + 190 + ], + [ + 359, + 153 + ], + [ + 380, + 201 + ] + ] + ] + }, + { + "task_id": "0004633", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104122062823#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004633/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 295, + 16 + ], + [ + 315, + 0 + ], + [ + 324, + 23 + ], + [ + 270, + 9 + ], + [ + 303, + 63 + ], + [ + 252, + 44 + ] + ], + [ + [ + 277, + 6 + ], + [ + 314, + 1 + ], + [ + 274, + 0 + ], + [ + 281, + 45 + ], + [ + 238, + 0 + ], + [ + 246, + 49 + ] + ], + [ + [ + 264, + 2 + ], + [ + 233, + 4 + ], + [ + 293, + 4 + ], + [ + 238, + 0 + ], + [ + 321, + 2 + ], + [ + 266, + 0 + ] + ], + [ + [ + 271, + 23 + ], + [ + 268, + 0 + ], + [ + 303, + 33 + ], + [ + 240, + 14 + ], + [ + 304, + 69 + ], + [ + 244, + 49 + ] + ] + ] + }, + { + "task_id": "0004634", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104122062823#159", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004634/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 12 + ], + [ + 251, + 0 + ], + [ + 290, + 0 + ], + [ + 260, + 43 + ], + [ + 315, + 0 + ], + [ + 282, + 63 + ] + ], + [ + [ + 292, + 27 + ], + [ + 288, + 0 + ], + [ + 325, + 31 + ], + [ + 260, + 23 + ], + [ + 327, + 67 + ], + [ + 265, + 57 + ] + ], + [ + [ + 308, + 42 + ], + [ + 286, + 63 + ], + [ + 311, + 13 + ], + [ + 305, + 68 + ], + [ + 332, + 0 + ], + [ + 325, + 49 + ] + ], + [ + [ + 296, + 0 + ], + [ + 297, + 0 + ], + [ + 324, + 0 + ], + [ + 270, + 11 + ], + [ + 325, + 0 + ], + [ + 268, + 30 + ] + ] + ] + }, + { + "task_id": "0004635", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0009_cfg_0004_104122062823#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004635/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 22 + ], + [ + 310, + 12 + ], + [ + 258, + 0 + ], + [ + 290, + 59 + ], + [ + 220, + 0 + ], + [ + 252, + 71 + ] + ], + [ + [ + 256, + 54 + ], + [ + 259, + 14 + ], + [ + 291, + 64 + ], + [ + 222, + 45 + ], + [ + 287, + 101 + ], + [ + 221, + 82 + ] + ], + [ + [ + 276, + 45 + ], + [ + 246, + 69 + ], + [ + 256, + 14 + ], + [ + 296, + 77 + ], + [ + 287, + 0 + ], + [ + 328, + 52 + ] + ], + [ + [ + 251, + 57 + ], + [ + 273, + 78 + ], + [ + 269, + 75 + ], + [ + 231, + 36 + ], + [ + 249, + 56 + ], + [ + 210, + 16 + ] + ] + ] + }, + { + "task_id": "0004638", + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_750612070853#112", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004638/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 348, + 46 + ], + [ + 371, + 51 + ], + [ + 355, + 87 + ], + [ + 339, + 3 + ], + [ + 335, + 80 + ], + [ + 319, + 2 + ] + ], + [ + [ + 293, + 61 + ], + [ + 294, + 92 + ], + [ + 277, + 82 + ], + [ + 310, + 37 + ], + [ + 276, + 54 + ], + [ + 308, + 6 + ] + ], + [ + [ + 312, + 50 + ], + [ + 305, + 11 + ], + [ + 271, + 50 + ], + [ + 354, + 50 + ], + [ + 279, + 85 + ], + [ + 357, + 86 + ] + ], + [ + [ + 321, + 63 + ], + [ + 289, + 91 + ], + [ + 319, + 59 + ], + [ + 323, + 67 + ], + [ + 350, + 32 + ], + [ + 359, + 35 + ] + ] + ] + }, + { + "task_id": "0004639", + "id": "RH20T_cfg1_task_0094_user_0014_scene_0010_cfg_0001_750612070853#152", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004639/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 34 + ], + [ + 334, + 69 + ], + [ + 372, + 66 + ], + [ + 319, + 0 + ], + [ + 382, + 35 + ], + [ + 333, + 0 + ] + ], + [ + [ + 370, + 12 + ], + [ + 394, + 23 + ], + [ + 405, + 15 + ], + [ + 331, + 9 + ], + [ + 383, + 6 + ], + [ + 312, + 0 + ] + ], + [ + [ + 360, + 51 + ], + [ + 350, + 11 + ], + [ + 322, + 53 + ], + [ + 401, + 48 + ], + [ + 333, + 86 + ], + [ + 408, + 84 + ] + ], + [ + [ + 398, + 80 + ], + [ + 396, + 37 + ], + [ + 375, + 92 + ], + [ + 419, + 69 + ], + [ + 378, + 134 + ], + [ + 420, + 107 + ] + ] + ] + }, + { + "task_id": "0004642", + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_038522062547#217", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004642/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 372, + 306 + ], + [ + 373, + 301 + ], + [ + 397, + 325 + ], + [ + 349, + 288 + ], + [ + 395, + 328 + ], + [ + 350, + 294 + ] + ], + [ + [ + 376, + 277 + ], + [ + 363, + 254 + ], + [ + 390, + 279 + ], + [ + 360, + 276 + ], + [ + 404, + 302 + ], + [ + 375, + 302 + ] + ], + [ + [ + 361, + 309 + ], + [ + 384, + 293 + ], + [ + 348, + 288 + ], + [ + 373, + 330 + ], + [ + 324, + 304 + ], + [ + 350, + 346 + ] + ], + [ + [ + 387, + 326 + ], + [ + 409, + 346 + ], + [ + 372, + 348 + ], + [ + 403, + 304 + ], + [ + 353, + 329 + ], + [ + 382, + 286 + ] + ] + ] + }, + { + "task_id": "0004643", + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_038522062547#250", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004643/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 386, + 319 + ], + [ + 363, + 298 + ], + [ + 389, + 317 + ], + [ + 383, + 321 + ], + [ + 412, + 337 + ], + [ + 409, + 344 + ] + ], + [ + [ + 386, + 285 + ], + [ + 388, + 278 + ], + [ + 410, + 304 + ], + [ + 363, + 266 + ], + [ + 406, + 309 + ], + [ + 362, + 273 + ] + ], + [ + [ + 377, + 279 + ], + [ + 358, + 283 + ], + [ + 393, + 307 + ], + [ + 363, + 253 + ], + [ + 410, + 303 + ], + [ + 380, + 252 + ] + ], + [ + [ + 397, + 301 + ], + [ + 396, + 312 + ], + [ + 426, + 317 + ], + [ + 370, + 285 + ], + [ + 425, + 306 + ], + [ + 373, + 276 + ] + ] + ] + }, + { + "task_id": "0004646", + "id": "RH20T_cfg6_task_0032_user_0020_scene_0009_cfg_0006_104122061602#212", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004646/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 556, + 270 + ], + [ + 585, + 255 + ], + [ + 576, + 298 + ], + [ + 537, + 244 + ], + [ + 547, + 312 + ], + [ + 511, + 258 + ] + ], + [ + [ + 537, + 258 + ], + [ + 548, + 241 + ], + [ + 555, + 283 + ], + [ + 521, + 235 + ], + [ + 544, + 298 + ], + [ + 512, + 251 + ] + ], + [ + [ + 549, + 267 + ], + [ + 583, + 268 + ], + [ + 552, + 294 + ], + [ + 545, + 240 + ], + [ + 521, + 291 + ], + [ + 513, + 239 + ] + ], + [ + [ + 520, + 242 + ], + [ + 539, + 262 + ], + [ + 528, + 257 + ], + [ + 513, + 226 + ], + [ + 512, + 238 + ], + [ + 496, + 208 + ] + ] + ] + }, + { + "task_id": "0004647", + "id": "RH20T_cfg6_task_0209_user_0014_scene_0004_cfg_0006_104122060811#68", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slice the carrots'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004647/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 302, + 149 + ], + [ + 302, + 139 + ], + [ + 318, + 148 + ], + [ + 287, + 150 + ], + [ + 318, + 158 + ], + [ + 287, + 160 + ] + ], + [ + [ + 289, + 153 + ], + [ + 277, + 163 + ], + [ + 299, + 165 + ], + [ + 280, + 141 + ], + [ + 311, + 155 + ], + [ + 292, + 132 + ] + ], + [ + [ + 289, + 142 + ], + [ + 303, + 135 + ], + [ + 287, + 137 + ], + [ + 290, + 147 + ], + [ + 273, + 145 + ], + [ + 276, + 155 + ] + ], + [ + [ + 306, + 149 + ], + [ + 298, + 163 + ], + [ + 295, + 140 + ], + [ + 316, + 158 + ], + [ + 303, + 127 + ], + [ + 323, + 145 + ] + ] + ] + }, + { + "task_id": "0004649", + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122061602#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004649/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 484, + 197 + ], + [ + 496, + 199 + ], + [ + 505, + 213 + ], + [ + 465, + 182 + ], + [ + 491, + 211 + ], + [ + 451, + 180 + ] + ], + [ + [ + 439, + 188 + ], + [ + 420, + 188 + ], + [ + 421, + 174 + ], + [ + 458, + 203 + ], + [ + 439, + 175 + ], + [ + 476, + 203 + ] + ], + [ + [ + 441, + 213 + ], + [ + 443, + 193 + ], + [ + 447, + 197 + ], + [ + 434, + 229 + ], + [ + 445, + 215 + ], + [ + 433, + 247 + ] + ], + [ + [ + 461, + 205 + ], + [ + 476, + 200 + ], + [ + 470, + 231 + ], + [ + 451, + 181 + ], + [ + 455, + 235 + ], + [ + 438, + 187 + ] + ] + ] + }, + { + "task_id": "0004650", + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122061602#194", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004650/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 440, + 202 + ], + [ + 456, + 196 + ], + [ + 449, + 227 + ], + [ + 431, + 177 + ], + [ + 434, + 231 + ], + [ + 418, + 183 + ] + ], + [ + [ + 463, + 168 + ], + [ + 481, + 150 + ], + [ + 459, + 153 + ], + [ + 468, + 182 + ], + [ + 440, + 172 + ], + [ + 451, + 200 + ] + ], + [ + [ + 436, + 203 + ], + [ + 445, + 199 + ], + [ + 461, + 196 + ], + [ + 412, + 211 + ], + [ + 452, + 200 + ], + [ + 406, + 214 + ] + ], + [ + [ + 452, + 175 + ], + [ + 437, + 159 + ], + [ + 472, + 174 + ], + [ + 434, + 175 + ], + [ + 487, + 191 + ], + [ + 448, + 190 + ] + ] + ] + }, + { + "task_id": "0004651", + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122061602#235", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004651/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 413, + 219 + ], + [ + 413, + 212 + ], + [ + 424, + 196 + ], + [ + 402, + 244 + ], + [ + 424, + 203 + ], + [ + 403, + 249 + ] + ], + [ + [ + 419, + 199 + ], + [ + 438, + 199 + ], + [ + 398, + 187 + ], + [ + 439, + 211 + ], + [ + 382, + 187 + ], + [ + 421, + 210 + ] + ], + [ + [ + 416, + 197 + ], + [ + 433, + 192 + ], + [ + 424, + 222 + ], + [ + 409, + 172 + ], + [ + 409, + 226 + ], + [ + 394, + 178 + ] + ], + [ + [ + 430, + 181 + ], + [ + 412, + 195 + ], + [ + 412, + 183 + ], + [ + 450, + 180 + ], + [ + 429, + 171 + ], + [ + 467, + 167 + ] + ] + ] + }, + { + "task_id": "0004652", + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122061602#256", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004652/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 397, + 194 + ], + [ + 412, + 190 + ], + [ + 404, + 219 + ], + [ + 390, + 169 + ], + [ + 390, + 222 + ], + [ + 376, + 174 + ] + ], + [ + [ + 398, + 185 + ], + [ + 387, + 190 + ], + [ + 416, + 202 + ], + [ + 380, + 168 + ], + [ + 429, + 197 + ], + [ + 391, + 162 + ] + ], + [ + [ + 397, + 202 + ], + [ + 418, + 196 + ], + [ + 388, + 214 + ], + [ + 407, + 189 + ], + [ + 368, + 220 + ], + [ + 385, + 195 + ] + ], + [ + [ + 376, + 177 + ], + [ + 355, + 185 + ], + [ + 390, + 172 + ], + [ + 364, + 182 + ], + [ + 411, + 164 + ], + [ + 384, + 174 + ] + ] + ] + }, + { + "task_id": "0004653", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#378", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004653/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 493, + 151 + ], + [ + 471, + 145 + ], + [ + 498, + 159 + ], + [ + 488, + 143 + ], + [ + 520, + 165 + ], + [ + 512, + 150 + ] + ], + [ + [ + 496, + 152 + ], + [ + 516, + 159 + ], + [ + 488, + 174 + ], + [ + 503, + 130 + ], + [ + 467, + 167 + ], + [ + 482, + 123 + ] + ], + [ + [ + 464, + 164 + ], + [ + 473, + 144 + ], + [ + 453, + 154 + ], + [ + 477, + 176 + ], + [ + 444, + 173 + ], + [ + 468, + 196 + ] + ], + [ + [ + 487, + 159 + ], + [ + 495, + 142 + ], + [ + 506, + 172 + ], + [ + 469, + 147 + ], + [ + 498, + 188 + ], + [ + 462, + 163 + ] + ] + ] + }, + { + "task_id": "0004655", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#915", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004655/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 466, + 143 + ], + [ + 472, + 126 + ], + [ + 483, + 156 + ], + [ + 449, + 131 + ], + [ + 477, + 171 + ], + [ + 444, + 147 + ] + ], + [ + [ + 458, + 159 + ], + [ + 444, + 174 + ], + [ + 464, + 150 + ], + [ + 452, + 167 + ], + [ + 477, + 135 + ], + [ + 465, + 154 + ] + ], + [ + [ + 466, + 145 + ], + [ + 466, + 163 + ], + [ + 470, + 134 + ], + [ + 461, + 158 + ], + [ + 470, + 115 + ], + [ + 460, + 139 + ] + ], + [ + [ + 453, + 134 + ], + [ + 448, + 142 + ], + [ + 455, + 155 + ], + [ + 451, + 112 + ], + [ + 460, + 148 + ], + [ + 456, + 103 + ] + ] + ] + }, + { + "task_id": "0004659", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#1447", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004659/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 428, + 142 + ], + [ + 448, + 152 + ], + [ + 430, + 157 + ], + [ + 426, + 126 + ], + [ + 412, + 147 + ], + [ + 406, + 117 + ] + ], + [ + [ + 442, + 165 + ], + [ + 454, + 154 + ], + [ + 458, + 181 + ], + [ + 426, + 149 + ], + [ + 445, + 194 + ], + [ + 413, + 161 + ] + ], + [ + [ + 451, + 155 + ], + [ + 458, + 137 + ], + [ + 471, + 168 + ], + [ + 432, + 144 + ], + [ + 463, + 185 + ], + [ + 426, + 161 + ] + ], + [ + [ + 432, + 158 + ], + [ + 420, + 168 + ], + [ + 452, + 168 + ], + [ + 412, + 148 + ], + [ + 462, + 158 + ], + [ + 423, + 139 + ] + ] + ] + }, + { + "task_id": "0004660", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#1523", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004660/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 443, + 132 + ], + [ + 453, + 123 + ], + [ + 443, + 150 + ], + [ + 443, + 114 + ], + [ + 433, + 159 + ], + [ + 432, + 124 + ] + ], + [ + [ + 442, + 123 + ], + [ + 455, + 141 + ], + [ + 456, + 115 + ], + [ + 428, + 132 + ], + [ + 444, + 97 + ], + [ + 415, + 113 + ] + ], + [ + [ + 434, + 125 + ], + [ + 429, + 144 + ], + [ + 415, + 120 + ], + [ + 454, + 130 + ], + [ + 421, + 100 + ], + [ + 460, + 110 + ] + ], + [ + [ + 434, + 124 + ], + [ + 438, + 106 + ], + [ + 451, + 134 + ], + [ + 417, + 114 + ], + [ + 447, + 150 + ], + [ + 414, + 130 + ] + ] + ] + }, + { + "task_id": "0004662", + "id": "RH20T_cfg1_task_0065_user_0010_scene_0005_cfg_0001_038522063145#1720", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004662/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 421, + 105 + ], + [ + 412, + 113 + ], + [ + 413, + 86 + ], + [ + 429, + 122 + ], + [ + 422, + 78 + ], + [ + 437, + 114 + ] + ], + [ + [ + 420, + 113 + ], + [ + 437, + 112 + ], + [ + 408, + 106 + ], + [ + 432, + 121 + ], + [ + 391, + 108 + ], + [ + 414, + 123 + ] + ], + [ + [ + 426, + 116 + ], + [ + 432, + 99 + ], + [ + 442, + 127 + ], + [ + 412, + 106 + ], + [ + 437, + 144 + ], + [ + 408, + 122 + ] + ], + [ + [ + 423, + 101 + ], + [ + 408, + 89 + ], + [ + 411, + 111 + ], + [ + 435, + 92 + ], + [ + 428, + 123 + ], + [ + 451, + 104 + ] + ] + ] + }, + { + "task_id": "0004665", + "id": "RH20T_cfg4_task_0012_user_0010_scene_0001_cfg_0004_038522062288#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push the soccer ball into the goal'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004665/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 272, + 215 + ], + [ + 275, + 197 + ], + [ + 280, + 240 + ], + [ + 264, + 192 + ], + [ + 276, + 256 + ], + [ + 261, + 209 + ] + ], + [ + [ + 290, + 203 + ], + [ + 277, + 193 + ], + [ + 263, + 203 + ], + [ + 318, + 204 + ], + [ + 275, + 213 + ], + [ + 327, + 214 + ] + ], + [ + [ + 258, + 178 + ], + [ + 248, + 169 + ], + [ + 252, + 207 + ], + [ + 264, + 149 + ], + [ + 263, + 219 + ], + [ + 276, + 158 + ] + ], + [ + [ + 279, + 191 + ], + [ + 289, + 202 + ], + [ + 263, + 171 + ], + [ + 294, + 211 + ], + [ + 255, + 161 + ], + [ + 284, + 200 + ] + ] + ] + }, + { + "task_id": "0004673", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0005_cfg_0003_104122062295#269", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004673/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 113, + 124 + ], + [ + 114, + 85 + ], + [ + 84, + 107 + ], + [ + 144, + 142 + ], + [ + 84, + 142 + ], + [ + 142, + 177 + ] + ], + [ + [ + 148, + 142 + ], + [ + 108, + 154 + ], + [ + 140, + 147 + ], + [ + 157, + 137 + ], + [ + 175, + 137 + ], + [ + 196, + 126 + ] + ], + [ + [ + 141, + 119 + ], + [ + 156, + 147 + ], + [ + 136, + 147 + ], + [ + 146, + 93 + ], + [ + 118, + 114 + ], + [ + 130, + 61 + ] + ], + [ + [ + 128, + 111 + ], + [ + 169, + 118 + ], + [ + 116, + 153 + ], + [ + 140, + 71 + ], + [ + 75, + 146 + ], + [ + 101, + 66 + ] + ] + ] + }, + { + "task_id": "0004674", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0005_cfg_0003_104122062295#303", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004674/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 199, + 203 + ], + [ + 204, + 167 + ], + [ + 173, + 176 + ], + [ + 227, + 232 + ], + [ + 169, + 207 + ], + [ + 221, + 263 + ] + ], + [ + [ + 207, + 170 + ], + [ + 186, + 153 + ], + [ + 253, + 164 + ], + [ + 162, + 176 + ], + [ + 267, + 178 + ], + [ + 184, + 189 + ] + ], + [ + [ + 217, + 157 + ], + [ + 195, + 167 + ], + [ + 236, + 120 + ], + [ + 199, + 191 + ], + [ + 253, + 115 + ], + [ + 219, + 180 + ] + ], + [ + [ + 257, + 199 + ], + [ + 272, + 176 + ], + [ + 299, + 219 + ], + [ + 215, + 179 + ], + [ + 286, + 248 + ], + [ + 193, + 204 + ] + ] + ] + }, + { + "task_id": "0004676", + "id": "RH20T_cfg6_task_0130_user_0020_scene_0002_cfg_0006_104122061602#297", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket and turn on the switch of the bulb'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004676/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 88 + ], + [ + 419, + 97 + ], + [ + 402, + 109 + ], + [ + 411, + 66 + ], + [ + 390, + 100 + ], + [ + 399, + 58 + ] + ], + [ + [ + 422, + 70 + ], + [ + 431, + 76 + ], + [ + 419, + 48 + ], + [ + 424, + 91 + ], + [ + 410, + 44 + ], + [ + 415, + 85 + ] + ], + [ + [ + 407, + 79 + ], + [ + 401, + 88 + ], + [ + 429, + 87 + ], + [ + 384, + 71 + ], + [ + 437, + 78 + ], + [ + 389, + 62 + ] + ], + [ + [ + 386, + 85 + ], + [ + 393, + 91 + ], + [ + 406, + 89 + ], + [ + 366, + 82 + ], + [ + 399, + 83 + ], + [ + 357, + 76 + ] + ] + ] + }, + { + "task_id": "0004677", + "id": "RH20T_cfg5_task_0002_user_0014_scene_0010_cfg_0005_104122061850#193", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004677/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 387, + 160 + ], + [ + 415, + 168 + ], + [ + 380, + 185 + ], + [ + 394, + 135 + ], + [ + 352, + 177 + ], + [ + 367, + 127 + ] + ], + [ + [ + 383, + 130 + ], + [ + 385, + 120 + ], + [ + 369, + 107 + ], + [ + 396, + 152 + ], + [ + 366, + 116 + ], + [ + 394, + 163 + ] + ], + [ + [ + 390, + 166 + ], + [ + 391, + 150 + ], + [ + 379, + 187 + ], + [ + 400, + 146 + ], + [ + 378, + 200 + ], + [ + 399, + 160 + ] + ], + [ + [ + 398, + 167 + ], + [ + 409, + 157 + ], + [ + 374, + 158 + ], + [ + 424, + 176 + ], + [ + 362, + 168 + ], + [ + 413, + 187 + ] + ] + ] + }, + { + "task_id": "0004679", + "id": "RH20T_cfg2_task_0011_user_0013_scene_0001_cfg_0002_037522062165#310", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004679/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 375, + 301 + ], + [ + 362, + 272 + ], + [ + 393, + 307 + ], + [ + 360, + 296 + ], + [ + 407, + 338 + ], + [ + 373, + 323 + ] + ], + [ + [ + 392, + 269 + ], + [ + 379, + 274 + ], + [ + 382, + 299 + ], + [ + 400, + 241 + ], + [ + 397, + 295 + ], + [ + 415, + 233 + ] + ], + [ + [ + 386, + 287 + ], + [ + 395, + 273 + ], + [ + 415, + 277 + ], + [ + 357, + 297 + ], + [ + 406, + 289 + ], + [ + 350, + 309 + ] + ], + [ + [ + 375, + 322 + ], + [ + 389, + 354 + ], + [ + 367, + 329 + ], + [ + 381, + 316 + ], + [ + 353, + 296 + ], + [ + 369, + 287 + ] + ] + ] + }, + { + "task_id": "0004681", + "id": "RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_819112071037#348", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004681/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 295, + 186 + ], + [ + 328, + 216 + ], + [ + 295, + 139 + ], + [ + 295, + 227 + ], + [ + 264, + 115 + ], + [ + 268, + 200 + ] + ], + [ + [ + 232, + 194 + ], + [ + 226, + 206 + ], + [ + 259, + 239 + ], + [ + 204, + 148 + ], + [ + 262, + 225 + ], + [ + 212, + 144 + ] + ], + [ + [ + 259, + 175 + ], + [ + 255, + 119 + ], + [ + 199, + 177 + ], + [ + 317, + 173 + ], + [ + 205, + 227 + ], + [ + 317, + 223 + ] + ], + [ + [ + 206, + 222 + ], + [ + 190, + 202 + ], + [ + 236, + 258 + ], + [ + 170, + 180 + ], + [ + 258, + 284 + ], + [ + 187, + 202 + ] + ] + ] + }, + { + "task_id": "0004682", + "id": "RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_819112071037#388", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004682/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 158, + 251 + ], + [ + 157, + 309 + ], + [ + 155, + 299 + ], + [ + 161, + 196 + ], + [ + 155, + 252 + ], + [ + 161, + 153 + ] + ], + [ + [ + 216, + 211 + ], + [ + 210, + 146 + ], + [ + 144, + 213 + ], + [ + 286, + 209 + ], + [ + 154, + 271 + ], + [ + 287, + 265 + ] + ], + [ + [ + 189, + 151 + ], + [ + 254, + 139 + ], + [ + 192, + 213 + ], + [ + 186, + 96 + ], + [ + 123, + 226 + ], + [ + 124, + 108 + ] + ], + [ + [ + 151, + 165 + ], + [ + 74, + 194 + ], + [ + 110, + 113 + ], + [ + 182, + 203 + ], + [ + 198, + 81 + ], + [ + 248, + 177 + ] + ] + ] + }, + { + "task_id": "0004684", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_035622060973#90", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004684/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 179, + 109 + ], + [ + 163, + 110 + ], + [ + 176, + 89 + ], + [ + 183, + 128 + ], + [ + 192, + 88 + ], + [ + 199, + 127 + ] + ], + [ + [ + 180, + 111 + ], + [ + 176, + 99 + ], + [ + 173, + 96 + ], + [ + 188, + 126 + ], + [ + 177, + 108 + ], + [ + 192, + 137 + ] + ], + [ + [ + 199, + 110 + ], + [ + 212, + 94 + ], + [ + 182, + 96 + ], + [ + 215, + 123 + ], + [ + 169, + 112 + ], + [ + 202, + 138 + ] + ], + [ + [ + 173, + 104 + ], + [ + 181, + 92 + ], + [ + 191, + 116 + ], + [ + 155, + 92 + ], + [ + 184, + 127 + ], + [ + 149, + 104 + ] + ] + ] + }, + { + "task_id": "0004686", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0005_cfg_0001_035622060973#376", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004686/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 197, + 136 + ], + [ + 177, + 143 + ], + [ + 191, + 133 + ], + [ + 204, + 140 + ], + [ + 210, + 127 + ], + [ + 224, + 134 + ] + ], + [ + [ + 192, + 149 + ], + [ + 189, + 138 + ], + [ + 185, + 133 + ], + [ + 200, + 165 + ], + [ + 188, + 143 + ], + [ + 203, + 175 + ] + ], + [ + [ + 175, + 158 + ], + [ + 160, + 169 + ], + [ + 162, + 141 + ], + [ + 188, + 175 + ], + [ + 178, + 129 + ], + [ + 204, + 164 + ] + ], + [ + [ + 187, + 155 + ], + [ + 172, + 166 + ], + [ + 201, + 164 + ], + [ + 172, + 145 + ], + [ + 215, + 153 + ], + [ + 187, + 133 + ] + ] + ] + }, + { + "task_id": "0004689", + "id": "RH20T_cfg5_task_0123_user_0010_scene_0004_cfg_0005_104122063678#259", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the knife on the knife rack'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004689/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 237, + 216 + ], + [ + 248, + 235 + ], + [ + 250, + 219 + ], + [ + 222, + 213 + ], + [ + 240, + 202 + ], + [ + 211, + 195 + ] + ], + [ + [ + 221, + 186 + ], + [ + 229, + 170 + ], + [ + 225, + 205 + ], + [ + 218, + 167 + ], + [ + 218, + 220 + ], + [ + 211, + 182 + ] + ], + [ + [ + 239, + 166 + ], + [ + 243, + 147 + ], + [ + 216, + 166 + ], + [ + 263, + 167 + ], + [ + 211, + 186 + ], + [ + 259, + 188 + ] + ], + [ + [ + 237, + 174 + ], + [ + 215, + 167 + ], + [ + 247, + 168 + ], + [ + 227, + 180 + ], + [ + 268, + 175 + ], + [ + 247, + 187 + ] + ] + ] + }, + { + "task_id": "0004690", + "id": "RH20T_cfg5_task_0123_user_0010_scene_0004_cfg_0005_104122063678#649", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the knife on the knife rack'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004690/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 249, + 209 + ], + [ + 252, + 184 + ], + [ + 243, + 221 + ], + [ + 254, + 198 + ], + [ + 240, + 246 + ], + [ + 251, + 220 + ] + ], + [ + [ + 229, + 187 + ], + [ + 251, + 176 + ], + [ + 224, + 163 + ], + [ + 234, + 211 + ], + [ + 203, + 174 + ], + [ + 211, + 222 + ] + ], + [ + [ + 249, + 191 + ], + [ + 259, + 168 + ], + [ + 225, + 179 + ], + [ + 274, + 202 + ], + [ + 216, + 201 + ], + [ + 264, + 223 + ] + ], + [ + [ + 220, + 208 + ], + [ + 223, + 208 + ], + [ + 229, + 236 + ], + [ + 212, + 181 + ], + [ + 226, + 235 + ], + [ + 210, + 184 + ] + ] + ] + }, + { + "task_id": "0004699", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0002_cfg_0001_819112070044#163", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004699/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 440, + 158 + ], + [ + 437, + 139 + ], + [ + 435, + 154 + ], + [ + 444, + 161 + ], + [ + 438, + 174 + ], + [ + 446, + 179 + ] + ], + [ + [ + 399, + 139 + ], + [ + 407, + 134 + ], + [ + 383, + 145 + ], + [ + 416, + 133 + ], + [ + 374, + 150 + ], + [ + 408, + 138 + ] + ], + [ + [ + 414, + 146 + ], + [ + 416, + 130 + ], + [ + 429, + 140 + ], + [ + 398, + 153 + ], + [ + 426, + 155 + ], + [ + 396, + 168 + ] + ], + [ + [ + 392, + 144 + ], + [ + 410, + 149 + ], + [ + 394, + 126 + ], + [ + 390, + 161 + ], + [ + 377, + 121 + ], + [ + 374, + 156 + ] + ] + ] + }, + { + "task_id": "0004700", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0002_cfg_0001_819112070044#266", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004700/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 318, + 186 + ], + [ + 318, + 169 + ], + [ + 336, + 179 + ], + [ + 299, + 193 + ], + [ + 335, + 194 + ], + [ + 299, + 208 + ] + ], + [ + [ + 325, + 183 + ], + [ + 329, + 199 + ], + [ + 305, + 181 + ], + [ + 344, + 185 + ], + [ + 302, + 166 + ], + [ + 340, + 170 + ] + ], + [ + [ + 340, + 186 + ], + [ + 328, + 180 + ], + [ + 322, + 186 + ], + [ + 357, + 186 + ], + [ + 335, + 193 + ], + [ + 370, + 192 + ] + ], + [ + [ + 334, + 171 + ], + [ + 325, + 152 + ], + [ + 347, + 158 + ], + [ + 322, + 182 + ], + [ + 358, + 177 + ], + [ + 332, + 200 + ] + ] + ] + }, + { + "task_id": "0004702", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_819112070044#186", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004702/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 446, + 158 + ], + [ + 458, + 169 + ], + [ + 430, + 164 + ], + [ + 462, + 152 + ], + [ + 419, + 154 + ], + [ + 450, + 142 + ] + ], + [ + [ + 417, + 172 + ], + [ + 411, + 187 + ], + [ + 403, + 174 + ], + [ + 431, + 170 + ], + [ + 409, + 160 + ], + [ + 437, + 155 + ] + ], + [ + [ + 442, + 176 + ], + [ + 458, + 187 + ], + [ + 443, + 172 + ], + [ + 440, + 180 + ], + [ + 427, + 162 + ], + [ + 425, + 170 + ] + ], + [ + [ + 419, + 162 + ], + [ + 423, + 148 + ], + [ + 434, + 156 + ], + [ + 404, + 168 + ], + [ + 430, + 170 + ], + [ + 401, + 181 + ] + ] + ] + }, + { + "task_id": "0004703", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_819112070044#289", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004703/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 395, + 162 + ], + [ + 390, + 174 + ], + [ + 394, + 147 + ], + [ + 395, + 176 + ], + [ + 399, + 135 + ], + [ + 400, + 164 + ] + ], + [ + [ + 428, + 167 + ], + [ + 432, + 156 + ], + [ + 442, + 157 + ], + [ + 413, + 177 + ], + [ + 438, + 168 + ], + [ + 410, + 188 + ] + ], + [ + [ + 388, + 171 + ], + [ + 386, + 154 + ], + [ + 400, + 175 + ], + [ + 376, + 167 + ], + [ + 402, + 192 + ], + [ + 378, + 185 + ] + ], + [ + [ + 402, + 173 + ], + [ + 405, + 158 + ], + [ + 418, + 168 + ], + [ + 385, + 178 + ], + [ + 414, + 181 + ], + [ + 382, + 191 + ] + ] + ] + }, + { + "task_id": "0004704", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_819112070044#398", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004704/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 386, + 179 + ], + [ + 394, + 195 + ], + [ + 388, + 188 + ], + [ + 384, + 169 + ], + [ + 380, + 172 + ], + [ + 376, + 153 + ] + ], + [ + [ + 385, + 179 + ], + [ + 388, + 165 + ], + [ + 402, + 174 + ], + [ + 369, + 185 + ], + [ + 399, + 187 + ], + [ + 366, + 198 + ] + ], + [ + [ + 384, + 201 + ], + [ + 366, + 210 + ], + [ + 375, + 189 + ], + [ + 392, + 213 + ], + [ + 392, + 180 + ], + [ + 410, + 203 + ] + ], + [ + [ + 389, + 185 + ], + [ + 397, + 204 + ], + [ + 407, + 178 + ], + [ + 370, + 193 + ], + [ + 399, + 160 + ], + [ + 362, + 175 + ] + ] + ] + }, + { + "task_id": "0004705", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_819112070044#529", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004705/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 365, + 203 + ], + [ + 369, + 210 + ], + [ + 364, + 221 + ], + [ + 366, + 184 + ], + [ + 359, + 214 + ], + [ + 362, + 178 + ] + ], + [ + [ + 367, + 183 + ], + [ + 369, + 169 + ], + [ + 383, + 177 + ], + [ + 350, + 189 + ], + [ + 380, + 190 + ], + [ + 348, + 202 + ] + ], + [ + [ + 358, + 185 + ], + [ + 366, + 167 + ], + [ + 374, + 192 + ], + [ + 342, + 177 + ], + [ + 366, + 210 + ], + [ + 334, + 195 + ] + ], + [ + [ + 357, + 183 + ], + [ + 348, + 169 + ], + [ + 370, + 169 + ], + [ + 344, + 196 + ], + [ + 379, + 183 + ], + [ + 353, + 210 + ] + ] + ] + }, + { + "task_id": "0004706", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_104122063678#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004706/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 372, + 200 + ], + [ + 353, + 203 + ], + [ + 353, + 208 + ], + [ + 392, + 192 + ], + [ + 371, + 205 + ], + [ + 410, + 190 + ] + ], + [ + [ + 351, + 195 + ], + [ + 329, + 179 + ], + [ + 346, + 193 + ], + [ + 356, + 197 + ], + [ + 366, + 208 + ], + [ + 378, + 214 + ] + ], + [ + [ + 374, + 189 + ], + [ + 380, + 199 + ], + [ + 348, + 194 + ], + [ + 402, + 184 + ], + [ + 340, + 184 + ], + [ + 397, + 172 + ] + ], + [ + [ + 373, + 184 + ], + [ + 382, + 166 + ], + [ + 367, + 203 + ], + [ + 378, + 166 + ], + [ + 359, + 219 + ], + [ + 370, + 183 + ] + ] + ] + }, + { + "task_id": "0004707", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_104122063678#246", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004707/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 383, + 199 + ], + [ + 391, + 179 + ], + [ + 380, + 219 + ], + [ + 385, + 180 + ], + [ + 371, + 237 + ], + [ + 377, + 198 + ] + ], + [ + [ + 390, + 208 + ], + [ + 364, + 202 + ], + [ + 388, + 200 + ], + [ + 394, + 218 + ], + [ + 413, + 205 + ], + [ + 422, + 224 + ] + ], + [ + [ + 389, + 182 + ], + [ + 398, + 207 + ], + [ + 395, + 176 + ], + [ + 383, + 189 + ], + [ + 386, + 151 + ], + [ + 373, + 162 + ] + ], + [ + [ + 394, + 212 + ], + [ + 375, + 211 + ], + [ + 403, + 234 + ], + [ + 384, + 189 + ], + [ + 425, + 237 + ], + [ + 405, + 189 + ] + ] + ] + }, + { + "task_id": "0004708", + "id": "RH20T_cfg5_task_0094_user_0007_scene_0006_cfg_0005_104122063678#307", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004708/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 359, + 204 + ], + [ + 367, + 185 + ], + [ + 355, + 224 + ], + [ + 363, + 185 + ], + [ + 348, + 241 + ], + [ + 356, + 202 + ] + ], + [ + [ + 334, + 193 + ], + [ + 360, + 195 + ], + [ + 334, + 168 + ], + [ + 335, + 218 + ], + [ + 309, + 167 + ], + [ + 310, + 216 + ] + ], + [ + [ + 329, + 219 + ], + [ + 330, + 233 + ], + [ + 347, + 236 + ], + [ + 311, + 203 + ], + [ + 346, + 220 + ], + [ + 309, + 187 + ] + ], + [ + [ + 364, + 231 + ], + [ + 344, + 215 + ], + [ + 380, + 228 + ], + [ + 351, + 233 + ], + [ + 400, + 245 + ], + [ + 370, + 248 + ] + ] + ] + }, + { + "task_id": "0004713", + "id": "RH20T_cfg4_task_0014_user_0010_scene_0005_cfg_0004_036422060909#109", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the object from the scale'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004713/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 137 + ], + [ + 305, + 117 + ], + [ + 327, + 127 + ], + [ + 283, + 148 + ], + [ + 327, + 146 + ], + [ + 285, + 167 + ] + ], + [ + [ + 294, + 144 + ], + [ + 282, + 166 + ], + [ + 311, + 144 + ], + [ + 279, + 143 + ], + [ + 324, + 121 + ], + [ + 291, + 122 + ] + ], + [ + [ + 296, + 129 + ], + [ + 304, + 129 + ], + [ + 277, + 110 + ], + [ + 313, + 147 + ], + [ + 271, + 111 + ], + [ + 305, + 146 + ] + ], + [ + [ + 311, + 139 + ], + [ + 293, + 135 + ], + [ + 331, + 130 + ], + [ + 289, + 148 + ], + [ + 350, + 134 + ], + [ + 309, + 153 + ] + ] + ] + }, + { + "task_id": "0004714", + "id": "RH20T_cfg4_task_0013_user_0010_scene_0002_cfg_0004_038522062288#111", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004714/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 333, + 333 + ], + [ + 333, + 315 + ], + [ + 315, + 307 + ], + [ + 353, + 360 + ], + [ + 316, + 323 + ], + [ + 352, + 360 + ] + ], + [ + [ + 298, + 336 + ], + [ + 324, + 332 + ], + [ + 302, + 360 + ], + [ + 295, + 303 + ], + [ + 273, + 360 + ], + [ + 267, + 305 + ] + ], + [ + [ + 370, + 351 + ], + [ + 336, + 342 + ], + [ + 367, + 360 + ], + [ + 374, + 329 + ], + [ + 400, + 360 + ], + [ + 411, + 339 + ] + ], + [ + [ + 339, + 326 + ], + [ + 310, + 329 + ], + [ + 356, + 358 + ], + [ + 323, + 298 + ], + [ + 382, + 354 + ], + [ + 348, + 298 + ] + ] + ] + }, + { + "task_id": "0004716", + "id": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_104122062295#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004716/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 246, + 223 + ], + [ + 224, + 260 + ], + [ + 216, + 226 + ], + [ + 271, + 220 + ], + [ + 239, + 183 + ], + [ + 292, + 185 + ] + ], + [ + [ + 232, + 234 + ], + [ + 245, + 193 + ], + [ + 206, + 207 + ], + [ + 262, + 265 + ], + [ + 195, + 244 + ], + [ + 248, + 303 + ] + ], + [ + [ + 211, + 284 + ], + [ + 254, + 282 + ], + [ + 208, + 231 + ], + [ + 214, + 342 + ], + [ + 158, + 228 + ], + [ + 160, + 350 + ] + ], + [ + [ + 211, + 255 + ], + [ + 256, + 256 + ], + [ + 225, + 229 + ], + [ + 199, + 278 + ], + [ + 172, + 226 + ], + [ + 152, + 278 + ] + ] + ] + }, + { + "task_id": "0004718", + "id": "RH20T_cfg4_task_0037_user_0016_scene_0002_cfg_0004_104122062295#269", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004718/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 228, + 156 + ], + [ + 242, + 174 + ], + [ + 258, + 175 + ], + [ + 201, + 137 + ], + [ + 244, + 155 + ], + [ + 183, + 116 + ] + ], + [ + [ + 222, + 175 + ], + [ + 228, + 140 + ], + [ + 189, + 160 + ], + [ + 257, + 191 + ], + [ + 185, + 192 + ], + [ + 250, + 223 + ] + ], + [ + [ + 178, + 192 + ], + [ + 154, + 162 + ], + [ + 191, + 204 + ], + [ + 167, + 181 + ], + [ + 213, + 233 + ], + [ + 188, + 206 + ] + ], + [ + [ + 209, + 150 + ], + [ + 223, + 154 + ], + [ + 235, + 178 + ], + [ + 184, + 124 + ], + [ + 220, + 176 + ], + [ + 166, + 117 + ] + ] + ] + }, + { + "task_id": "0004723", + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_039422060546#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004723/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 156, + 258 + ], + [ + 181, + 295 + ], + [ + 159, + 254 + ], + [ + 153, + 261 + ], + [ + 131, + 213 + ], + [ + 130, + 227 + ] + ], + [ + [ + 146, + 255 + ], + [ + 128, + 234 + ], + [ + 160, + 294 + ], + [ + 134, + 220 + ], + [ + 175, + 309 + ], + [ + 150, + 240 + ] + ], + [ + [ + 163, + 200 + ], + [ + 200, + 181 + ], + [ + 137, + 161 + ], + [ + 187, + 237 + ], + [ + 101, + 180 + ], + [ + 152, + 254 + ] + ], + [ + [ + 126, + 257 + ], + [ + 115, + 302 + ], + [ + 118, + 274 + ], + [ + 137, + 237 + ], + [ + 128, + 236 + ], + [ + 147, + 195 + ] + ] + ] + }, + { + "task_id": "0004725", + "id": "RH20T_cfg1_task_0094_user_0001_scene_0009_cfg_0001_039422060546#558", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004725/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 171, + 266 + ], + [ + 127, + 282 + ], + [ + 150, + 231 + ], + [ + 189, + 298 + ], + [ + 196, + 214 + ], + [ + 231, + 282 + ] + ], + [ + [ + 140, + 263 + ], + [ + 120, + 246 + ], + [ + 145, + 304 + ], + [ + 136, + 226 + ], + [ + 162, + 315 + ], + [ + 153, + 242 + ] + ], + [ + [ + 188, + 260 + ], + [ + 184, + 267 + ], + [ + 212, + 227 + ], + [ + 163, + 297 + ], + [ + 218, + 216 + ], + [ + 165, + 291 + ] + ], + [ + [ + 144, + 295 + ], + [ + 124, + 307 + ], + [ + 123, + 258 + ], + [ + 165, + 333 + ], + [ + 145, + 243 + ], + [ + 189, + 322 + ] + ] + ] + }, + { + "task_id": "0004726", + "id": "RH20T_cfg4_task_0065_user_0016_scene_0005_cfg_0004_104122063550#53", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004726/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 385, + 272 + ], + [ + 390, + 267 + ], + [ + 410, + 289 + ], + [ + 360, + 255 + ], + [ + 404, + 292 + ], + [ + 357, + 260 + ] + ], + [ + [ + 383, + 254 + ], + [ + 375, + 225 + ], + [ + 362, + 257 + ], + [ + 402, + 251 + ], + [ + 370, + 288 + ], + [ + 410, + 279 + ] + ], + [ + [ + 405, + 272 + ], + [ + 419, + 285 + ], + [ + 426, + 284 + ], + [ + 383, + 260 + ], + [ + 414, + 272 + ], + [ + 372, + 249 + ] + ], + [ + [ + 364, + 261 + ], + [ + 386, + 242 + ], + [ + 346, + 240 + ], + [ + 382, + 281 + ], + [ + 324, + 259 + ], + [ + 360, + 300 + ] + ] + ] + }, + { + "task_id": "0004728", + "id": "RH20T_cfg4_task_0065_user_0016_scene_0005_cfg_0004_104122063550#193", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004728/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 378, + 242 + ], + [ + 379, + 215 + ], + [ + 374, + 237 + ], + [ + 383, + 248 + ], + [ + 373, + 262 + ], + [ + 382, + 275 + ] + ], + [ + [ + 383, + 229 + ], + [ + 385, + 222 + ], + [ + 392, + 256 + ], + [ + 375, + 204 + ], + [ + 390, + 260 + ], + [ + 373, + 211 + ] + ], + [ + [ + 388, + 220 + ], + [ + 406, + 202 + ], + [ + 407, + 238 + ], + [ + 368, + 203 + ], + [ + 389, + 256 + ], + [ + 350, + 222 + ] + ], + [ + [ + 403, + 249 + ], + [ + 374, + 254 + ], + [ + 408, + 278 + ], + [ + 397, + 220 + ], + [ + 438, + 274 + ], + [ + 426, + 215 + ] + ] + ] + }, + { + "task_id": "0004731", + "id": "RH20T_cfg4_task_0103_user_0007_scene_0003_cfg_0004_104122062295#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004731/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 195, + 94 + ], + [ + 217, + 85 + ], + [ + 212, + 74 + ], + [ + 180, + 112 + ], + [ + 187, + 82 + ], + [ + 155, + 122 + ] + ], + [ + [ + 192, + 109 + ], + [ + 201, + 81 + ], + [ + 167, + 89 + ], + [ + 218, + 130 + ], + [ + 160, + 114 + ], + [ + 209, + 154 + ] + ], + [ + [ + 154, + 124 + ], + [ + 149, + 116 + ], + [ + 175, + 150 + ], + [ + 133, + 96 + ], + [ + 182, + 160 + ], + [ + 137, + 104 + ] + ], + [ + [ + 188, + 81 + ], + [ + 157, + 93 + ], + [ + 173, + 47 + ], + [ + 203, + 113 + ], + [ + 202, + 37 + ], + [ + 229, + 101 + ] + ] + ] + }, + { + "task_id": "0004732", + "id": "RH20T_cfg4_task_0103_user_0007_scene_0003_cfg_0004_104122062295#131", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004732/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 206, + 145 + ], + [ + 194, + 179 + ], + [ + 172, + 130 + ], + [ + 240, + 160 + ], + [ + 184, + 94 + ], + [ + 254, + 125 + ] + ], + [ + [ + 218, + 129 + ], + [ + 191, + 138 + ], + [ + 195, + 114 + ], + [ + 242, + 144 + ], + [ + 220, + 107 + ], + [ + 266, + 135 + ] + ], + [ + [ + 228, + 144 + ], + [ + 234, + 114 + ], + [ + 203, + 125 + ], + [ + 253, + 165 + ], + [ + 199, + 153 + ], + [ + 246, + 192 + ] + ], + [ + [ + 256, + 151 + ], + [ + 221, + 144 + ], + [ + 258, + 165 + ], + [ + 255, + 139 + ], + [ + 294, + 172 + ], + [ + 287, + 146 + ] + ] + ] + }, + { + "task_id": "0004733", + "id": "RH20T_cfg6_task_0046_user_0020_scene_0010_cfg_0006_104122061602#318", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004733/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 492, + 155 + ], + [ + 506, + 177 + ], + [ + 476, + 169 + ], + [ + 509, + 139 + ], + [ + 464, + 148 + ], + [ + 495, + 118 + ] + ], + [ + [ + 513, + 161 + ], + [ + 532, + 158 + ], + [ + 535, + 169 + ], + [ + 492, + 154 + ], + [ + 514, + 173 + ], + [ + 473, + 157 + ] + ], + [ + [ + 515, + 160 + ], + [ + 522, + 137 + ], + [ + 511, + 177 + ], + [ + 519, + 144 + ], + [ + 505, + 199 + ], + [ + 513, + 165 + ] + ], + [ + [ + 481, + 177 + ], + [ + 494, + 157 + ], + [ + 473, + 183 + ], + [ + 489, + 171 + ], + [ + 460, + 202 + ], + [ + 474, + 192 + ] + ] + ] + }, + { + "task_id": "0004734", + "id": "RH20T_cfg1_task_0053_user_0014_scene_0003_cfg_0001_038522062547#175", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use a clamp'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004734/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 364, + 66 + ], + [ + 361, + 52 + ], + [ + 376, + 80 + ], + [ + 352, + 51 + ], + [ + 379, + 95 + ], + [ + 354, + 66 + ] + ], + [ + [ + 377, + 71 + ], + [ + 378, + 50 + ], + [ + 401, + 70 + ], + [ + 354, + 72 + ], + [ + 400, + 91 + ], + [ + 355, + 92 + ] + ], + [ + [ + 383, + 72 + ], + [ + 403, + 81 + ], + [ + 396, + 52 + ], + [ + 370, + 92 + ], + [ + 375, + 42 + ], + [ + 350, + 83 + ] + ], + [ + [ + 358, + 84 + ], + [ + 373, + 76 + ], + [ + 375, + 94 + ], + [ + 341, + 74 + ], + [ + 362, + 101 + ], + [ + 328, + 82 + ] + ] + ] + }, + { + "task_id": "0004735", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_037522061512#119", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004735/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 254, + 242 + ], + [ + 259, + 250 + ], + [ + 261, + 217 + ], + [ + 247, + 269 + ], + [ + 256, + 207 + ], + [ + 241, + 261 + ] + ], + [ + [ + 246, + 218 + ], + [ + 227, + 216 + ], + [ + 233, + 201 + ], + [ + 258, + 234 + ], + [ + 254, + 203 + ], + [ + 278, + 237 + ] + ], + [ + [ + 255, + 220 + ], + [ + 262, + 195 + ], + [ + 267, + 215 + ], + [ + 241, + 225 + ], + [ + 260, + 238 + ], + [ + 234, + 250 + ] + ], + [ + [ + 261, + 246 + ], + [ + 263, + 222 + ], + [ + 259, + 256 + ], + [ + 264, + 235 + ], + [ + 257, + 280 + ], + [ + 262, + 262 + ] + ] + ] + }, + { + "task_id": "0004736", + "id": "RH20T_cfg6_task_0081_user_0020_scene_0004_cfg_0006_037522061512#311", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004736/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 230, + 187 + ], + [ + 245, + 201 + ], + [ + 212, + 189 + ], + [ + 246, + 185 + ], + [ + 198, + 175 + ], + [ + 232, + 172 + ] + ], + [ + [ + 228, + 177 + ], + [ + 244, + 166 + ], + [ + 225, + 193 + ], + [ + 230, + 160 + ], + [ + 208, + 204 + ], + [ + 211, + 171 + ] + ], + [ + [ + 242, + 181 + ], + [ + 222, + 172 + ], + [ + 237, + 182 + ], + [ + 247, + 180 + ], + [ + 259, + 192 + ], + [ + 267, + 189 + ] + ], + [ + [ + 229, + 206 + ], + [ + 246, + 214 + ], + [ + 217, + 193 + ], + [ + 240, + 217 + ], + [ + 201, + 185 + ], + [ + 224, + 209 + ] + ] + ] + }, + { + "task_id": "0004739", + "id": "RH20T_cfg5_task_0002_user_0010_scene_0001_cfg_0005_104122061850#100", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004739/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 116 + ], + [ + 355, + 91 + ], + [ + 336, + 117 + ], + [ + 358, + 114 + ], + [ + 328, + 142 + ], + [ + 351, + 137 + ] + ], + [ + [ + 348, + 140 + ], + [ + 337, + 118 + ], + [ + 331, + 148 + ], + [ + 366, + 132 + ], + [ + 342, + 169 + ], + [ + 378, + 155 + ] + ], + [ + [ + 351, + 97 + ], + [ + 350, + 104 + ], + [ + 337, + 75 + ], + [ + 365, + 118 + ], + [ + 337, + 66 + ], + [ + 367, + 112 + ] + ], + [ + [ + 339, + 133 + ], + [ + 324, + 150 + ], + [ + 352, + 134 + ], + [ + 327, + 132 + ], + [ + 366, + 116 + ], + [ + 341, + 115 + ] + ] + ] + }, + { + "task_id": "0004740", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122061602#227", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004740/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 369, + 177 + ], + [ + 380, + 170 + ], + [ + 379, + 201 + ], + [ + 359, + 154 + ], + [ + 369, + 206 + ], + [ + 349, + 162 + ] + ], + [ + [ + 369, + 149 + ], + [ + 350, + 132 + ], + [ + 375, + 152 + ], + [ + 362, + 145 + ], + [ + 394, + 168 + ], + [ + 383, + 163 + ] + ], + [ + [ + 352, + 206 + ], + [ + 334, + 214 + ], + [ + 359, + 190 + ], + [ + 345, + 224 + ], + [ + 378, + 181 + ], + [ + 365, + 216 + ] + ], + [ + [ + 345, + 171 + ], + [ + 331, + 183 + ], + [ + 338, + 189 + ], + [ + 352, + 152 + ], + [ + 351, + 178 + ], + [ + 365, + 141 + ] + ] + ] + }, + { + "task_id": "0004743", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0009_cfg_0006_104122061602#1324", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004743/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 484, + 230 + ], + [ + 461, + 224 + ], + [ + 471, + 252 + ], + [ + 499, + 207 + ], + [ + 493, + 257 + ], + [ + 521, + 213 + ] + ], + [ + [ + 499, + 270 + ], + [ + 528, + 257 + ], + [ + 492, + 265 + ], + [ + 505, + 273 + ], + [ + 463, + 278 + ], + [ + 478, + 284 + ] + ], + [ + [ + 508, + 263 + ], + [ + 528, + 249 + ], + [ + 486, + 255 + ], + [ + 533, + 271 + ], + [ + 466, + 268 + ], + [ + 512, + 286 + ] + ], + [ + [ + 490, + 247 + ], + [ + 509, + 243 + ], + [ + 503, + 276 + ], + [ + 478, + 220 + ], + [ + 485, + 278 + ], + [ + 463, + 225 + ] + ] + ] + }, + { + "task_id": "0004744", + "id": "RH20T_cfg1_task_0012_user_0014_scene_0004_cfg_0001_038522063145#101", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push the soccer ball into the goal'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004744/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 418, + 131 + ], + [ + 398, + 122 + ], + [ + 411, + 151 + ], + [ + 426, + 110 + ], + [ + 432, + 160 + ], + [ + 447, + 119 + ] + ], + [ + [ + 418, + 127 + ], + [ + 427, + 110 + ], + [ + 432, + 141 + ], + [ + 404, + 113 + ], + [ + 423, + 156 + ], + [ + 395, + 129 + ] + ], + [ + [ + 417, + 114 + ], + [ + 413, + 95 + ], + [ + 397, + 119 + ], + [ + 437, + 108 + ], + [ + 401, + 139 + ], + [ + 442, + 128 + ] + ], + [ + [ + 431, + 129 + ], + [ + 442, + 117 + ], + [ + 438, + 148 + ], + [ + 423, + 109 + ], + [ + 426, + 161 + ], + [ + 410, + 122 + ] + ] + ] + }, + { + "task_id": "0004747", + "id": "RH20T_cfg5_task_0075_user_0007_scene_0004_cfg_0005_104122061850#599", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Finish setting up the starting position of a chessboard that is almost arranged'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004747/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 440, + 155 + ], + [ + 440, + 141 + ], + [ + 418, + 170 + ], + [ + 462, + 140 + ], + [ + 419, + 182 + ], + [ + 460, + 154 + ] + ], + [ + [ + 466, + 156 + ], + [ + 449, + 138 + ], + [ + 482, + 133 + ], + [ + 452, + 177 + ], + [ + 502, + 152 + ], + [ + 469, + 196 + ] + ], + [ + [ + 437, + 146 + ], + [ + 456, + 167 + ], + [ + 456, + 128 + ], + [ + 419, + 163 + ], + [ + 436, + 107 + ], + [ + 402, + 143 + ] + ], + [ + [ + 413, + 172 + ], + [ + 432, + 157 + ], + [ + 396, + 168 + ], + [ + 428, + 176 + ], + [ + 378, + 182 + ], + [ + 410, + 189 + ] + ] + ] + }, + { + "task_id": "0004752", + "id": "RH20T_cfg4_task_0094_user_0007_scene_0003_cfg_0004_104122062823#336", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004752/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 247, + 9 + ], + [ + 254, + 0 + ], + [ + 280, + 23 + ], + [ + 215, + 0 + ], + [ + 272, + 55 + ], + [ + 211, + 28 + ] + ], + [ + [ + 253, + 26 + ], + [ + 227, + 43 + ], + [ + 231, + 1 + ], + [ + 275, + 52 + ], + [ + 257, + 0 + ], + [ + 300, + 36 + ] + ], + [ + [ + 233, + 0 + ], + [ + 226, + 0 + ], + [ + 265, + 0 + ], + [ + 199, + 0 + ], + [ + 275, + 0 + ], + [ + 205, + 0 + ] + ], + [ + [ + 247, + 32 + ], + [ + 274, + 27 + ], + [ + 266, + 49 + ], + [ + 229, + 16 + ], + [ + 235, + 55 + ], + [ + 199, + 20 + ] + ] + ] + }, + { + "task_id": "0004753", + "id": "RH20T_cfg7_task_0046_user_0014_scene_0001_cfg_0007_037522061512#97", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004753/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 209, + 110 + ], + [ + 192, + 124 + ], + [ + 227, + 129 + ], + [ + 191, + 90 + ], + [ + 246, + 116 + ], + [ + 210, + 75 + ] + ], + [ + [ + 226, + 142 + ], + [ + 237, + 134 + ], + [ + 230, + 165 + ], + [ + 221, + 118 + ], + [ + 218, + 175 + ], + [ + 208, + 126 + ] + ], + [ + [ + 219, + 136 + ], + [ + 210, + 119 + ], + [ + 233, + 118 + ], + [ + 205, + 155 + ], + [ + 240, + 135 + ], + [ + 214, + 170 + ] + ], + [ + [ + 226, + 157 + ], + [ + 205, + 151 + ], + [ + 241, + 158 + ], + [ + 210, + 157 + ], + [ + 262, + 164 + ], + [ + 233, + 163 + ] + ] + ] + }, + { + "task_id": "0004755", + "id": "RH20T_cfg6_task_0046_user_0020_scene_0004_cfg_0006_037522061512#372", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004755/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 251, + 117 + ], + [ + 273, + 121 + ], + [ + 261, + 128 + ], + [ + 240, + 106 + ], + [ + 241, + 125 + ], + [ + 218, + 103 + ] + ], + [ + [ + 236, + 112 + ], + [ + 250, + 110 + ], + [ + 236, + 136 + ], + [ + 236, + 86 + ], + [ + 220, + 140 + ], + [ + 220, + 88 + ] + ], + [ + [ + 233, + 108 + ], + [ + 247, + 117 + ], + [ + 248, + 124 + ], + [ + 219, + 93 + ], + [ + 233, + 115 + ], + [ + 203, + 83 + ] + ], + [ + [ + 223, + 101 + ], + [ + 233, + 111 + ], + [ + 234, + 79 + ], + [ + 212, + 123 + ], + [ + 223, + 67 + ], + [ + 201, + 113 + ] + ] + ] + }, + { + "task_id": "0004756", + "id": "RH20T_cfg3_task_0018_user_0010_scene_0003_cfg_0003_038522062288#179", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004756/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 237, + 223 + ], + [ + 224, + 225 + ], + [ + 254, + 200 + ], + [ + 219, + 247 + ], + [ + 269, + 197 + ], + [ + 234, + 246 + ] + ], + [ + [ + 249, + 220 + ], + [ + 244, + 194 + ], + [ + 265, + 227 + ], + [ + 235, + 214 + ], + [ + 271, + 253 + ], + [ + 240, + 239 + ] + ], + [ + [ + 229, + 226 + ], + [ + 225, + 210 + ], + [ + 208, + 209 + ], + [ + 251, + 243 + ], + [ + 213, + 224 + ], + [ + 253, + 256 + ] + ], + [ + [ + 259, + 244 + ], + [ + 258, + 260 + ], + [ + 282, + 260 + ], + [ + 236, + 230 + ], + [ + 283, + 243 + ], + [ + 235, + 212 + ] + ] + ] + }, + { + "task_id": "0004759", + "id": "RH20T_cfg1_task_0061_user_0015_scene_0001_cfg_0001_038522063145#531", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004759/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 369, + 103 + ], + [ + 361, + 92 + ], + [ + 384, + 111 + ], + [ + 354, + 95 + ], + [ + 394, + 122 + ], + [ + 363, + 107 + ] + ], + [ + [ + 364, + 111 + ], + [ + 369, + 93 + ], + [ + 385, + 120 + ], + [ + 343, + 104 + ], + [ + 379, + 136 + ], + [ + 339, + 120 + ] + ], + [ + [ + 359, + 113 + ], + [ + 359, + 93 + ], + [ + 335, + 112 + ], + [ + 383, + 115 + ], + [ + 335, + 132 + ], + [ + 383, + 136 + ] + ], + [ + [ + 389, + 89 + ], + [ + 399, + 71 + ], + [ + 384, + 101 + ], + [ + 394, + 78 + ], + [ + 375, + 118 + ], + [ + 385, + 95 + ] + ] + ] + }, + { + "task_id": "0004760", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0009_cfg_0005_104422070044#126", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004760/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 403, + 131 + ], + [ + 402, + 116 + ], + [ + 413, + 142 + ], + [ + 393, + 121 + ], + [ + 413, + 156 + ], + [ + 393, + 135 + ] + ], + [ + [ + 400, + 129 + ], + [ + 406, + 111 + ], + [ + 405, + 125 + ], + [ + 395, + 132 + ], + [ + 399, + 144 + ], + [ + 389, + 149 + ] + ], + [ + [ + 414, + 147 + ], + [ + 407, + 159 + ], + [ + 397, + 137 + ], + [ + 430, + 157 + ], + [ + 404, + 126 + ], + [ + 436, + 145 + ] + ], + [ + [ + 409, + 126 + ], + [ + 395, + 135 + ], + [ + 421, + 129 + ], + [ + 398, + 124 + ], + [ + 436, + 120 + ], + [ + 412, + 116 + ] + ] + ] + }, + { + "task_id": "0004762", + "id": "RH20T_cfg4_task_0064_user_0016_scene_0003_cfg_0004_038522062288#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004762/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 172, + 163 + ], + [ + 176, + 191 + ], + [ + 171, + 171 + ], + [ + 172, + 156 + ], + [ + 165, + 140 + ], + [ + 168, + 128 + ] + ], + [ + [ + 123, + 145 + ], + [ + 107, + 120 + ], + [ + 148, + 125 + ], + [ + 99, + 164 + ], + [ + 164, + 151 + ], + [ + 115, + 190 + ] + ], + [ + [ + 105, + 161 + ], + [ + 89, + 136 + ], + [ + 76, + 177 + ], + [ + 133, + 146 + ], + [ + 92, + 200 + ], + [ + 146, + 169 + ] + ], + [ + [ + 140, + 154 + ], + [ + 130, + 131 + ], + [ + 113, + 171 + ], + [ + 165, + 138 + ], + [ + 123, + 192 + ], + [ + 172, + 159 + ] + ] + ] + }, + { + "task_id": "0004763", + "id": "RH20T_cfg4_task_0111_user_0007_scene_0003_cfg_0004_038522062288#124", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object with the corresponding shape into the corresponding hole'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004763/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 196, + 306 + ], + [ + 165, + 329 + ], + [ + 209, + 332 + ], + [ + 183, + 278 + ], + [ + 236, + 310 + ], + [ + 213, + 257 + ] + ], + [ + [ + 224, + 329 + ], + [ + 242, + 299 + ], + [ + 242, + 348 + ], + [ + 205, + 309 + ], + [ + 224, + 360 + ], + [ + 183, + 343 + ] + ], + [ + [ + 227, + 313 + ], + [ + 222, + 300 + ], + [ + 203, + 293 + ], + [ + 252, + 334 + ], + [ + 209, + 306 + ], + [ + 255, + 345 + ] + ], + [ + [ + 245, + 307 + ], + [ + 214, + 316 + ], + [ + 251, + 311 + ], + [ + 239, + 303 + ], + [ + 280, + 303 + ], + [ + 272, + 294 + ] + ] + ] + }, + { + "task_id": "0004764", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_038522062547#111", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004764/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 400, + 315 + ], + [ + 379, + 328 + ], + [ + 389, + 333 + ], + [ + 412, + 295 + ], + [ + 407, + 321 + ], + [ + 431, + 283 + ] + ], + [ + [ + 423, + 302 + ], + [ + 455, + 291 + ], + [ + 411, + 290 + ], + [ + 437, + 315 + ], + [ + 380, + 300 + ], + [ + 401, + 326 + ] + ], + [ + [ + 426, + 290 + ], + [ + 396, + 294 + ], + [ + 437, + 321 + ], + [ + 416, + 261 + ], + [ + 466, + 316 + ], + [ + 443, + 257 + ] + ], + [ + [ + 426, + 285 + ], + [ + 432, + 278 + ], + [ + 441, + 316 + ], + [ + 411, + 256 + ], + [ + 435, + 321 + ], + [ + 407, + 264 + ] + ] + ] + }, + { + "task_id": "0004767", + "id": "RH20T_cfg4_task_0026_user_0016_scene_0001_cfg_0004_104422070011#128", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push and close the drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004767/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 122 + ], + [ + 351, + 114 + ], + [ + 350, + 125 + ], + [ + 390, + 120 + ], + [ + 373, + 135 + ], + [ + 412, + 129 + ] + ], + [ + [ + 421, + 99 + ], + [ + 402, + 114 + ], + [ + 403, + 77 + ], + [ + 440, + 121 + ], + [ + 423, + 59 + ], + [ + 462, + 106 + ] + ], + [ + [ + 379, + 125 + ], + [ + 380, + 97 + ], + [ + 353, + 131 + ], + [ + 407, + 119 + ], + [ + 352, + 159 + ], + [ + 407, + 149 + ] + ], + [ + [ + 392, + 128 + ], + [ + 402, + 111 + ], + [ + 364, + 126 + ], + [ + 420, + 131 + ], + [ + 357, + 142 + ], + [ + 410, + 146 + ] + ] + ] + }, + { + "task_id": "0004768", + "id": "RH20T_cfg4_task_0009_user_0010_scene_0006_cfg_0004_038522062288#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004768/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 215, + 179 + ], + [ + 209, + 161 + ], + [ + 190, + 166 + ], + [ + 241, + 193 + ], + [ + 197, + 183 + ], + [ + 245, + 208 + ] + ], + [ + [ + 233, + 193 + ], + [ + 232, + 184 + ], + [ + 204, + 180 + ], + [ + 262, + 207 + ], + [ + 206, + 189 + ], + [ + 261, + 214 + ] + ], + [ + [ + 246, + 164 + ], + [ + 239, + 159 + ], + [ + 236, + 194 + ], + [ + 257, + 133 + ], + [ + 243, + 201 + ], + [ + 266, + 137 + ] + ], + [ + [ + 206, + 174 + ], + [ + 231, + 179 + ], + [ + 212, + 142 + ], + [ + 201, + 206 + ], + [ + 185, + 136 + ], + [ + 173, + 203 + ] + ] + ] + }, + { + "task_id": "0004769", + "id": "RH20T_cfg6_task_0046_user_0020_scene_0004_cfg_0006_104122061330#373", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004769/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 303, + 151 + ], + [ + 327, + 169 + ], + [ + 290, + 157 + ], + [ + 318, + 145 + ], + [ + 266, + 140 + ], + [ + 292, + 126 + ] + ], + [ + [ + 312, + 149 + ], + [ + 289, + 135 + ], + [ + 310, + 170 + ], + [ + 313, + 126 + ], + [ + 333, + 184 + ], + [ + 338, + 140 + ] + ], + [ + [ + 327, + 133 + ], + [ + 325, + 118 + ], + [ + 349, + 152 + ], + [ + 307, + 114 + ], + [ + 350, + 164 + ], + [ + 310, + 128 + ] + ], + [ + [ + 315, + 145 + ], + [ + 306, + 173 + ], + [ + 321, + 146 + ], + [ + 307, + 145 + ], + [ + 330, + 119 + ], + [ + 316, + 114 + ] + ] + ] + }, + { + "task_id": "0004771", + "id": "RH20T_cfg6_task_0039_user_0014_scene_0003_cfg_0006_104122064161#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004771/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 140 + ], + [ + 311, + 159 + ], + [ + 344, + 155 + ], + [ + 307, + 125 + ], + [ + 359, + 137 + ], + [ + 322, + 107 + ] + ], + [ + [ + 343, + 112 + ], + [ + 346, + 93 + ], + [ + 341, + 99 + ], + [ + 346, + 124 + ], + [ + 338, + 120 + ], + [ + 344, + 144 + ] + ], + [ + [ + 321, + 122 + ], + [ + 304, + 129 + ], + [ + 326, + 105 + ], + [ + 316, + 138 + ], + [ + 343, + 99 + ], + [ + 332, + 132 + ] + ], + [ + [ + 340, + 118 + ], + [ + 341, + 101 + ], + [ + 340, + 136 + ], + [ + 341, + 102 + ], + [ + 339, + 151 + ], + [ + 340, + 118 + ] + ] + ] + }, + { + "task_id": "0004772", + "id": "RH20T_cfg5_task_0101_user_0007_scene_0007_cfg_0005_036422060215#175", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the table with a cloth'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004772/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 183, + 178 + ], + [ + 185, + 161 + ], + [ + 207, + 176 + ], + [ + 157, + 179 + ], + [ + 204, + 192 + ], + [ + 156, + 195 + ] + ], + [ + [ + 197, + 181 + ], + [ + 202, + 200 + ], + [ + 190, + 193 + ], + [ + 204, + 168 + ], + [ + 185, + 175 + ], + [ + 199, + 149 + ] + ], + [ + [ + 204, + 196 + ], + [ + 197, + 183 + ], + [ + 201, + 215 + ], + [ + 207, + 175 + ], + [ + 208, + 230 + ], + [ + 215, + 189 + ] + ], + [ + [ + 179, + 195 + ], + [ + 199, + 191 + ], + [ + 173, + 172 + ], + [ + 185, + 219 + ], + [ + 151, + 176 + ], + [ + 163, + 225 + ] + ] + ] + }, + { + "task_id": "0004773", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_036422060215#179", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004773/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 271, + 173 + ], + [ + 277, + 158 + ], + [ + 290, + 186 + ], + [ + 253, + 161 + ], + [ + 284, + 200 + ], + [ + 249, + 176 + ] + ], + [ + [ + 284, + 186 + ], + [ + 303, + 182 + ], + [ + 291, + 172 + ], + [ + 275, + 200 + ], + [ + 274, + 176 + ], + [ + 257, + 203 + ] + ], + [ + [ + 273, + 176 + ], + [ + 271, + 197 + ], + [ + 257, + 170 + ], + [ + 289, + 183 + ], + [ + 260, + 149 + ], + [ + 291, + 163 + ] + ], + [ + [ + 277, + 186 + ], + [ + 278, + 191 + ], + [ + 267, + 166 + ], + [ + 286, + 206 + ], + [ + 265, + 159 + ], + [ + 285, + 202 + ] + ] + ] + }, + { + "task_id": "0004775", + "id": "RH20T_cfg6_task_0078_user_0014_scene_0008_cfg_0006_104122064161#121", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close the microwave door'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004775/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 384, + 101 + ], + [ + 402, + 125 + ], + [ + 380, + 105 + ], + [ + 389, + 96 + ], + [ + 363, + 83 + ], + [ + 370, + 70 + ] + ], + [ + [ + 396, + 136 + ], + [ + 405, + 112 + ], + [ + 402, + 157 + ], + [ + 390, + 116 + ], + [ + 393, + 179 + ], + [ + 383, + 137 + ] + ], + [ + [ + 414, + 155 + ], + [ + 443, + 153 + ], + [ + 404, + 165 + ], + [ + 425, + 143 + ], + [ + 376, + 167 + ], + [ + 394, + 144 + ] + ], + [ + [ + 396, + 121 + ], + [ + 402, + 107 + ], + [ + 391, + 92 + ], + [ + 399, + 147 + ], + [ + 384, + 107 + ], + [ + 393, + 164 + ] + ] + ] + }, + { + "task_id": "0004776", + "id": "RH20T_cfg6_task_0129_user_0020_scene_0009_cfg_0006_104122061330#215", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004776/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 382, + 111 + ], + [ + 375, + 118 + ], + [ + 372, + 132 + ], + [ + 393, + 88 + ], + [ + 379, + 125 + ], + [ + 399, + 82 + ] + ], + [ + [ + 386, + 131 + ], + [ + 411, + 123 + ], + [ + 387, + 123 + ], + [ + 385, + 138 + ], + [ + 360, + 132 + ], + [ + 361, + 146 + ] + ], + [ + [ + 397, + 123 + ], + [ + 388, + 105 + ], + [ + 400, + 105 + ], + [ + 394, + 142 + ], + [ + 409, + 122 + ], + [ + 403, + 159 + ] + ], + [ + [ + 394, + 108 + ], + [ + 402, + 91 + ], + [ + 384, + 128 + ], + [ + 403, + 89 + ], + [ + 377, + 142 + ], + [ + 395, + 105 + ] + ] + ] + }, + { + "task_id": "0004777", + "id": "RH20T_cfg6_task_0082_user_0020_scene_0001_cfg_0006_104122064161#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004777/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 400, + 159 + ], + [ + 389, + 141 + ], + [ + 385, + 152 + ], + [ + 414, + 165 + ], + [ + 398, + 171 + ], + [ + 426, + 182 + ] + ], + [ + [ + 385, + 170 + ], + [ + 383, + 196 + ], + [ + 383, + 176 + ], + [ + 386, + 162 + ], + [ + 385, + 152 + ], + [ + 388, + 135 + ] + ], + [ + [ + 390, + 155 + ], + [ + 418, + 158 + ], + [ + 390, + 164 + ], + [ + 389, + 147 + ], + [ + 361, + 161 + ], + [ + 363, + 144 + ] + ], + [ + [ + 381, + 184 + ], + [ + 358, + 168 + ], + [ + 379, + 191 + ], + [ + 383, + 176 + ], + [ + 401, + 206 + ], + [ + 408, + 193 + ] + ] + ] + }, + { + "task_id": "0004778", + "id": "RH20T_cfg6_task_0082_user_0020_scene_0001_cfg_0006_104122064161#444", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004778/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 360, + 193 + ], + [ + 341, + 185 + ], + [ + 367, + 170 + ], + [ + 354, + 216 + ], + [ + 389, + 178 + ], + [ + 374, + 226 + ] + ], + [ + [ + 341, + 201 + ], + [ + 319, + 213 + ], + [ + 346, + 200 + ], + [ + 337, + 203 + ], + [ + 369, + 188 + ], + [ + 358, + 192 + ] + ], + [ + [ + 347, + 197 + ], + [ + 343, + 219 + ], + [ + 347, + 213 + ], + [ + 346, + 180 + ], + [ + 350, + 194 + ], + [ + 350, + 160 + ] + ], + [ + [ + 339, + 175 + ], + [ + 340, + 154 + ], + [ + 324, + 190 + ], + [ + 352, + 160 + ], + [ + 323, + 209 + ], + [ + 350, + 179 + ] + ] + ] + }, + { + "task_id": "0004779", + "id": "RH20T_cfg3_task_0038_user_0010_scene_0005_cfg_0003_036422060909#197", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004779/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 257, + 119 + ], + [ + 280, + 121 + ], + [ + 255, + 148 + ], + [ + 260, + 90 + ], + [ + 234, + 145 + ], + [ + 239, + 89 + ] + ], + [ + [ + 258, + 138 + ], + [ + 257, + 117 + ], + [ + 271, + 121 + ], + [ + 243, + 156 + ], + [ + 272, + 141 + ], + [ + 245, + 176 + ] + ], + [ + [ + 276, + 133 + ], + [ + 294, + 153 + ], + [ + 289, + 130 + ], + [ + 261, + 137 + ], + [ + 272, + 112 + ], + [ + 242, + 117 + ] + ], + [ + [ + 241, + 140 + ], + [ + 261, + 154 + ], + [ + 231, + 163 + ], + [ + 252, + 116 + ], + [ + 213, + 149 + ], + [ + 233, + 103 + ] + ] + ] + }, + { + "task_id": "0004780", + "id": "RH20T_cfg4_task_0111_user_0007_scene_0001_cfg_0004_104122062823#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object with the corresponding shape into the corresponding hole'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004780/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 447, + 73 + ], + [ + 432, + 97 + ], + [ + 420, + 60 + ], + [ + 474, + 85 + ], + [ + 434, + 33 + ], + [ + 491, + 59 + ] + ], + [ + [ + 453, + 68 + ], + [ + 456, + 89 + ], + [ + 423, + 72 + ], + [ + 484, + 64 + ], + [ + 421, + 53 + ], + [ + 480, + 44 + ] + ], + [ + [ + 420, + 80 + ], + [ + 422, + 53 + ], + [ + 451, + 84 + ], + [ + 390, + 77 + ], + [ + 447, + 109 + ], + [ + 389, + 101 + ] + ], + [ + [ + 431, + 73 + ], + [ + 440, + 62 + ], + [ + 422, + 97 + ], + [ + 441, + 47 + ], + [ + 413, + 109 + ], + [ + 432, + 58 + ] + ] + ] + }, + { + "task_id": "0004781", + "id": "RH20T_cfg5_task_0001_user_0020_scene_0001_cfg_0005_036422060215#270", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004781/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 269, + 143 + ], + [ + 274, + 130 + ], + [ + 254, + 128 + ], + [ + 283, + 159 + ], + [ + 250, + 141 + ], + [ + 278, + 171 + ] + ], + [ + [ + 286, + 132 + ], + [ + 302, + 120 + ], + [ + 281, + 136 + ], + [ + 291, + 128 + ], + [ + 264, + 148 + ], + [ + 274, + 141 + ] + ], + [ + [ + 263, + 124 + ], + [ + 248, + 126 + ], + [ + 278, + 113 + ], + [ + 249, + 133 + ], + [ + 293, + 111 + ], + [ + 264, + 131 + ] + ], + [ + [ + 276, + 136 + ], + [ + 281, + 118 + ], + [ + 295, + 132 + ], + [ + 257, + 141 + ], + [ + 291, + 149 + ], + [ + 253, + 158 + ] + ] + ] + }, + { + "task_id": "0004782", + "id": "RH20T_cfg7_task_0038_user_0020_scene_0006_cfg_0007_104122061018#175", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004782/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 127 + ], + [ + 254, + 124 + ], + [ + 292, + 130 + ], + [ + 278, + 124 + ], + [ + 321, + 134 + ], + [ + 312, + 128 + ] + ], + [ + [ + 285, + 115 + ], + [ + 254, + 121 + ], + [ + 291, + 148 + ], + [ + 279, + 83 + ], + [ + 321, + 142 + ], + [ + 310, + 78 + ] + ], + [ + [ + 291, + 143 + ], + [ + 318, + 138 + ], + [ + 307, + 145 + ], + [ + 277, + 141 + ], + [ + 277, + 151 + ], + [ + 249, + 146 + ] + ], + [ + [ + 288, + 136 + ], + [ + 290, + 111 + ], + [ + 275, + 116 + ], + [ + 302, + 158 + ], + [ + 274, + 138 + ], + [ + 299, + 180 + ] + ] + ] + }, + { + "task_id": "0004786", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0002_cfg_0006_037522061512#397", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004786/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 278, + 97 + ], + [ + 272, + 87 + ], + [ + 272, + 80 + ], + [ + 284, + 113 + ], + [ + 279, + 90 + ], + [ + 291, + 124 + ] + ], + [ + [ + 283, + 110 + ], + [ + 283, + 90 + ], + [ + 293, + 103 + ], + [ + 272, + 117 + ], + [ + 293, + 123 + ], + [ + 272, + 138 + ] + ], + [ + [ + 271, + 119 + ], + [ + 265, + 100 + ], + [ + 267, + 128 + ], + [ + 275, + 110 + ], + [ + 272, + 147 + ], + [ + 281, + 131 + ] + ], + [ + [ + 273, + 104 + ], + [ + 252, + 111 + ], + [ + 281, + 123 + ], + [ + 265, + 84 + ], + [ + 302, + 117 + ], + [ + 287, + 77 + ] + ] + ] + }, + { + "task_id": "0004790", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_037522061512#203", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004790/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 298, + 74 + ], + [ + 299, + 60 + ], + [ + 276, + 68 + ], + [ + 319, + 80 + ], + [ + 274, + 83 + ], + [ + 319, + 95 + ] + ], + [ + [ + 293, + 80 + ], + [ + 289, + 99 + ], + [ + 298, + 92 + ], + [ + 288, + 69 + ], + [ + 303, + 72 + ], + [ + 292, + 50 + ] + ], + [ + [ + 296, + 69 + ], + [ + 307, + 55 + ], + [ + 292, + 52 + ], + [ + 299, + 86 + ], + [ + 279, + 67 + ], + [ + 288, + 100 + ] + ], + [ + [ + 315, + 81 + ], + [ + 315, + 61 + ], + [ + 324, + 71 + ], + [ + 304, + 92 + ], + [ + 324, + 89 + ], + [ + 304, + 110 + ] + ] + ] + }, + { + "task_id": "0004791", + "id": "RH20T_cfg1_task_0015_user_0008_scene_0002_cfg_0001_038522062547#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004791/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 412, + 174 + ], + [ + 390, + 186 + ], + [ + 425, + 195 + ], + [ + 399, + 152 + ], + [ + 448, + 184 + ], + [ + 422, + 139 + ] + ], + [ + [ + 379, + 157 + ], + [ + 390, + 179 + ], + [ + 386, + 161 + ], + [ + 371, + 154 + ], + [ + 376, + 141 + ], + [ + 360, + 132 + ] + ], + [ + [ + 397, + 152 + ], + [ + 401, + 141 + ], + [ + 400, + 176 + ], + [ + 394, + 130 + ], + [ + 396, + 185 + ], + [ + 390, + 141 + ] + ], + [ + [ + 418, + 157 + ], + [ + 438, + 172 + ], + [ + 418, + 147 + ], + [ + 419, + 168 + ], + [ + 399, + 133 + ], + [ + 398, + 152 + ] + ] + ] + }, + { + "task_id": "0004792", + "id": "RH20T_cfg2_task_0002_user_0013_scene_0006_cfg_0002_037522062165#129", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004792/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 202, + 204 + ], + [ + 187, + 228 + ], + [ + 210, + 210 + ], + [ + 193, + 197 + ], + [ + 223, + 187 + ], + [ + 208, + 172 + ] + ], + [ + [ + 203, + 209 + ], + [ + 205, + 235 + ], + [ + 202, + 221 + ], + [ + 205, + 195 + ], + [ + 200, + 197 + ], + [ + 203, + 169 + ] + ], + [ + [ + 218, + 214 + ], + [ + 214, + 198 + ], + [ + 195, + 202 + ], + [ + 241, + 227 + ], + [ + 200, + 217 + ], + [ + 243, + 242 + ] + ], + [ + [ + 224, + 209 + ], + [ + 196, + 216 + ], + [ + 214, + 202 + ], + [ + 233, + 215 + ], + [ + 244, + 196 + ], + [ + 260, + 209 + ] + ] + ] + }, + { + "task_id": "0004794", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0006_cfg_0003_104122062295#263", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004794/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 201, + 201 + ], + [ + 211, + 162 + ], + [ + 186, + 175 + ], + [ + 219, + 231 + ], + [ + 178, + 210 + ], + [ + 209, + 268 + ] + ], + [ + [ + 164, + 219 + ], + [ + 215, + 218 + ], + [ + 167, + 192 + ], + [ + 162, + 242 + ], + [ + 107, + 193 + ], + [ + 112, + 244 + ] + ], + [ + [ + 193, + 189 + ], + [ + 216, + 147 + ], + [ + 143, + 164 + ], + [ + 239, + 213 + ], + [ + 114, + 211 + ], + [ + 216, + 260 + ] + ], + [ + [ + 243, + 246 + ], + [ + 254, + 223 + ], + [ + 220, + 209 + ], + [ + 264, + 281 + ], + [ + 204, + 234 + ], + [ + 253, + 312 + ] + ] + ] + }, + { + "task_id": "0004797", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0007_cfg_0004_038522062288#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004797/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 233, + 303 + ], + [ + 228, + 281 + ], + [ + 208, + 324 + ], + [ + 259, + 282 + ], + [ + 212, + 350 + ], + [ + 266, + 306 + ] + ], + [ + [ + 249, + 279 + ], + [ + 234, + 308 + ], + [ + 237, + 261 + ], + [ + 260, + 295 + ], + [ + 252, + 234 + ], + [ + 273, + 269 + ] + ], + [ + [ + 254, + 278 + ], + [ + 250, + 262 + ], + [ + 223, + 275 + ], + [ + 286, + 282 + ], + [ + 228, + 289 + ], + [ + 288, + 297 + ] + ], + [ + [ + 242, + 282 + ], + [ + 223, + 260 + ], + [ + 268, + 267 + ], + [ + 215, + 299 + ], + [ + 288, + 289 + ], + [ + 235, + 324 + ] + ] + ] + }, + { + "task_id": "0004798", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0007_cfg_0004_038522062288#226", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004798/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 264, + 171 + ], + [ + 270, + 171 + ], + [ + 291, + 161 + ], + [ + 237, + 182 + ], + [ + 284, + 162 + ], + [ + 234, + 182 + ] + ], + [ + [ + 295, + 153 + ], + [ + 269, + 164 + ], + [ + 305, + 146 + ], + [ + 285, + 160 + ], + [ + 331, + 136 + ], + [ + 309, + 150 + ] + ], + [ + [ + 276, + 147 + ], + [ + 275, + 125 + ], + [ + 248, + 145 + ], + [ + 304, + 149 + ], + [ + 251, + 165 + ], + [ + 304, + 169 + ] + ], + [ + [ + 254, + 134 + ], + [ + 279, + 145 + ], + [ + 270, + 131 + ], + [ + 241, + 137 + ], + [ + 242, + 119 + ], + [ + 215, + 127 + ] + ] + ] + }, + { + "task_id": "0004799", + "id": "RH20T_cfg4_task_0107_user_0015_scene_0007_cfg_0004_038522062288#359", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the garbage in the trash can'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004799/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 152, + 151 + ], + [ + 145, + 131 + ], + [ + 124, + 148 + ], + [ + 180, + 154 + ], + [ + 132, + 166 + ], + [ + 186, + 172 + ] + ], + [ + [ + 138, + 161 + ], + [ + 139, + 134 + ], + [ + 119, + 155 + ], + [ + 158, + 166 + ], + [ + 118, + 181 + ], + [ + 156, + 193 + ] + ], + [ + [ + 136, + 138 + ], + [ + 142, + 113 + ], + [ + 159, + 147 + ], + [ + 111, + 129 + ], + [ + 153, + 172 + ], + [ + 103, + 156 + ] + ], + [ + [ + 161, + 141 + ], + [ + 160, + 159 + ], + [ + 192, + 149 + ], + [ + 130, + 132 + ], + [ + 195, + 130 + ], + [ + 130, + 110 + ] + ] + ] + }, + { + "task_id": "0004801", + "id": "RH20T_cfg1_task_0030_user_0015_scene_0001_cfg_0001_819112070044#248", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour out the test tube'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004801/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 291, + 235 + ], + [ + 310, + 231 + ], + [ + 285, + 213 + ], + [ + 298, + 258 + ], + [ + 267, + 218 + ], + [ + 279, + 262 + ] + ], + [ + [ + 290, + 185 + ], + [ + 301, + 167 + ], + [ + 292, + 194 + ], + [ + 288, + 175 + ], + [ + 280, + 212 + ], + [ + 275, + 195 + ] + ], + [ + [ + 283, + 236 + ], + [ + 272, + 240 + ], + [ + 262, + 232 + ], + [ + 306, + 240 + ], + [ + 273, + 228 + ], + [ + 316, + 236 + ] + ], + [ + [ + 280, + 208 + ], + [ + 290, + 187 + ], + [ + 282, + 202 + ], + [ + 279, + 214 + ], + [ + 272, + 221 + ], + [ + 269, + 236 + ] + ] + ] + }, + { + "task_id": "0004802", + "id": "RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_104122061602#151", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004802/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 454, + 245 + ], + [ + 439, + 268 + ], + [ + 465, + 254 + ], + [ + 442, + 235 + ], + [ + 479, + 232 + ], + [ + 457, + 211 + ] + ], + [ + [ + 443, + 219 + ], + [ + 451, + 199 + ], + [ + 462, + 239 + ], + [ + 426, + 201 + ], + [ + 454, + 257 + ], + [ + 420, + 219 + ] + ], + [ + [ + 442, + 213 + ], + [ + 456, + 192 + ], + [ + 462, + 227 + ], + [ + 423, + 198 + ], + [ + 449, + 248 + ], + [ + 409, + 220 + ] + ], + [ + [ + 438, + 230 + ], + [ + 432, + 229 + ], + [ + 410, + 231 + ], + [ + 467, + 229 + ], + [ + 415, + 233 + ], + [ + 477, + 230 + ] + ] + ] + }, + { + "task_id": "0004803", + "id": "RH20T_cfg6_task_0057_user_0014_scene_0008_cfg_0006_104122061602#177", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004803/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 455, + 220 + ], + [ + 463, + 201 + ], + [ + 472, + 240 + ], + [ + 439, + 202 + ], + [ + 463, + 258 + ], + [ + 432, + 220 + ] + ], + [ + [ + 473, + 214 + ], + [ + 454, + 233 + ], + [ + 484, + 217 + ], + [ + 463, + 210 + ], + [ + 505, + 197 + ], + [ + 482, + 192 + ] + ], + [ + [ + 458, + 218 + ], + [ + 478, + 230 + ], + [ + 480, + 213 + ], + [ + 438, + 222 + ], + [ + 458, + 200 + ], + [ + 417, + 210 + ] + ], + [ + [ + 456, + 216 + ], + [ + 432, + 228 + ], + [ + 450, + 223 + ], + [ + 462, + 208 + ], + [ + 473, + 212 + ], + [ + 488, + 195 + ] + ] + ] + }, + { + "task_id": "0004806", + "id": "RH20T_cfg6_task_0020_user_0014_scene_0001_cfg_0006_104122061018#226", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004806/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 482, + 266 + ], + [ + 449, + 274 + ], + [ + 473, + 233 + ], + [ + 492, + 299 + ], + [ + 507, + 224 + ], + [ + 526, + 291 + ] + ], + [ + [ + 499, + 253 + ], + [ + 470, + 257 + ], + [ + 521, + 254 + ], + [ + 480, + 252 + ], + [ + 551, + 250 + ], + [ + 506, + 249 + ] + ], + [ + [ + 527, + 254 + ], + [ + 504, + 259 + ], + [ + 528, + 293 + ], + [ + 525, + 216 + ], + [ + 549, + 287 + ], + [ + 546, + 214 + ] + ], + [ + [ + 483, + 263 + ], + [ + 486, + 237 + ], + [ + 474, + 238 + ], + [ + 494, + 291 + ], + [ + 472, + 261 + ], + [ + 491, + 314 + ] + ] + ] + }, + { + "task_id": "0004807", + "id": "RH20T_cfg5_task_0083_user_0007_scene_0010_cfg_0005_036422060215#288", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Swat with a flyswatter'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004807/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 324, + 220 + ], + [ + 330, + 205 + ], + [ + 343, + 215 + ], + [ + 305, + 226 + ], + [ + 338, + 229 + ], + [ + 300, + 240 + ] + ], + [ + [ + 336, + 222 + ], + [ + 320, + 240 + ], + [ + 317, + 206 + ], + [ + 354, + 238 + ], + [ + 333, + 188 + ], + [ + 369, + 220 + ] + ], + [ + [ + 315, + 209 + ], + [ + 308, + 189 + ], + [ + 329, + 209 + ], + [ + 301, + 209 + ], + [ + 336, + 228 + ], + [ + 309, + 230 + ] + ], + [ + [ + 339, + 237 + ], + [ + 330, + 236 + ], + [ + 318, + 234 + ], + [ + 360, + 241 + ], + [ + 326, + 234 + ], + [ + 367, + 241 + ] + ] + ] + }, + { + "task_id": "0004808", + "id": "RH20T_cfg5_task_0075_user_0007_scene_0004_cfg_0005_037522062165#238", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Finish setting up the starting position of a chessboard that is almost arranged'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004808/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 204 + ], + [ + 322, + 190 + ], + [ + 301, + 188 + ], + [ + 338, + 221 + ], + [ + 300, + 201 + ], + [ + 335, + 234 + ] + ], + [ + [ + 342, + 206 + ], + [ + 360, + 190 + ], + [ + 359, + 209 + ], + [ + 327, + 204 + ], + [ + 340, + 227 + ], + [ + 309, + 220 + ] + ], + [ + [ + 313, + 230 + ], + [ + 309, + 205 + ], + [ + 293, + 240 + ], + [ + 332, + 220 + ], + [ + 296, + 264 + ], + [ + 335, + 244 + ] + ], + [ + [ + 292, + 187 + ], + [ + 292, + 166 + ], + [ + 314, + 199 + ], + [ + 272, + 177 + ], + [ + 314, + 219 + ], + [ + 273, + 196 + ] + ] + ] + }, + { + "task_id": "0004809", + "id": "RH20T_cfg5_task_0075_user_0007_scene_0004_cfg_0005_037522062165#582", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Finish setting up the starting position of a chessboard that is almost arranged'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004809/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 214, + 355 + ], + [ + 239, + 341 + ], + [ + 229, + 350 + ], + [ + 201, + 360 + ], + [ + 202, + 360 + ], + [ + 175, + 360 + ] + ], + [ + [ + 211, + 334 + ], + [ + 210, + 360 + ], + [ + 234, + 339 + ], + [ + 190, + 329 + ], + [ + 234, + 303 + ], + [ + 191, + 296 + ] + ], + [ + [ + 217, + 341 + ], + [ + 213, + 330 + ], + [ + 206, + 314 + ], + [ + 229, + 360 + ], + [ + 210, + 326 + ], + [ + 232, + 360 + ] + ], + [ + [ + 235, + 341 + ], + [ + 206, + 360 + ], + [ + 224, + 331 + ], + [ + 247, + 352 + ], + [ + 250, + 314 + ], + [ + 275, + 333 + ] + ] + ] + }, + { + "task_id": "0004810", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0008_cfg_0006_104122061330#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004810/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 87 + ], + [ + 314, + 67 + ], + [ + 341, + 77 + ], + [ + 297, + 99 + ], + [ + 345, + 96 + ], + [ + 303, + 117 + ] + ], + [ + [ + 308, + 72 + ], + [ + 325, + 90 + ], + [ + 308, + 75 + ], + [ + 308, + 69 + ], + [ + 291, + 57 + ], + [ + 289, + 50 + ] + ], + [ + [ + 326, + 102 + ], + [ + 320, + 89 + ], + [ + 315, + 123 + ], + [ + 337, + 82 + ], + [ + 321, + 134 + ], + [ + 341, + 94 + ] + ], + [ + [ + 296, + 109 + ], + [ + 272, + 105 + ], + [ + 294, + 107 + ], + [ + 299, + 110 + ], + [ + 320, + 110 + ], + [ + 322, + 114 + ] + ] + ] + }, + { + "task_id": "0004811", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0008_cfg_0006_104122061330#238", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004811/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 278, + 123 + ], + [ + 279, + 140 + ], + [ + 258, + 112 + ], + [ + 299, + 134 + ], + [ + 256, + 94 + ], + [ + 299, + 116 + ] + ], + [ + [ + 286, + 117 + ], + [ + 275, + 113 + ], + [ + 280, + 141 + ], + [ + 291, + 93 + ], + [ + 290, + 143 + ], + [ + 301, + 98 + ] + ], + [ + [ + 281, + 101 + ], + [ + 259, + 96 + ], + [ + 290, + 110 + ], + [ + 270, + 91 + ], + [ + 311, + 114 + ], + [ + 293, + 96 + ] + ], + [ + [ + 302, + 101 + ], + [ + 318, + 104 + ], + [ + 302, + 75 + ], + [ + 303, + 125 + ], + [ + 287, + 73 + ], + [ + 289, + 121 + ] + ] + ] + }, + { + "task_id": "0004812", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0008_cfg_0006_104122061330#400", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004812/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 335, + 109 + ], + [ + 348, + 93 + ], + [ + 313, + 98 + ], + [ + 357, + 121 + ], + [ + 298, + 115 + ], + [ + 343, + 140 + ] + ], + [ + [ + 327, + 138 + ], + [ + 320, + 123 + ], + [ + 317, + 159 + ], + [ + 337, + 116 + ], + [ + 325, + 171 + ], + [ + 343, + 130 + ] + ], + [ + [ + 322, + 135 + ], + [ + 317, + 131 + ], + [ + 308, + 156 + ], + [ + 335, + 114 + ], + [ + 313, + 159 + ], + [ + 338, + 118 + ] + ], + [ + [ + 313, + 149 + ], + [ + 313, + 127 + ], + [ + 339, + 147 + ], + [ + 287, + 152 + ], + [ + 339, + 170 + ], + [ + 286, + 175 + ] + ] + ] + }, + { + "task_id": "0004813", + "id": "RH20T_cfg6_task_0037_user_0014_scene_0008_cfg_0006_104122061330#679", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004813/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 275, + 233 + ], + [ + 288, + 257 + ], + [ + 249, + 247 + ], + [ + 299, + 220 + ], + [ + 237, + 223 + ], + [ + 286, + 198 + ] + ], + [ + [ + 286, + 209 + ], + [ + 272, + 207 + ], + [ + 278, + 235 + ], + [ + 295, + 182 + ], + [ + 291, + 236 + ], + [ + 307, + 185 + ] + ], + [ + [ + 281, + 229 + ], + [ + 300, + 242 + ], + [ + 289, + 205 + ], + [ + 274, + 253 + ], + [ + 271, + 194 + ], + [ + 256, + 240 + ] + ], + [ + [ + 318, + 189 + ], + [ + 314, + 217 + ], + [ + 342, + 194 + ], + [ + 293, + 184 + ], + [ + 345, + 167 + ], + [ + 296, + 156 + ] + ] + ] + }, + { + "task_id": "0004814", + "id": "RH20T_cfg1_task_0015_user_0001_scene_0007_cfg_0001_038522062547#261", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004814/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 469, + 160 + ], + [ + 473, + 160 + ], + [ + 492, + 144 + ], + [ + 448, + 175 + ], + [ + 489, + 142 + ], + [ + 443, + 175 + ] + ], + [ + [ + 464, + 131 + ], + [ + 468, + 155 + ], + [ + 466, + 122 + ], + [ + 461, + 139 + ], + [ + 461, + 97 + ], + [ + 457, + 116 + ] + ], + [ + [ + 492, + 163 + ], + [ + 511, + 180 + ], + [ + 475, + 167 + ], + [ + 508, + 160 + ], + [ + 457, + 151 + ], + [ + 490, + 145 + ] + ], + [ + [ + 467, + 149 + ], + [ + 472, + 138 + ], + [ + 493, + 147 + ], + [ + 441, + 150 + ], + [ + 487, + 157 + ], + [ + 437, + 160 + ] + ] + ] + }, + { + "task_id": "0004816", + "id": "RH20T_cfg7_task_0015_user_0014_scene_0008_cfg_0007_104122060811#144", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004816/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 396, + 114 + ], + [ + 386, + 107 + ], + [ + 406, + 110 + ], + [ + 386, + 118 + ], + [ + 416, + 117 + ], + [ + 395, + 125 + ] + ], + [ + [ + 384, + 117 + ], + [ + 384, + 110 + ], + [ + 395, + 125 + ], + [ + 373, + 109 + ], + [ + 395, + 133 + ], + [ + 372, + 116 + ] + ], + [ + [ + 400, + 121 + ], + [ + 414, + 121 + ], + [ + 400, + 127 + ], + [ + 399, + 115 + ], + [ + 385, + 127 + ], + [ + 385, + 116 + ] + ], + [ + [ + 391, + 124 + ], + [ + 394, + 113 + ], + [ + 403, + 122 + ], + [ + 378, + 126 + ], + [ + 399, + 133 + ], + [ + 375, + 136 + ] + ] + ] + }, + { + "task_id": "0004817", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0002_cfg_0001_039422060546#335", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004817/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 197 + ], + [ + 341, + 189 + ], + [ + 310, + 230 + ], + [ + 301, + 163 + ], + [ + 275, + 238 + ], + [ + 263, + 170 + ] + ], + [ + [ + 328, + 187 + ], + [ + 331, + 169 + ], + [ + 318, + 216 + ], + [ + 337, + 160 + ], + [ + 316, + 230 + ], + [ + 334, + 176 + ] + ], + [ + [ + 318, + 169 + ], + [ + 345, + 182 + ], + [ + 329, + 140 + ], + [ + 308, + 197 + ], + [ + 303, + 127 + ], + [ + 283, + 184 + ] + ], + [ + [ + 321, + 208 + ], + [ + 322, + 176 + ], + [ + 319, + 207 + ], + [ + 322, + 209 + ], + [ + 318, + 237 + ], + [ + 321, + 243 + ] + ] + ] + }, + { + "task_id": "0004819", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0002_cfg_0001_039422060546#751", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004819/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 389, + 38 + ], + [ + 387, + 9 + ], + [ + 415, + 31 + ], + [ + 365, + 45 + ], + [ + 416, + 62 + ], + [ + 366, + 74 + ] + ], + [ + [ + 375, + 58 + ], + [ + 394, + 41 + ], + [ + 371, + 70 + ], + [ + 380, + 44 + ], + [ + 353, + 87 + ], + [ + 360, + 62 + ] + ], + [ + [ + 383, + 38 + ], + [ + 405, + 55 + ], + [ + 395, + 15 + ], + [ + 370, + 61 + ], + [ + 372, + 0 + ], + [ + 347, + 43 + ] + ], + [ + [ + 368, + 69 + ], + [ + 368, + 46 + ], + [ + 364, + 89 + ], + [ + 372, + 50 + ], + [ + 364, + 110 + ], + [ + 372, + 71 + ] + ] + ] + }, + { + "task_id": "0004821", + "id": "RH20T_cfg7_task_0039_user_0014_scene_0008_cfg_0007_104122060811#106", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004821/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 407, + 119 + ], + [ + 404, + 120 + ], + [ + 402, + 106 + ], + [ + 412, + 132 + ], + [ + 405, + 104 + ], + [ + 416, + 131 + ] + ], + [ + [ + 420, + 111 + ], + [ + 424, + 112 + ], + [ + 432, + 118 + ], + [ + 407, + 104 + ], + [ + 428, + 117 + ], + [ + 404, + 103 + ] + ], + [ + [ + 406, + 131 + ], + [ + 412, + 144 + ], + [ + 395, + 138 + ], + [ + 417, + 124 + ], + [ + 389, + 126 + ], + [ + 411, + 110 + ] + ], + [ + [ + 408, + 124 + ], + [ + 410, + 116 + ], + [ + 421, + 120 + ], + [ + 394, + 129 + ], + [ + 418, + 128 + ], + [ + 392, + 136 + ] + ] + ] + }, + { + "task_id": "0004823", + "id": "RH20T_cfg7_task_0031_user_0020_scene_0007_cfg_0007_104122061330#142", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Cover the box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004823/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 282, + 204 + ], + [ + 289, + 227 + ], + [ + 303, + 191 + ], + [ + 260, + 218 + ], + [ + 297, + 167 + ], + [ + 253, + 194 + ] + ], + [ + [ + 304, + 225 + ], + [ + 301, + 206 + ], + [ + 289, + 245 + ], + [ + 318, + 207 + ], + [ + 292, + 263 + ], + [ + 319, + 225 + ] + ], + [ + [ + 320, + 237 + ], + [ + 344, + 250 + ], + [ + 327, + 235 + ], + [ + 312, + 240 + ], + [ + 305, + 223 + ], + [ + 287, + 226 + ] + ], + [ + [ + 318, + 232 + ], + [ + 343, + 241 + ], + [ + 327, + 244 + ], + [ + 308, + 219 + ], + [ + 305, + 235 + ], + [ + 285, + 210 + ] + ] + ] + }, + { + "task_id": "0004824", + "id": "RH20T_cfg5_task_0045_user_0010_scene_0004_cfg_0005_104122061850#389", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the brush on the pen rack'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004824/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 386, + 158 + ], + [ + 386, + 144 + ], + [ + 375, + 180 + ], + [ + 396, + 138 + ], + [ + 376, + 192 + ], + [ + 396, + 152 + ] + ], + [ + [ + 408, + 168 + ], + [ + 386, + 170 + ], + [ + 416, + 146 + ], + [ + 401, + 189 + ], + [ + 437, + 145 + ], + [ + 421, + 187 + ] + ], + [ + [ + 384, + 173 + ], + [ + 378, + 197 + ], + [ + 371, + 165 + ], + [ + 396, + 180 + ], + [ + 378, + 139 + ], + [ + 402, + 157 + ] + ], + [ + [ + 407, + 168 + ], + [ + 389, + 174 + ], + [ + 428, + 168 + ], + [ + 388, + 167 + ], + [ + 444, + 161 + ], + [ + 404, + 161 + ] + ] + ] + }, + { + "task_id": "0004826", + "id": "RH20T_cfg7_task_0029_user_0020_scene_0002_cfg_0007_104122060811#215", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004826/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 336, + 112 + ], + [ + 343, + 119 + ], + [ + 348, + 118 + ], + [ + 325, + 105 + ], + [ + 342, + 110 + ], + [ + 318, + 97 + ] + ], + [ + [ + 337, + 120 + ], + [ + 324, + 116 + ], + [ + 328, + 130 + ], + [ + 347, + 111 + ], + [ + 340, + 133 + ], + [ + 360, + 115 + ] + ], + [ + [ + 329, + 126 + ], + [ + 334, + 131 + ], + [ + 315, + 134 + ], + [ + 342, + 117 + ], + [ + 309, + 129 + ], + [ + 336, + 110 + ] + ], + [ + [ + 322, + 136 + ], + [ + 323, + 121 + ], + [ + 330, + 132 + ], + [ + 314, + 139 + ], + [ + 328, + 148 + ], + [ + 312, + 154 + ] + ] + ] + }, + { + "task_id": "0004827", + "id": "RH20T_cfg7_task_0039_user_0014_scene_0008_cfg_0007_104122064161#109", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004827/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 360, + 178 + ], + [ + 350, + 154 + ], + [ + 361, + 160 + ], + [ + 359, + 197 + ], + [ + 370, + 182 + ], + [ + 369, + 219 + ] + ], + [ + [ + 370, + 195 + ], + [ + 376, + 175 + ], + [ + 374, + 217 + ], + [ + 366, + 175 + ], + [ + 369, + 235 + ], + [ + 361, + 193 + ] + ], + [ + [ + 353, + 170 + ], + [ + 373, + 162 + ], + [ + 374, + 181 + ], + [ + 331, + 157 + ], + [ + 355, + 188 + ], + [ + 313, + 166 + ] + ], + [ + [ + 387, + 185 + ], + [ + 357, + 187 + ], + [ + 387, + 163 + ], + [ + 386, + 205 + ], + [ + 418, + 160 + ], + [ + 414, + 202 + ] + ] + ] + }, + { + "task_id": "0004829", + "id": "RH20T_cfg5_task_0123_user_0010_scene_0004_cfg_0005_037522062165#613", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the knife on the knife rack'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004829/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 236, + 275 + ], + [ + 207, + 285 + ], + [ + 248, + 270 + ], + [ + 224, + 279 + ], + [ + 276, + 260 + ], + [ + 250, + 270 + ] + ], + [ + [ + 246, + 274 + ], + [ + 273, + 263 + ], + [ + 232, + 247 + ], + [ + 261, + 302 + ], + [ + 203, + 258 + ], + [ + 232, + 316 + ] + ], + [ + [ + 242, + 265 + ], + [ + 224, + 268 + ], + [ + 266, + 279 + ], + [ + 217, + 250 + ], + [ + 287, + 276 + ], + [ + 237, + 245 + ] + ], + [ + [ + 239, + 246 + ], + [ + 251, + 238 + ], + [ + 228, + 218 + ], + [ + 250, + 274 + ], + [ + 218, + 226 + ], + [ + 239, + 279 + ] + ] + ] + }, + { + "task_id": "0004830", + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_037522061512#92", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004830/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 258, + 90 + ], + [ + 239, + 76 + ], + [ + 252, + 90 + ], + [ + 264, + 90 + ], + [ + 272, + 104 + ], + [ + 282, + 104 + ] + ], + [ + [ + 261, + 72 + ], + [ + 244, + 70 + ], + [ + 277, + 67 + ], + [ + 244, + 77 + ], + [ + 295, + 69 + ], + [ + 262, + 80 + ] + ], + [ + [ + 258, + 82 + ], + [ + 255, + 61 + ], + [ + 270, + 73 + ], + [ + 244, + 91 + ], + [ + 273, + 92 + ], + [ + 247, + 112 + ] + ], + [ + [ + 262, + 60 + ], + [ + 279, + 71 + ], + [ + 246, + 64 + ], + [ + 277, + 55 + ], + [ + 230, + 53 + ], + [ + 261, + 44 + ] + ] + ] + }, + { + "task_id": "0004831", + "id": "RH20T_cfg6_task_0003_user_0014_scene_0010_cfg_0006_037522061512#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004831/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 90 + ], + [ + 272, + 71 + ], + [ + 302, + 82 + ], + [ + 265, + 99 + ], + [ + 313, + 101 + ], + [ + 277, + 119 + ] + ], + [ + [ + 258, + 103 + ], + [ + 260, + 80 + ], + [ + 270, + 101 + ], + [ + 245, + 105 + ], + [ + 268, + 122 + ], + [ + 244, + 128 + ] + ], + [ + [ + 281, + 83 + ], + [ + 263, + 87 + ], + [ + 291, + 78 + ], + [ + 270, + 89 + ], + [ + 310, + 74 + ], + [ + 289, + 85 + ] + ], + [ + [ + 242, + 96 + ], + [ + 230, + 89 + ], + [ + 222, + 105 + ], + [ + 260, + 88 + ], + [ + 235, + 112 + ], + [ + 274, + 95 + ] + ] + ] + }, + { + "task_id": "0004835", + "id": "RH20T_cfg4_task_0125_user_0007_scene_0004_cfg_0004_104422070011#120", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull up the lever'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004835/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 359, + 175 + ], + [ + 348, + 156 + ], + [ + 382, + 163 + ], + [ + 335, + 188 + ], + [ + 394, + 183 + ], + [ + 345, + 208 + ] + ], + [ + [ + 345, + 169 + ], + [ + 356, + 156 + ], + [ + 326, + 184 + ], + [ + 363, + 155 + ], + [ + 318, + 196 + ], + [ + 353, + 167 + ] + ], + [ + [ + 327, + 152 + ], + [ + 339, + 175 + ], + [ + 318, + 169 + ], + [ + 338, + 134 + ], + [ + 308, + 147 + ], + [ + 326, + 111 + ] + ], + [ + [ + 355, + 180 + ], + [ + 362, + 161 + ], + [ + 377, + 194 + ], + [ + 333, + 166 + ], + [ + 370, + 214 + ], + [ + 324, + 185 + ] + ] + ] + }, + { + "task_id": "0004836", + "id": "RH20T_cfg4_task_0125_user_0007_scene_0004_cfg_0004_104422070011#185", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull up the lever'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004836/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 335, + 107 + ], + [ + 332, + 105 + ], + [ + 355, + 85 + ], + [ + 316, + 129 + ], + [ + 356, + 88 + ], + [ + 319, + 130 + ] + ], + [ + [ + 333, + 135 + ], + [ + 350, + 121 + ], + [ + 313, + 146 + ], + [ + 351, + 124 + ], + [ + 298, + 158 + ], + [ + 335, + 136 + ] + ], + [ + [ + 338, + 136 + ], + [ + 344, + 163 + ], + [ + 354, + 128 + ], + [ + 321, + 145 + ], + [ + 349, + 101 + ], + [ + 314, + 115 + ] + ], + [ + [ + 334, + 148 + ], + [ + 307, + 144 + ], + [ + 326, + 173 + ], + [ + 342, + 124 + ], + [ + 355, + 178 + ], + [ + 369, + 127 + ] + ] + ] + }, + { + "task_id": "0004840", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0002_cfg_0006_104122063633#567", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004840/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 204, + 117 + ], + [ + 185, + 120 + ], + [ + 210, + 126 + ], + [ + 198, + 108 + ], + [ + 229, + 124 + ], + [ + 216, + 106 + ] + ], + [ + [ + 190, + 111 + ], + [ + 190, + 93 + ], + [ + 194, + 101 + ], + [ + 184, + 121 + ], + [ + 194, + 117 + ], + [ + 184, + 138 + ] + ], + [ + [ + 197, + 102 + ], + [ + 196, + 81 + ], + [ + 200, + 103 + ], + [ + 195, + 102 + ], + [ + 202, + 124 + ], + [ + 197, + 122 + ] + ], + [ + [ + 203, + 122 + ], + [ + 222, + 122 + ], + [ + 201, + 142 + ], + [ + 204, + 102 + ], + [ + 181, + 141 + ], + [ + 185, + 102 + ] + ] + ] + }, + { + "task_id": "0004841", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0002_cfg_0006_104122063633#704", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004841/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 170, + 131 + ], + [ + 165, + 127 + ], + [ + 184, + 116 + ], + [ + 155, + 145 + ], + [ + 189, + 120 + ], + [ + 160, + 150 + ] + ], + [ + [ + 191, + 115 + ], + [ + 171, + 116 + ], + [ + 183, + 117 + ], + [ + 199, + 113 + ], + [ + 204, + 116 + ], + [ + 218, + 111 + ] + ], + [ + [ + 182, + 128 + ], + [ + 187, + 117 + ], + [ + 185, + 145 + ], + [ + 180, + 110 + ], + [ + 180, + 157 + ], + [ + 174, + 120 + ] + ], + [ + [ + 174, + 123 + ], + [ + 193, + 123 + ], + [ + 170, + 103 + ], + [ + 178, + 142 + ], + [ + 151, + 104 + ], + [ + 159, + 142 + ] + ] + ] + }, + { + "task_id": "0004843", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0009_cfg_0005_036422060215#180", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004843/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 102 + ], + [ + 284, + 102 + ], + [ + 297, + 82 + ], + [ + 273, + 121 + ], + [ + 298, + 81 + ], + [ + 273, + 122 + ] + ], + [ + [ + 295, + 117 + ], + [ + 316, + 120 + ], + [ + 300, + 129 + ], + [ + 288, + 103 + ], + [ + 281, + 126 + ], + [ + 267, + 100 + ] + ], + [ + [ + 261, + 98 + ], + [ + 265, + 111 + ], + [ + 254, + 117 + ], + [ + 269, + 78 + ], + [ + 251, + 104 + ], + [ + 265, + 66 + ] + ], + [ + [ + 273, + 110 + ], + [ + 275, + 90 + ], + [ + 295, + 110 + ], + [ + 250, + 109 + ], + [ + 292, + 128 + ], + [ + 248, + 128 + ] + ] + ] + }, + { + "task_id": "0004844", + "id": "RH20T_cfg5_task_0073_user_0007_scene_0009_cfg_0005_036422060215#433", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004844/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 241, + 209 + ], + [ + 244, + 192 + ], + [ + 265, + 216 + ], + [ + 218, + 201 + ], + [ + 261, + 232 + ], + [ + 217, + 217 + ] + ], + [ + [ + 228, + 183 + ], + [ + 209, + 198 + ], + [ + 213, + 163 + ], + [ + 244, + 203 + ], + [ + 232, + 148 + ], + [ + 263, + 188 + ] + ], + [ + [ + 248, + 188 + ], + [ + 254, + 194 + ], + [ + 271, + 195 + ], + [ + 225, + 182 + ], + [ + 265, + 189 + ], + [ + 221, + 176 + ] + ], + [ + [ + 218, + 192 + ], + [ + 194, + 184 + ], + [ + 212, + 194 + ], + [ + 224, + 190 + ], + [ + 238, + 202 + ], + [ + 247, + 197 + ] + ] + ] + }, + { + "task_id": "0004852", + "id": "RH20T_cfg4_task_0065_user_0016_scene_0005_cfg_0004_104122062823#261", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004852/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 382, + 106 + ], + [ + 405, + 96 + ], + [ + 368, + 77 + ], + [ + 396, + 135 + ], + [ + 343, + 87 + ], + [ + 371, + 146 + ] + ], + [ + [ + 385, + 74 + ], + [ + 354, + 88 + ], + [ + 398, + 104 + ], + [ + 372, + 43 + ], + [ + 430, + 91 + ], + [ + 405, + 27 + ] + ], + [ + [ + 411, + 100 + ], + [ + 415, + 74 + ], + [ + 444, + 109 + ], + [ + 381, + 93 + ], + [ + 439, + 132 + ], + [ + 380, + 116 + ] + ], + [ + [ + 372, + 111 + ], + [ + 381, + 120 + ], + [ + 390, + 83 + ], + [ + 355, + 139 + ], + [ + 381, + 71 + ], + [ + 344, + 131 + ] + ] + ] + }, + { + "task_id": "0004854", + "id": "RH20T_cfg4_task_0065_user_0016_scene_0005_cfg_0004_104122062823#403", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004854/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 107 + ], + [ + 397, + 114 + ], + [ + 370, + 130 + ], + [ + 373, + 84 + ], + [ + 343, + 122 + ], + [ + 347, + 76 + ] + ], + [ + [ + 366, + 63 + ], + [ + 360, + 91 + ], + [ + 389, + 67 + ], + [ + 341, + 59 + ], + [ + 396, + 40 + ], + [ + 347, + 30 + ] + ], + [ + [ + 377, + 87 + ], + [ + 370, + 111 + ], + [ + 385, + 101 + ], + [ + 369, + 71 + ], + [ + 392, + 79 + ], + [ + 377, + 46 + ] + ], + [ + [ + 381, + 76 + ], + [ + 382, + 52 + ], + [ + 409, + 84 + ], + [ + 355, + 69 + ], + [ + 407, + 106 + ], + [ + 355, + 91 + ] + ] + ] + }, + { + "task_id": "0004855", + "id": "RH20T_cfg4_task_0065_user_0016_scene_0005_cfg_0004_104122062823#591", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004855/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 383, + 90 + ], + [ + 368, + 110 + ], + [ + 364, + 105 + ], + [ + 404, + 72 + ], + [ + 377, + 87 + ], + [ + 417, + 54 + ] + ], + [ + [ + 369, + 76 + ], + [ + 372, + 48 + ], + [ + 375, + 92 + ], + [ + 362, + 59 + ], + [ + 372, + 120 + ], + [ + 358, + 90 + ] + ], + [ + [ + 352, + 66 + ], + [ + 343, + 53 + ], + [ + 332, + 44 + ], + [ + 369, + 86 + ], + [ + 341, + 59 + ], + [ + 379, + 102 + ] + ], + [ + [ + 380, + 88 + ], + [ + 381, + 62 + ], + [ + 409, + 97 + ], + [ + 352, + 80 + ], + [ + 406, + 120 + ], + [ + 352, + 103 + ] + ] + ] + }, + { + "task_id": "0004856", + "id": "RH20T_cfg4_task_0065_user_0016_scene_0005_cfg_0004_104122062823#694", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004856/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 385, + 68 + ], + [ + 387, + 44 + ], + [ + 412, + 76 + ], + [ + 359, + 61 + ], + [ + 409, + 99 + ], + [ + 358, + 83 + ] + ], + [ + [ + 358, + 76 + ], + [ + 357, + 48 + ], + [ + 331, + 80 + ], + [ + 385, + 71 + ], + [ + 334, + 107 + ], + [ + 386, + 98 + ] + ], + [ + [ + 376, + 42 + ], + [ + 389, + 46 + ], + [ + 379, + 70 + ], + [ + 373, + 11 + ], + [ + 368, + 65 + ], + [ + 361, + 9 + ] + ], + [ + [ + 393, + 69 + ], + [ + 363, + 78 + ], + [ + 401, + 99 + ], + [ + 384, + 38 + ], + [ + 431, + 91 + ], + [ + 416, + 28 + ] + ] + ] + }, + { + "task_id": "0004857", + "id": "RH20T_cfg4_task_0065_user_0016_scene_0005_cfg_0004_104122062823#884", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004857/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 392, + 92 + ], + [ + 373, + 98 + ], + [ + 405, + 65 + ], + [ + 379, + 117 + ], + [ + 422, + 60 + ], + [ + 396, + 110 + ] + ], + [ + [ + 386, + 107 + ], + [ + 389, + 80 + ], + [ + 418, + 116 + ], + [ + 356, + 98 + ], + [ + 413, + 140 + ], + [ + 354, + 122 + ] + ], + [ + [ + 378, + 94 + ], + [ + 364, + 99 + ], + [ + 353, + 81 + ], + [ + 402, + 107 + ], + [ + 369, + 75 + ], + [ + 420, + 103 + ] + ], + [ + [ + 380, + 107 + ], + [ + 373, + 85 + ], + [ + 356, + 126 + ], + [ + 405, + 87 + ], + [ + 362, + 151 + ], + [ + 414, + 111 + ] + ] + ] + }, + { + "task_id": "0004858", + "id": "RH20T_cfg4_task_0065_user_0016_scene_0005_cfg_0004_104122062823#943", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004858/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 405, + 139 + ], + [ + 402, + 169 + ], + [ + 429, + 135 + ], + [ + 380, + 144 + ], + [ + 433, + 105 + ], + [ + 383, + 113 + ] + ], + [ + [ + 386, + 108 + ], + [ + 389, + 81 + ], + [ + 419, + 116 + ], + [ + 355, + 101 + ], + [ + 415, + 140 + ], + [ + 355, + 125 + ] + ], + [ + [ + 362, + 114 + ], + [ + 375, + 94 + ], + [ + 336, + 118 + ], + [ + 391, + 110 + ], + [ + 322, + 139 + ], + [ + 377, + 133 + ] + ], + [ + [ + 363, + 73 + ], + [ + 366, + 42 + ], + [ + 386, + 63 + ], + [ + 339, + 85 + ], + [ + 382, + 91 + ], + [ + 338, + 113 + ] + ] + ] + }, + { + "task_id": "0004859", + "id": "RH20T_cfg4_task_0065_user_0016_scene_0005_cfg_0004_104122062823#967", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004859/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 399, + 83 + ], + [ + 385, + 103 + ], + [ + 412, + 110 + ], + [ + 386, + 54 + ], + [ + 425, + 91 + ], + [ + 401, + 37 + ] + ], + [ + [ + 388, + 71 + ], + [ + 374, + 40 + ], + [ + 409, + 59 + ], + [ + 370, + 81 + ], + [ + 424, + 91 + ], + [ + 383, + 109 + ] + ], + [ + [ + 415, + 91 + ], + [ + 421, + 59 + ], + [ + 384, + 96 + ], + [ + 445, + 86 + ], + [ + 380, + 125 + ], + [ + 438, + 115 + ] + ], + [ + [ + 378, + 106 + ], + [ + 381, + 79 + ], + [ + 409, + 115 + ], + [ + 348, + 97 + ], + [ + 405, + 139 + ], + [ + 347, + 120 + ] + ] + ] + }, + { + "task_id": "0004863", + "id": "RH20T_cfg2_task_0003_user_0006_scene_0007_cfg_0002_037522062165#358", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004863/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 287, + 260 + ], + [ + 270, + 279 + ], + [ + 307, + 269 + ], + [ + 264, + 251 + ], + [ + 324, + 250 + ], + [ + 282, + 231 + ] + ], + [ + [ + 319, + 248 + ], + [ + 321, + 236 + ], + [ + 303, + 226 + ], + [ + 335, + 272 + ], + [ + 302, + 238 + ], + [ + 332, + 282 + ] + ], + [ + [ + 326, + 267 + ], + [ + 298, + 278 + ], + [ + 317, + 239 + ], + [ + 336, + 296 + ], + [ + 345, + 229 + ], + [ + 365, + 286 + ] + ], + [ + [ + 336, + 249 + ], + [ + 312, + 264 + ], + [ + 349, + 272 + ], + [ + 323, + 225 + ], + [ + 372, + 257 + ], + [ + 346, + 210 + ] + ] + ] + }, + { + "task_id": "0004864", + "id": "RH20T_cfg6_task_0062_user_0014_scene_0003_cfg_0006_104122061602#114", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the toilet paper on its holder'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004864/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 549, + 227 + ], + [ + 526, + 218 + ], + [ + 528, + 222 + ], + [ + 573, + 233 + ], + [ + 549, + 231 + ], + [ + 596, + 242 + ] + ], + [ + [ + 525, + 223 + ], + [ + 539, + 211 + ], + [ + 545, + 247 + ], + [ + 507, + 201 + ], + [ + 531, + 258 + ], + [ + 495, + 213 + ] + ], + [ + [ + 488, + 218 + ], + [ + 503, + 223 + ], + [ + 497, + 192 + ], + [ + 479, + 243 + ], + [ + 483, + 189 + ], + [ + 466, + 237 + ] + ], + [ + [ + 533, + 224 + ], + [ + 509, + 236 + ], + [ + 548, + 252 + ], + [ + 518, + 198 + ], + [ + 575, + 241 + ], + [ + 543, + 184 + ] + ] + ] + }, + { + "task_id": "0004865", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061018#253", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004865/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 328, + 112 + ], + [ + 322, + 84 + ], + [ + 317, + 114 + ], + [ + 341, + 110 + ], + [ + 323, + 141 + ], + [ + 348, + 139 + ] + ], + [ + [ + 327, + 121 + ], + [ + 350, + 132 + ], + [ + 314, + 143 + ], + [ + 341, + 97 + ], + [ + 290, + 132 + ], + [ + 317, + 85 + ] + ], + [ + [ + 322, + 132 + ], + [ + 326, + 156 + ], + [ + 307, + 143 + ], + [ + 335, + 121 + ], + [ + 302, + 117 + ], + [ + 331, + 96 + ] + ], + [ + [ + 346, + 132 + ], + [ + 353, + 113 + ], + [ + 319, + 137 + ], + [ + 371, + 128 + ], + [ + 313, + 155 + ], + [ + 363, + 145 + ] + ] + ] + }, + { + "task_id": "0004868", + "id": "RH20T_cfg6_task_0037_user_0020_scene_0004_cfg_0006_104122061018#1133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004868/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 258, + 137 + ], + [ + 249, + 156 + ], + [ + 263, + 114 + ], + [ + 254, + 159 + ], + [ + 271, + 97 + ], + [ + 262, + 141 + ] + ], + [ + [ + 279, + 122 + ], + [ + 282, + 99 + ], + [ + 267, + 103 + ], + [ + 291, + 141 + ], + [ + 265, + 124 + ], + [ + 288, + 162 + ] + ], + [ + [ + 289, + 137 + ], + [ + 303, + 125 + ], + [ + 312, + 152 + ], + [ + 264, + 122 + ], + [ + 299, + 162 + ], + [ + 253, + 133 + ] + ], + [ + [ + 310, + 119 + ], + [ + 290, + 140 + ], + [ + 329, + 139 + ], + [ + 290, + 99 + ], + [ + 348, + 118 + ], + [ + 310, + 77 + ] + ] + ] + }, + { + "task_id": "0004869", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0010_cfg_0005_104122063678#200", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004869/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 415, + 243 + ], + [ + 425, + 229 + ], + [ + 424, + 268 + ], + [ + 408, + 220 + ], + [ + 414, + 281 + ], + [ + 399, + 234 + ] + ], + [ + [ + 415, + 218 + ], + [ + 396, + 210 + ], + [ + 431, + 230 + ], + [ + 401, + 207 + ], + [ + 449, + 238 + ], + [ + 419, + 214 + ] + ], + [ + [ + 409, + 258 + ], + [ + 436, + 258 + ], + [ + 414, + 257 + ], + [ + 404, + 259 + ], + [ + 388, + 257 + ], + [ + 376, + 259 + ] + ], + [ + [ + 419, + 260 + ], + [ + 433, + 245 + ], + [ + 402, + 264 + ], + [ + 438, + 255 + ], + [ + 387, + 279 + ], + [ + 424, + 272 + ] + ] + ] + }, + { + "task_id": "0004871", + "id": "RH20T_cfg5_task_0003_user_0014_scene_0010_cfg_0005_104122063678#375", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004871/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 391, + 154 + ], + [ + 409, + 143 + ], + [ + 409, + 163 + ], + [ + 372, + 145 + ], + [ + 392, + 173 + ], + [ + 356, + 157 + ] + ], + [ + [ + 385, + 176 + ], + [ + 395, + 160 + ], + [ + 382, + 194 + ], + [ + 387, + 159 + ], + [ + 372, + 208 + ], + [ + 377, + 173 + ] + ], + [ + [ + 391, + 174 + ], + [ + 386, + 178 + ], + [ + 375, + 156 + ], + [ + 408, + 192 + ], + [ + 380, + 153 + ], + [ + 411, + 187 + ] + ], + [ + [ + 363, + 174 + ], + [ + 386, + 171 + ], + [ + 365, + 155 + ], + [ + 362, + 191 + ], + [ + 340, + 158 + ], + [ + 339, + 194 + ] + ] + ] + }, + { + "task_id": "0004872", + "id": "RH20T_cfg3_task_0007_user_0016_scene_0007_cfg_0003_104122062823#163", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004872/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 103 + ], + [ + 327, + 125 + ], + [ + 321, + 83 + ], + [ + 354, + 121 + ], + [ + 333, + 61 + ], + [ + 364, + 100 + ] + ], + [ + [ + 334, + 92 + ], + [ + 339, + 101 + ], + [ + 307, + 78 + ], + [ + 360, + 105 + ], + [ + 304, + 70 + ], + [ + 355, + 96 + ] + ], + [ + [ + 318, + 105 + ], + [ + 342, + 108 + ], + [ + 317, + 134 + ], + [ + 319, + 76 + ], + [ + 291, + 132 + ], + [ + 294, + 71 + ] + ], + [ + [ + 335, + 97 + ], + [ + 337, + 72 + ], + [ + 365, + 102 + ], + [ + 307, + 92 + ], + [ + 362, + 125 + ], + [ + 307, + 114 + ] + ] + ] + }, + { + "task_id": "0004873", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0009_cfg_0007_104122061330#196", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004873/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 343, + 154 + ], + [ + 337, + 177 + ], + [ + 352, + 147 + ], + [ + 333, + 161 + ], + [ + 359, + 123 + ], + [ + 340, + 135 + ] + ], + [ + [ + 320, + 165 + ], + [ + 322, + 165 + ], + [ + 341, + 179 + ], + [ + 299, + 151 + ], + [ + 341, + 180 + ], + [ + 296, + 150 + ] + ], + [ + [ + 343, + 155 + ], + [ + 351, + 142 + ], + [ + 353, + 132 + ], + [ + 334, + 176 + ], + [ + 345, + 145 + ], + [ + 326, + 191 + ] + ], + [ + [ + 339, + 133 + ], + [ + 332, + 156 + ], + [ + 355, + 144 + ], + [ + 323, + 122 + ], + [ + 363, + 120 + ], + [ + 329, + 99 + ] + ] + ] + }, + { + "task_id": "0004874", + "id": "RH20T_cfg5_task_0044_user_0010_scene_0001_cfg_0005_036422060215#482", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up a bag of things'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004874/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 345, + 109 + ], + [ + 343, + 90 + ], + [ + 348, + 116 + ], + [ + 342, + 100 + ], + [ + 350, + 134 + ], + [ + 344, + 120 + ] + ], + [ + [ + 331, + 91 + ], + [ + 336, + 73 + ], + [ + 350, + 94 + ], + [ + 312, + 88 + ], + [ + 345, + 111 + ], + [ + 308, + 105 + ] + ], + [ + [ + 323, + 93 + ], + [ + 334, + 94 + ], + [ + 335, + 108 + ], + [ + 312, + 78 + ], + [ + 324, + 108 + ], + [ + 300, + 76 + ] + ], + [ + [ + 341, + 116 + ], + [ + 345, + 124 + ], + [ + 322, + 123 + ], + [ + 359, + 108 + ], + [ + 319, + 115 + ], + [ + 355, + 101 + ] + ] + ] + }, + { + "task_id": "0004879", + "id": "RH20T_cfg3_task_0006_user_0016_scene_0003_cfg_0003_104122063550#123", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004879/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 360, + 316 + ], + [ + 331, + 310 + ], + [ + 361, + 302 + ], + [ + 359, + 330 + ], + [ + 393, + 308 + ], + [ + 388, + 336 + ] + ], + [ + [ + 351, + 281 + ], + [ + 330, + 302 + ], + [ + 330, + 260 + ], + [ + 372, + 301 + ], + [ + 351, + 238 + ], + [ + 393, + 280 + ] + ], + [ + [ + 357, + 287 + ], + [ + 359, + 282 + ], + [ + 372, + 314 + ], + [ + 342, + 261 + ], + [ + 370, + 317 + ], + [ + 341, + 267 + ] + ], + [ + [ + 369, + 296 + ], + [ + 385, + 298 + ], + [ + 390, + 309 + ], + [ + 347, + 282 + ], + [ + 374, + 305 + ], + [ + 333, + 280 + ] + ] + ] + }, + { + "task_id": "0004880", + "id": "RH20T_cfg1_task_0030_user_0015_scene_0001_cfg_0001_039422060546#236", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour out the test tube'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004880/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 360, + 52 + ], + [ + 363, + 43 + ], + [ + 378, + 76 + ], + [ + 343, + 29 + ], + [ + 375, + 82 + ], + [ + 342, + 38 + ] + ], + [ + [ + 364, + 39 + ], + [ + 348, + 30 + ], + [ + 338, + 46 + ], + [ + 390, + 32 + ], + [ + 356, + 57 + ], + [ + 409, + 41 + ] + ], + [ + [ + 365, + 58 + ], + [ + 364, + 84 + ], + [ + 335, + 60 + ], + [ + 396, + 56 + ], + [ + 337, + 35 + ], + [ + 396, + 30 + ] + ], + [ + [ + 365, + 42 + ], + [ + 352, + 56 + ], + [ + 359, + 12 + ], + [ + 369, + 69 + ], + [ + 372, + 0 + ], + [ + 381, + 56 + ] + ] + ] + }, + { + "task_id": "0004885", + "id": "RH20T_cfg2_task_0051_user_0016_scene_0003_cfg_0002_104122063678#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the dish off the dish rack'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004885/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 392, + 103 + ], + [ + 371, + 100 + ], + [ + 384, + 88 + ], + [ + 400, + 116 + ], + [ + 408, + 91 + ], + [ + 422, + 120 + ] + ], + [ + [ + 420, + 110 + ], + [ + 421, + 91 + ], + [ + 413, + 126 + ], + [ + 427, + 96 + ], + [ + 412, + 143 + ], + [ + 426, + 113 + ] + ], + [ + [ + 436, + 117 + ], + [ + 453, + 136 + ], + [ + 450, + 113 + ], + [ + 420, + 122 + ], + [ + 434, + 96 + ], + [ + 404, + 104 + ] + ], + [ + [ + 422, + 124 + ], + [ + 445, + 119 + ], + [ + 416, + 131 + ], + [ + 427, + 117 + ], + [ + 394, + 135 + ], + [ + 407, + 121 + ] + ] + ] + }, + { + "task_id": "0004887", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_037522061512#206", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004887/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 254, + 106 + ], + [ + 265, + 120 + ], + [ + 234, + 119 + ], + [ + 273, + 93 + ], + [ + 223, + 105 + ], + [ + 261, + 79 + ] + ], + [ + [ + 245, + 106 + ], + [ + 263, + 95 + ], + [ + 252, + 120 + ], + [ + 238, + 91 + ], + [ + 234, + 132 + ], + [ + 219, + 104 + ] + ], + [ + [ + 263, + 93 + ], + [ + 276, + 84 + ], + [ + 280, + 96 + ], + [ + 246, + 91 + ], + [ + 265, + 105 + ], + [ + 231, + 100 + ] + ], + [ + [ + 261, + 106 + ], + [ + 267, + 84 + ], + [ + 261, + 108 + ], + [ + 261, + 103 + ], + [ + 255, + 129 + ], + [ + 254, + 126 + ] + ] + ] + }, + { + "task_id": "0004888", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_037522061512#260", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004888/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 247, + 78 + ], + [ + 226, + 85 + ], + [ + 235, + 71 + ], + [ + 259, + 85 + ], + [ + 254, + 65 + ], + [ + 279, + 78 + ] + ], + [ + [ + 254, + 88 + ], + [ + 233, + 91 + ], + [ + 245, + 71 + ], + [ + 263, + 104 + ], + [ + 267, + 68 + ], + [ + 284, + 102 + ] + ], + [ + [ + 245, + 81 + ], + [ + 263, + 76 + ], + [ + 250, + 65 + ], + [ + 241, + 96 + ], + [ + 231, + 69 + ], + [ + 224, + 101 + ] + ], + [ + [ + 261, + 80 + ], + [ + 269, + 64 + ], + [ + 255, + 91 + ], + [ + 267, + 68 + ], + [ + 247, + 107 + ], + [ + 258, + 85 + ] + ] + ] + }, + { + "task_id": "0004889", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_037522061512#288", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004889/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 258, + 79 + ], + [ + 267, + 65 + ], + [ + 253, + 92 + ], + [ + 263, + 64 + ], + [ + 243, + 107 + ], + [ + 253, + 78 + ] + ], + [ + [ + 255, + 70 + ], + [ + 256, + 54 + ], + [ + 240, + 57 + ], + [ + 270, + 84 + ], + [ + 240, + 72 + ], + [ + 269, + 98 + ] + ], + [ + [ + 246, + 78 + ], + [ + 267, + 87 + ], + [ + 253, + 60 + ], + [ + 240, + 97 + ], + [ + 232, + 50 + ], + [ + 218, + 87 + ] + ], + [ + [ + 243, + 64 + ], + [ + 243, + 52 + ], + [ + 228, + 47 + ], + [ + 258, + 81 + ], + [ + 228, + 60 + ], + [ + 258, + 94 + ] + ] + ] + }, + { + "task_id": "0004891", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_037522061512#340", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004891/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 267, + 85 + ], + [ + 276, + 66 + ], + [ + 266, + 92 + ], + [ + 269, + 77 + ], + [ + 258, + 111 + ], + [ + 261, + 98 + ] + ], + [ + [ + 256, + 95 + ], + [ + 241, + 77 + ], + [ + 255, + 104 + ], + [ + 257, + 86 + ], + [ + 269, + 122 + ], + [ + 270, + 102 + ] + ], + [ + [ + 247, + 104 + ], + [ + 236, + 101 + ], + [ + 246, + 83 + ], + [ + 248, + 126 + ], + [ + 256, + 87 + ], + [ + 259, + 128 + ] + ], + [ + [ + 282, + 81 + ], + [ + 261, + 77 + ], + [ + 286, + 60 + ], + [ + 279, + 102 + ], + [ + 307, + 64 + ], + [ + 300, + 106 + ] + ] + ] + }, + { + "task_id": "0004892", + "id": "RH20T_cfg7_task_0048_user_0020_scene_0004_cfg_0007_037522061512#417", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Rotate the steering wheel 90 degrees clockwise'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004892/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 258, + 98 + ], + [ + 272, + 88 + ], + [ + 244, + 99 + ], + [ + 273, + 97 + ], + [ + 229, + 110 + ], + [ + 259, + 109 + ] + ], + [ + [ + 252, + 90 + ], + [ + 271, + 103 + ], + [ + 245, + 103 + ], + [ + 260, + 77 + ], + [ + 225, + 91 + ], + [ + 239, + 63 + ] + ], + [ + [ + 282, + 48 + ], + [ + 276, + 62 + ], + [ + 282, + 30 + ], + [ + 282, + 67 + ], + [ + 289, + 14 + ], + [ + 289, + 51 + ] + ], + [ + [ + 259, + 70 + ], + [ + 269, + 54 + ], + [ + 254, + 80 + ], + [ + 264, + 58 + ], + [ + 243, + 96 + ], + [ + 253, + 76 + ] + ] + ] + }, + { + "task_id": "0004895", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0001_cfg_0003_104122063550#260", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004895/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 274, + 134 + ], + [ + 271, + 121 + ], + [ + 278, + 158 + ], + [ + 269, + 111 + ], + [ + 281, + 168 + ], + [ + 272, + 123 + ] + ], + [ + [ + 273, + 140 + ], + [ + 248, + 147 + ], + [ + 280, + 156 + ], + [ + 267, + 126 + ], + [ + 306, + 149 + ], + [ + 291, + 119 + ] + ], + [ + [ + 274, + 108 + ], + [ + 295, + 99 + ], + [ + 260, + 110 + ], + [ + 289, + 106 + ], + [ + 238, + 119 + ], + [ + 266, + 116 + ] + ], + [ + [ + 272, + 137 + ], + [ + 297, + 132 + ], + [ + 277, + 162 + ], + [ + 268, + 112 + ], + [ + 252, + 167 + ], + [ + 243, + 117 + ] + ] + ] + }, + { + "task_id": "0004896", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0007_cfg_0004_038522062288#142", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004896/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 207 + ], + [ + 262, + 238 + ], + [ + 295, + 203 + ], + [ + 240, + 212 + ], + [ + 298, + 173 + ], + [ + 244, + 184 + ] + ], + [ + [ + 311, + 201 + ], + [ + 310, + 232 + ], + [ + 289, + 202 + ], + [ + 332, + 200 + ], + [ + 290, + 169 + ], + [ + 333, + 170 + ] + ], + [ + [ + 324, + 222 + ], + [ + 321, + 252 + ], + [ + 296, + 211 + ], + [ + 351, + 233 + ], + [ + 300, + 183 + ], + [ + 353, + 205 + ] + ], + [ + [ + 296, + 229 + ], + [ + 293, + 212 + ], + [ + 264, + 235 + ], + [ + 327, + 223 + ], + [ + 268, + 250 + ], + [ + 328, + 239 + ] + ] + ] + }, + { + "task_id": "0004898", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0007_cfg_0004_038522062288#474", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004898/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 429, + 259 + ], + [ + 419, + 275 + ], + [ + 464, + 271 + ], + [ + 396, + 247 + ], + [ + 470, + 255 + ], + [ + 406, + 233 + ] + ], + [ + [ + 420, + 217 + ], + [ + 394, + 236 + ], + [ + 438, + 243 + ], + [ + 404, + 192 + ], + [ + 467, + 223 + ], + [ + 430, + 173 + ] + ], + [ + [ + 407, + 245 + ], + [ + 412, + 228 + ], + [ + 376, + 239 + ], + [ + 440, + 251 + ], + [ + 374, + 254 + ], + [ + 434, + 266 + ] + ], + [ + [ + 434, + 272 + ], + [ + 414, + 289 + ], + [ + 419, + 280 + ], + [ + 450, + 262 + ], + [ + 437, + 264 + ], + [ + 469, + 245 + ] + ] + ] + }, + { + "task_id": "0004899", + "id": "RH20T_cfg4_task_0037_user_0017_scene_0007_cfg_0004_038522062288#634", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004899/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 96 + ], + [ + 290, + 79 + ], + [ + 263, + 80 + ], + [ + 305, + 111 + ], + [ + 257, + 98 + ], + [ + 301, + 130 + ] + ], + [ + [ + 290, + 67 + ], + [ + 268, + 62 + ], + [ + 287, + 42 + ], + [ + 293, + 91 + ], + [ + 311, + 46 + ], + [ + 316, + 97 + ] + ], + [ + [ + 274, + 102 + ], + [ + 262, + 85 + ], + [ + 287, + 114 + ], + [ + 261, + 89 + ], + [ + 300, + 132 + ], + [ + 275, + 108 + ] + ], + [ + [ + 277, + 100 + ], + [ + 274, + 80 + ], + [ + 253, + 110 + ], + [ + 301, + 90 + ], + [ + 257, + 128 + ], + [ + 303, + 109 + ] + ] + ] + }, + { + "task_id": "0004901", + "id": "RH20T_cfg4_task_0108_user_0015_scene_0009_cfg_0004_104122062295#134", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Sharpen the pencil with a pencil sharpener'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004901/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 269, + 129 + ], + [ + 302, + 125 + ], + [ + 269, + 127 + ], + [ + 270, + 130 + ], + [ + 238, + 131 + ], + [ + 235, + 134 + ] + ], + [ + [ + 256, + 137 + ], + [ + 252, + 167 + ], + [ + 224, + 131 + ], + [ + 288, + 142 + ], + [ + 228, + 99 + ], + [ + 293, + 111 + ] + ], + [ + [ + 216, + 131 + ], + [ + 184, + 131 + ], + [ + 222, + 164 + ], + [ + 208, + 95 + ], + [ + 254, + 164 + ], + [ + 242, + 94 + ] + ], + [ + [ + 256, + 128 + ], + [ + 261, + 98 + ], + [ + 224, + 115 + ], + [ + 288, + 141 + ], + [ + 220, + 143 + ], + [ + 281, + 168 + ] + ] + ] + }, + { + "task_id": "0004903", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122061018#228", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004903/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 371, + 69 + ], + [ + 394, + 75 + ], + [ + 361, + 98 + ], + [ + 382, + 39 + ], + [ + 337, + 93 + ], + [ + 357, + 32 + ] + ], + [ + [ + 346, + 68 + ], + [ + 340, + 37 + ], + [ + 333, + 67 + ], + [ + 360, + 69 + ], + [ + 338, + 94 + ], + [ + 366, + 99 + ] + ], + [ + [ + 352, + 94 + ], + [ + 322, + 86 + ], + [ + 349, + 92 + ], + [ + 355, + 96 + ], + [ + 377, + 99 + ], + [ + 387, + 104 + ] + ], + [ + [ + 381, + 78 + ], + [ + 355, + 87 + ], + [ + 366, + 63 + ], + [ + 398, + 95 + ], + [ + 390, + 56 + ], + [ + 423, + 87 + ] + ] + ] + }, + { + "task_id": "0004905", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122061018#412", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004905/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 81 + ], + [ + 329, + 82 + ], + [ + 368, + 98 + ], + [ + 327, + 64 + ], + [ + 384, + 96 + ], + [ + 344, + 63 + ] + ], + [ + [ + 335, + 82 + ], + [ + 363, + 100 + ], + [ + 347, + 58 + ], + [ + 324, + 104 + ], + [ + 318, + 39 + ], + [ + 298, + 87 + ] + ], + [ + [ + 340, + 67 + ], + [ + 325, + 44 + ], + [ + 331, + 85 + ], + [ + 350, + 46 + ], + [ + 345, + 107 + ], + [ + 366, + 71 + ] + ], + [ + [ + 308, + 51 + ], + [ + 284, + 66 + ], + [ + 297, + 22 + ], + [ + 318, + 78 + ], + [ + 321, + 8 + ], + [ + 339, + 63 + ] + ] + ] + }, + { + "task_id": "0004906", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122061018#482", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004906/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 342, + 60 + ], + [ + 332, + 29 + ], + [ + 327, + 65 + ], + [ + 358, + 53 + ], + [ + 337, + 93 + ], + [ + 368, + 85 + ] + ], + [ + [ + 327, + 74 + ], + [ + 323, + 43 + ], + [ + 328, + 60 + ], + [ + 327, + 86 + ], + [ + 332, + 94 + ], + [ + 331, + 116 + ] + ], + [ + [ + 372, + 49 + ], + [ + 390, + 32 + ], + [ + 378, + 25 + ], + [ + 366, + 69 + ], + [ + 357, + 43 + ], + [ + 348, + 86 + ] + ], + [ + [ + 348, + 90 + ], + [ + 322, + 107 + ], + [ + 357, + 97 + ], + [ + 338, + 82 + ], + [ + 382, + 81 + ], + [ + 366, + 63 + ] + ] + ] + }, + { + "task_id": "0004907", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122061018#530", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004907/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 350, + 39 + ], + [ + 340, + 70 + ], + [ + 322, + 27 + ], + [ + 377, + 51 + ], + [ + 332, + 0 + ], + [ + 387, + 21 + ] + ], + [ + [ + 369, + 77 + ], + [ + 375, + 76 + ], + [ + 340, + 84 + ], + [ + 399, + 70 + ], + [ + 337, + 84 + ], + [ + 393, + 71 + ] + ], + [ + [ + 350, + 75 + ], + [ + 338, + 47 + ], + [ + 338, + 86 + ], + [ + 363, + 62 + ], + [ + 349, + 112 + ], + [ + 375, + 91 + ] + ], + [ + [ + 373, + 49 + ], + [ + 397, + 59 + ], + [ + 386, + 19 + ], + [ + 361, + 78 + ], + [ + 360, + 7 + ], + [ + 335, + 69 + ] + ] + ] + }, + { + "task_id": "0004910", + "id": "RH20T_cfg7_task_0042_user_0020_scene_0003_cfg_0007_104122063633#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004910/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 196, + 174 + ], + [ + 181, + 187 + ], + [ + 201, + 188 + ], + [ + 192, + 162 + ], + [ + 217, + 175 + ], + [ + 207, + 150 + ] + ], + [ + [ + 174, + 168 + ], + [ + 182, + 150 + ], + [ + 160, + 168 + ], + [ + 189, + 168 + ], + [ + 151, + 186 + ], + [ + 182, + 187 + ] + ], + [ + [ + 184, + 171 + ], + [ + 185, + 152 + ], + [ + 189, + 159 + ], + [ + 180, + 183 + ], + [ + 188, + 176 + ], + [ + 179, + 200 + ] + ], + [ + [ + 198, + 169 + ], + [ + 178, + 176 + ], + [ + 190, + 166 + ], + [ + 206, + 173 + ], + [ + 209, + 160 + ], + [ + 226, + 167 + ] + ] + ] + }, + { + "task_id": "0004913", + "id": "RH20T_cfg1_task_0046_user_0001_scene_0003_cfg_0001_750612070853#197", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004913/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 108, + 267 + ], + [ + 151, + 274 + ], + [ + 105, + 310 + ], + [ + 111, + 222 + ], + [ + 63, + 303 + ], + [ + 69, + 216 + ] + ], + [ + [ + 56, + 221 + ], + [ + 5, + 209 + ], + [ + 57, + 186 + ], + [ + 56, + 252 + ], + [ + 110, + 200 + ], + [ + 103, + 262 + ] + ], + [ + [ + 99, + 229 + ], + [ + 86, + 195 + ], + [ + 55, + 255 + ], + [ + 137, + 206 + ], + [ + 69, + 285 + ], + [ + 145, + 235 + ] + ], + [ + [ + 135, + 233 + ], + [ + 158, + 267 + ], + [ + 99, + 251 + ], + [ + 172, + 215 + ], + [ + 75, + 216 + ], + [ + 149, + 179 + ] + ] + ] + }, + { + "task_id": "0004914", + "id": "RH20T_cfg1_task_0046_user_0001_scene_0003_cfg_0001_750612070853#418", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004914/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 228, + 216 + ], + [ + 234, + 186 + ], + [ + 185, + 212 + ], + [ + 268, + 219 + ], + [ + 182, + 240 + ], + [ + 260, + 245 + ] + ], + [ + [ + 249, + 247 + ], + [ + 212, + 245 + ], + [ + 255, + 210 + ], + [ + 243, + 287 + ], + [ + 292, + 212 + ], + [ + 283, + 291 + ] + ], + [ + [ + 223, + 204 + ], + [ + 214, + 184 + ], + [ + 214, + 236 + ], + [ + 233, + 169 + ], + [ + 223, + 260 + ], + [ + 244, + 190 + ] + ], + [ + [ + 232, + 204 + ], + [ + 231, + 180 + ], + [ + 227, + 232 + ], + [ + 239, + 174 + ], + [ + 228, + 260 + ], + [ + 241, + 201 + ] + ] + ] + }, + { + "task_id": "0004915", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_750612070851#104", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004915/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 121, + 242 + ], + [ + 73, + 254 + ], + [ + 111, + 200 + ], + [ + 130, + 283 + ], + [ + 155, + 190 + ], + [ + 172, + 270 + ] + ], + [ + [ + 155, + 206 + ], + [ + 149, + 178 + ], + [ + 183, + 181 + ], + [ + 128, + 229 + ], + [ + 191, + 212 + ], + [ + 132, + 261 + ] + ], + [ + [ + 163, + 238 + ], + [ + 161, + 264 + ], + [ + 119, + 231 + ], + [ + 205, + 245 + ], + [ + 124, + 209 + ], + [ + 204, + 224 + ] + ], + [ + [ + 144, + 208 + ], + [ + 131, + 185 + ], + [ + 102, + 222 + ], + [ + 184, + 195 + ], + [ + 116, + 241 + ], + [ + 192, + 215 + ] + ] + ] + }, + { + "task_id": "0004916", + "id": "RH20T_cfg1_task_0003_user_0008_scene_0002_cfg_0001_750612070851#258", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004916/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 368, + 206 + ], + [ + 349, + 241 + ], + [ + 398, + 228 + ], + [ + 336, + 182 + ], + [ + 414, + 195 + ], + [ + 355, + 149 + ] + ], + [ + [ + 344, + 215 + ], + [ + 344, + 192 + ], + [ + 303, + 215 + ], + [ + 384, + 215 + ], + [ + 306, + 235 + ], + [ + 382, + 236 + ] + ], + [ + [ + 309, + 174 + ], + [ + 316, + 212 + ], + [ + 269, + 186 + ], + [ + 346, + 163 + ], + [ + 260, + 145 + ], + [ + 340, + 123 + ] + ], + [ + [ + 346, + 250 + ], + [ + 368, + 284 + ], + [ + 317, + 262 + ], + [ + 372, + 239 + ], + [ + 297, + 228 + ], + [ + 353, + 209 + ] + ] + ] + }, + { + "task_id": "0004919", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0005_cfg_0005_037522062165#709", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004919/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 223, + 149 + ], + [ + 216, + 173 + ], + [ + 247, + 155 + ], + [ + 197, + 143 + ], + [ + 254, + 131 + ], + [ + 204, + 118 + ] + ], + [ + [ + 220, + 142 + ], + [ + 241, + 140 + ], + [ + 207, + 144 + ], + [ + 234, + 139 + ], + [ + 185, + 147 + ], + [ + 210, + 141 + ] + ], + [ + [ + 198, + 135 + ], + [ + 224, + 138 + ], + [ + 203, + 110 + ], + [ + 194, + 160 + ], + [ + 176, + 106 + ], + [ + 167, + 157 + ] + ], + [ + [ + 224, + 157 + ], + [ + 223, + 140 + ], + [ + 217, + 138 + ], + [ + 232, + 177 + ], + [ + 218, + 153 + ], + [ + 233, + 191 + ] + ] + ] + }, + { + "task_id": "0004921", + "id": "RH20T_cfg2_task_0002_user_0013_scene_0006_cfg_0002_036422060215#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004921/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 271, + 167 + ], + [ + 263, + 148 + ], + [ + 252, + 175 + ], + [ + 290, + 159 + ], + [ + 260, + 194 + ], + [ + 298, + 178 + ] + ], + [ + [ + 300, + 169 + ], + [ + 319, + 179 + ], + [ + 290, + 185 + ], + [ + 311, + 152 + ], + [ + 272, + 174 + ], + [ + 292, + 141 + ] + ], + [ + [ + 296, + 157 + ], + [ + 301, + 141 + ], + [ + 313, + 153 + ], + [ + 278, + 161 + ], + [ + 308, + 168 + ], + [ + 274, + 177 + ] + ], + [ + [ + 286, + 165 + ], + [ + 294, + 148 + ], + [ + 296, + 177 + ], + [ + 277, + 154 + ], + [ + 287, + 193 + ], + [ + 269, + 171 + ] + ] + ] + }, + { + "task_id": "0004924", + "id": "RH20T_cfg5_task_0004_user_0014_scene_0007_cfg_0005_036422060215#238", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004924/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 272, + 128 + ], + [ + 276, + 113 + ], + [ + 287, + 121 + ], + [ + 257, + 135 + ], + [ + 284, + 135 + ], + [ + 254, + 149 + ] + ], + [ + [ + 273, + 124 + ], + [ + 253, + 122 + ], + [ + 270, + 144 + ], + [ + 277, + 104 + ], + [ + 291, + 147 + ], + [ + 297, + 106 + ] + ], + [ + [ + 267, + 142 + ], + [ + 283, + 132 + ], + [ + 254, + 127 + ], + [ + 279, + 157 + ], + [ + 238, + 138 + ], + [ + 263, + 167 + ] + ], + [ + [ + 285, + 134 + ], + [ + 299, + 129 + ], + [ + 277, + 115 + ], + [ + 294, + 152 + ], + [ + 264, + 120 + ], + [ + 281, + 156 + ] + ] + ] + }, + { + "task_id": "0004926", + "id": "RH20T_cfg1_task_0065_user_0014_scene_0004_cfg_0001_038522062547#101", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004926/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 453, + 261 + ], + [ + 425, + 267 + ], + [ + 454, + 255 + ], + [ + 452, + 268 + ], + [ + 481, + 249 + ], + [ + 483, + 262 + ] + ], + [ + [ + 451, + 259 + ], + [ + 447, + 232 + ], + [ + 424, + 262 + ], + [ + 479, + 255 + ], + [ + 428, + 289 + ], + [ + 483, + 281 + ] + ], + [ + [ + 436, + 241 + ], + [ + 442, + 232 + ], + [ + 441, + 268 + ], + [ + 431, + 215 + ], + [ + 436, + 274 + ], + [ + 426, + 223 + ] + ], + [ + [ + 448, + 271 + ], + [ + 423, + 280 + ], + [ + 461, + 273 + ], + [ + 436, + 269 + ], + [ + 486, + 263 + ], + [ + 458, + 260 + ] + ] + ] + }, + { + "task_id": "0004927", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104122061850#239", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004927/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 331, + 154 + ], + [ + 344, + 171 + ], + [ + 313, + 146 + ], + [ + 348, + 161 + ], + [ + 302, + 129 + ], + [ + 337, + 146 + ] + ], + [ + [ + 324, + 171 + ], + [ + 319, + 157 + ], + [ + 312, + 193 + ], + [ + 336, + 150 + ], + [ + 318, + 205 + ], + [ + 340, + 164 + ] + ], + [ + [ + 296, + 190 + ], + [ + 276, + 184 + ], + [ + 280, + 201 + ], + [ + 311, + 181 + ], + [ + 302, + 208 + ], + [ + 333, + 187 + ] + ], + [ + [ + 327, + 194 + ], + [ + 317, + 201 + ], + [ + 342, + 217 + ], + [ + 312, + 172 + ], + [ + 350, + 209 + ], + [ + 321, + 167 + ] + ] + ] + }, + { + "task_id": "0004929", + "id": "RH20T_cfg5_task_0037_user_0007_scene_0008_cfg_0005_104122061850#694", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004929/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 402, + 190 + ], + [ + 400, + 176 + ], + [ + 374, + 195 + ], + [ + 431, + 185 + ], + [ + 378, + 207 + ], + [ + 432, + 199 + ] + ], + [ + [ + 378, + 192 + ], + [ + 374, + 207 + ], + [ + 390, + 170 + ], + [ + 366, + 213 + ], + [ + 393, + 158 + ], + [ + 370, + 200 + ] + ], + [ + [ + 380, + 209 + ], + [ + 390, + 234 + ], + [ + 374, + 196 + ], + [ + 386, + 221 + ], + [ + 365, + 172 + ], + [ + 378, + 198 + ] + ], + [ + [ + 377, + 200 + ], + [ + 377, + 186 + ], + [ + 392, + 182 + ], + [ + 362, + 219 + ], + [ + 392, + 195 + ], + [ + 363, + 231 + ] + ] + ] + }, + { + "task_id": "0004931", + "id": "RH20T_cfg6_task_0082_user_0020_scene_0001_cfg_0006_104122061330#240", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004931/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 167 + ], + [ + 328, + 185 + ], + [ + 322, + 148 + ], + [ + 287, + 188 + ], + [ + 300, + 130 + ], + [ + 264, + 170 + ] + ], + [ + [ + 319, + 168 + ], + [ + 311, + 198 + ], + [ + 335, + 171 + ], + [ + 302, + 165 + ], + [ + 343, + 143 + ], + [ + 311, + 133 + ] + ], + [ + [ + 306, + 193 + ], + [ + 326, + 182 + ], + [ + 325, + 213 + ], + [ + 287, + 172 + ], + [ + 307, + 221 + ], + [ + 270, + 183 + ] + ], + [ + [ + 296, + 186 + ], + [ + 285, + 183 + ], + [ + 324, + 201 + ], + [ + 270, + 171 + ], + [ + 332, + 203 + ], + [ + 282, + 175 + ] + ] + ] + }, + { + "task_id": "0004933", + "id": "RH20T_cfg5_task_0039_user_0007_scene_0007_cfg_0005_036422060215#274", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pour the water from one cup into another empty cup'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004933/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 190, + 178 + ], + [ + 191, + 159 + ], + [ + 214, + 176 + ], + [ + 165, + 180 + ], + [ + 212, + 192 + ], + [ + 164, + 197 + ] + ], + [ + [ + 197, + 173 + ], + [ + 193, + 156 + ], + [ + 184, + 187 + ], + [ + 211, + 158 + ], + [ + 187, + 205 + ], + [ + 215, + 176 + ] + ], + [ + [ + 198, + 167 + ], + [ + 195, + 147 + ], + [ + 172, + 170 + ], + [ + 223, + 164 + ], + [ + 174, + 192 + ], + [ + 226, + 185 + ] + ], + [ + [ + 180, + 181 + ], + [ + 192, + 160 + ], + [ + 159, + 176 + ], + [ + 199, + 186 + ], + [ + 147, + 197 + ], + [ + 188, + 206 + ] + ] + ] + }, + { + "task_id": "0004934", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0002_cfg_0003_036422060909#137", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004934/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 403, + 93 + ], + [ + 384, + 104 + ], + [ + 402, + 109 + ], + [ + 404, + 75 + ], + [ + 420, + 98 + ], + [ + 423, + 64 + ] + ], + [ + [ + 409, + 61 + ], + [ + 420, + 73 + ], + [ + 387, + 63 + ], + [ + 430, + 58 + ], + [ + 377, + 51 + ], + [ + 419, + 47 + ] + ], + [ + [ + 409, + 92 + ], + [ + 414, + 74 + ], + [ + 427, + 83 + ], + [ + 390, + 102 + ], + [ + 422, + 100 + ], + [ + 386, + 119 + ] + ], + [ + [ + 402, + 62 + ], + [ + 389, + 71 + ], + [ + 387, + 78 + ], + [ + 418, + 46 + ], + [ + 399, + 70 + ], + [ + 430, + 39 + ] + ] + ] + }, + { + "task_id": "0004938", + "id": "RH20T_cfg1_task_0070_user_0015_scene_0002_cfg_0001_038522063145#300", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004938/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 422, + 129 + ], + [ + 429, + 111 + ], + [ + 441, + 139 + ], + [ + 404, + 119 + ], + [ + 434, + 156 + ], + [ + 398, + 135 + ] + ], + [ + [ + 391, + 142 + ], + [ + 374, + 151 + ], + [ + 379, + 133 + ], + [ + 404, + 151 + ], + [ + 396, + 125 + ], + [ + 422, + 142 + ] + ], + [ + [ + 418, + 147 + ], + [ + 427, + 165 + ], + [ + 412, + 144 + ], + [ + 423, + 149 + ], + [ + 403, + 125 + ], + [ + 414, + 132 + ] + ], + [ + [ + 418, + 116 + ], + [ + 427, + 102 + ], + [ + 402, + 123 + ], + [ + 433, + 108 + ], + [ + 394, + 136 + ], + [ + 424, + 121 + ] + ] + ] + }, + { + "task_id": "0004940", + "id": "RH20T_cfg1_task_0079_user_0015_scene_0003_cfg_0001_038522062547#199", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open the microwave door'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004940/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 323, + 0 + ], + [ + 331, + 18 + ], + [ + 313, + 0 + ], + [ + 332, + 25 + ], + [ + 304, + 0 + ], + [ + 324, + 6 + ] + ], + [ + [ + 327, + 2 + ], + [ + 321, + 3 + ], + [ + 343, + 0 + ], + [ + 313, + 33 + ], + [ + 347, + 0 + ], + [ + 319, + 30 + ] + ], + [ + [ + 353, + 22 + ], + [ + 359, + 0 + ], + [ + 358, + 47 + ], + [ + 347, + 0 + ], + [ + 352, + 68 + ], + [ + 342, + 20 + ] + ], + [ + [ + 361, + 0 + ], + [ + 356, + 1 + ], + [ + 393, + 0 + ], + [ + 328, + 0 + ], + [ + 401, + 0 + ], + [ + 332, + 0 + ] + ] + ] + }, + { + "task_id": "0004941", + "id": "RH20T_cfg7_task_0007_user_0014_scene_0005_cfg_0007_104122061330#171", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004941/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 332, + 261 + ], + [ + 329, + 263 + ], + [ + 321, + 235 + ], + [ + 343, + 288 + ], + [ + 324, + 235 + ], + [ + 345, + 285 + ] + ], + [ + [ + 319, + 221 + ], + [ + 315, + 209 + ], + [ + 306, + 245 + ], + [ + 331, + 198 + ], + [ + 311, + 255 + ], + [ + 334, + 210 + ] + ], + [ + [ + 308, + 202 + ], + [ + 331, + 216 + ], + [ + 304, + 206 + ], + [ + 311, + 199 + ], + [ + 280, + 191 + ], + [ + 289, + 185 + ] + ], + [ + [ + 313, + 225 + ], + [ + 326, + 232 + ], + [ + 294, + 241 + ], + [ + 332, + 208 + ], + [ + 278, + 234 + ], + [ + 318, + 200 + ] + ] + ] + }, + { + "task_id": "0004942", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0007_cfg_0006_104122060811#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004942/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 378, + 131 + ], + [ + 385, + 140 + ], + [ + 376, + 120 + ], + [ + 381, + 142 + ], + [ + 370, + 111 + ], + [ + 375, + 133 + ] + ], + [ + [ + 391, + 141 + ], + [ + 388, + 132 + ], + [ + 380, + 151 + ], + [ + 403, + 130 + ], + [ + 384, + 160 + ], + [ + 407, + 139 + ] + ], + [ + [ + 393, + 143 + ], + [ + 406, + 134 + ], + [ + 399, + 147 + ], + [ + 386, + 138 + ], + [ + 387, + 154 + ], + [ + 373, + 147 + ] + ], + [ + [ + 388, + 144 + ], + [ + 400, + 135 + ], + [ + 377, + 133 + ], + [ + 398, + 154 + ], + [ + 366, + 142 + ], + [ + 386, + 162 + ] + ] + ] + }, + { + "task_id": "0004943", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122061602#284", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004943/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 511, + 257 + ], + [ + 523, + 239 + ], + [ + 538, + 280 + ], + [ + 487, + 236 + ], + [ + 526, + 296 + ], + [ + 478, + 252 + ] + ], + [ + [ + 509, + 255 + ], + [ + 501, + 242 + ], + [ + 490, + 279 + ], + [ + 528, + 231 + ], + [ + 497, + 295 + ], + [ + 537, + 245 + ] + ], + [ + [ + 500, + 236 + ], + [ + 522, + 227 + ], + [ + 524, + 259 + ], + [ + 479, + 216 + ], + [ + 501, + 270 + ], + [ + 456, + 225 + ] + ], + [ + [ + 483, + 233 + ], + [ + 478, + 254 + ], + [ + 488, + 258 + ], + [ + 478, + 211 + ], + [ + 493, + 235 + ], + [ + 483, + 189 + ] + ] + ] + }, + { + "task_id": "0004944", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122061602#725", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004944/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 475, + 224 + ], + [ + 494, + 208 + ], + [ + 473, + 243 + ], + [ + 477, + 204 + ], + [ + 454, + 259 + ], + [ + 457, + 221 + ] + ], + [ + [ + 451, + 224 + ], + [ + 435, + 212 + ], + [ + 433, + 246 + ], + [ + 469, + 202 + ], + [ + 449, + 262 + ], + [ + 488, + 215 + ] + ], + [ + [ + 470, + 248 + ], + [ + 486, + 274 + ], + [ + 457, + 262 + ], + [ + 483, + 233 + ], + [ + 444, + 239 + ], + [ + 467, + 209 + ] + ], + [ + [ + 463, + 242 + ], + [ + 481, + 233 + ], + [ + 470, + 222 + ], + [ + 456, + 262 + ], + [ + 454, + 231 + ], + [ + 439, + 270 + ] + ] + ] + }, + { + "task_id": "0004946", + "id": "RH20T_cfg7_task_0046_user_0014_scene_0001_cfg_0007_104122064161#91", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the cup on the cup rack'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004946/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 354, + 115 + ], + [ + 366, + 129 + ], + [ + 352, + 99 + ], + [ + 357, + 131 + ], + [ + 339, + 84 + ], + [ + 344, + 116 + ] + ], + [ + [ + 371, + 134 + ], + [ + 376, + 113 + ], + [ + 372, + 148 + ], + [ + 370, + 120 + ], + [ + 368, + 168 + ], + [ + 366, + 139 + ] + ], + [ + [ + 370, + 145 + ], + [ + 377, + 168 + ], + [ + 376, + 148 + ], + [ + 364, + 142 + ], + [ + 368, + 122 + ], + [ + 358, + 120 + ] + ], + [ + [ + 362, + 150 + ], + [ + 349, + 168 + ], + [ + 351, + 133 + ], + [ + 373, + 166 + ], + [ + 364, + 116 + ], + [ + 385, + 149 + ] + ] + ] + }, + { + "task_id": "0004948", + "id": "RH20T_cfg1_task_0076_user_0010_scene_0001_cfg_0001_819112070044#172", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack blocks (small Lego) one on top of the other every time'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004948/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 402, + 162 + ], + [ + 407, + 144 + ], + [ + 397, + 161 + ], + [ + 408, + 162 + ], + [ + 393, + 178 + ], + [ + 403, + 180 + ] + ], + [ + [ + 397, + 169 + ], + [ + 401, + 156 + ], + [ + 412, + 163 + ], + [ + 382, + 175 + ], + [ + 408, + 176 + ], + [ + 379, + 188 + ] + ], + [ + [ + 397, + 169 + ], + [ + 398, + 184 + ], + [ + 400, + 157 + ], + [ + 393, + 180 + ], + [ + 398, + 143 + ], + [ + 392, + 167 + ] + ], + [ + [ + 377, + 163 + ], + [ + 364, + 153 + ], + [ + 372, + 154 + ], + [ + 381, + 171 + ], + [ + 386, + 164 + ], + [ + 394, + 181 + ] + ] + ] + }, + { + "task_id": "0004949", + "id": "RH20T_cfg1_task_0119_user_0014_scene_0004_cfg_0001_750612070853#96", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn off the water tap'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004949/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 251 + ], + [ + 287, + 208 + ], + [ + 253, + 238 + ], + [ + 351, + 266 + ], + [ + 266, + 274 + ], + [ + 359, + 304 + ] + ], + [ + [ + 286, + 194 + ], + [ + 308, + 192 + ], + [ + 330, + 164 + ], + [ + 236, + 227 + ], + [ + 309, + 169 + ], + [ + 223, + 226 + ] + ], + [ + [ + 320, + 261 + ], + [ + 277, + 238 + ], + [ + 290, + 312 + ], + [ + 348, + 213 + ], + [ + 337, + 343 + ], + [ + 397, + 236 + ] + ], + [ + [ + 249, + 242 + ], + [ + 196, + 216 + ], + [ + 238, + 224 + ], + [ + 261, + 265 + ], + [ + 281, + 245 + ], + [ + 314, + 289 + ] + ] + ] + }, + { + "task_id": "0004952", + "id": "RH20T_cfg5_task_0034_user_0007_scene_0005_cfg_0005_104122063678#796", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004952/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 303, + 293 + ], + [ + 310, + 267 + ], + [ + 273, + 290 + ], + [ + 332, + 296 + ], + [ + 266, + 316 + ], + [ + 324, + 320 + ] + ], + [ + [ + 300, + 289 + ], + [ + 317, + 316 + ], + [ + 302, + 293 + ], + [ + 297, + 284 + ], + [ + 287, + 269 + ], + [ + 280, + 257 + ] + ], + [ + [ + 289, + 286 + ], + [ + 266, + 306 + ], + [ + 277, + 274 + ], + [ + 303, + 298 + ], + [ + 298, + 256 + ], + [ + 326, + 278 + ] + ], + [ + [ + 289, + 288 + ], + [ + 293, + 272 + ], + [ + 274, + 311 + ], + [ + 302, + 266 + ], + [ + 271, + 325 + ], + [ + 298, + 281 + ] + ] + ] + }, + { + "task_id": "0004958", + "id": "RH20T_cfg1_task_0108_user_0014_scene_0004_cfg_0001_039422060546#103", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Sharpen the pencil with a pencil sharpener'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004958/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 290, + 97 + ], + [ + 260, + 88 + ], + [ + 299, + 91 + ], + [ + 279, + 103 + ], + [ + 328, + 100 + ], + [ + 312, + 113 + ] + ], + [ + [ + 326, + 71 + ], + [ + 310, + 42 + ], + [ + 340, + 62 + ], + [ + 314, + 78 + ], + [ + 357, + 92 + ], + [ + 329, + 105 + ] + ], + [ + [ + 306, + 101 + ], + [ + 304, + 77 + ], + [ + 323, + 120 + ], + [ + 291, + 82 + ], + [ + 324, + 141 + ], + [ + 293, + 104 + ] + ], + [ + [ + 310, + 90 + ], + [ + 339, + 82 + ], + [ + 302, + 60 + ], + [ + 317, + 119 + ], + [ + 272, + 67 + ], + [ + 288, + 127 + ] + ] + ] + }, + { + "task_id": "0004959", + "id": "RH20T_cfg1_task_0066_user_0015_scene_0006_cfg_0001_038522063145#149", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004959/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 366, + 72 + ], + [ + 358, + 89 + ], + [ + 380, + 78 + ], + [ + 353, + 66 + ], + [ + 389, + 60 + ], + [ + 361, + 49 + ] + ], + [ + [ + 366, + 67 + ], + [ + 370, + 51 + ], + [ + 381, + 78 + ], + [ + 352, + 58 + ], + [ + 376, + 93 + ], + [ + 348, + 73 + ] + ], + [ + [ + 343, + 76 + ], + [ + 341, + 57 + ], + [ + 335, + 71 + ], + [ + 351, + 81 + ], + [ + 336, + 88 + ], + [ + 352, + 99 + ] + ], + [ + [ + 360, + 53 + ], + [ + 362, + 36 + ], + [ + 352, + 64 + ], + [ + 368, + 41 + ], + [ + 351, + 80 + ], + [ + 366, + 57 + ] + ] + ] + }, + { + "task_id": "0004962", + "id": "RH20T_cfg1_task_0038_user_0015_scene_0010_cfg_0001_819112071037#159", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004962/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 448, + 85 + ], + [ + 449, + 8 + ], + [ + 386, + 93 + ], + [ + 503, + 78 + ], + [ + 388, + 167 + ], + [ + 500, + 145 + ] + ], + [ + [ + 410, + 164 + ], + [ + 377, + 190 + ], + [ + 475, + 208 + ], + [ + 351, + 123 + ], + [ + 495, + 182 + ], + [ + 382, + 108 + ] + ], + [ + [ + 491, + 96 + ], + [ + 436, + 118 + ], + [ + 446, + 55 + ], + [ + 544, + 143 + ], + [ + 495, + 39 + ], + [ + 594, + 119 + ] + ], + [ + [ + 413, + 73 + ], + [ + 490, + 83 + ], + [ + 419, + 84 + ], + [ + 408, + 65 + ], + [ + 330, + 72 + ], + [ + 343, + 56 + ] + ] + ] + }, + { + "task_id": "0004963", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0003_cfg_0005_104122063678#248", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004963/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 297, + 170 + ], + [ + 292, + 152 + ], + [ + 276, + 181 + ], + [ + 318, + 159 + ], + [ + 280, + 200 + ], + [ + 324, + 178 + ] + ], + [ + [ + 301, + 160 + ], + [ + 315, + 162 + ], + [ + 318, + 166 + ], + [ + 283, + 152 + ], + [ + 305, + 164 + ], + [ + 270, + 151 + ] + ], + [ + [ + 318, + 148 + ], + [ + 296, + 159 + ], + [ + 306, + 141 + ], + [ + 329, + 155 + ], + [ + 329, + 129 + ], + [ + 350, + 144 + ] + ], + [ + [ + 294, + 157 + ], + [ + 297, + 140 + ], + [ + 289, + 176 + ], + [ + 299, + 139 + ], + [ + 287, + 191 + ], + [ + 297, + 155 + ] + ] + ] + }, + { + "task_id": "0004964", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0003_cfg_0005_104122063678#414", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004964/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 173 + ], + [ + 300, + 157 + ], + [ + 285, + 188 + ], + [ + 311, + 159 + ], + [ + 285, + 204 + ], + [ + 310, + 175 + ] + ], + [ + [ + 316, + 163 + ], + [ + 332, + 146 + ], + [ + 302, + 147 + ], + [ + 331, + 179 + ], + [ + 287, + 164 + ], + [ + 315, + 196 + ] + ], + [ + [ + 315, + 184 + ], + [ + 297, + 173 + ], + [ + 315, + 172 + ], + [ + 316, + 195 + ], + [ + 335, + 184 + ], + [ + 334, + 206 + ] + ], + [ + [ + 311, + 163 + ], + [ + 296, + 159 + ], + [ + 314, + 139 + ], + [ + 308, + 187 + ], + [ + 327, + 143 + ], + [ + 322, + 190 + ] + ] + ] + }, + { + "task_id": "0004965", + "id": "RH20T_cfg6_task_0014_user_0014_scene_0008_cfg_0006_104122064161#57", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Remove the object from the scale'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004965/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 341, + 152 + ], + [ + 346, + 137 + ], + [ + 341, + 172 + ], + [ + 340, + 134 + ], + [ + 336, + 185 + ], + [ + 335, + 148 + ] + ], + [ + [ + 328, + 158 + ], + [ + 316, + 151 + ], + [ + 348, + 160 + ], + [ + 307, + 155 + ], + [ + 361, + 168 + ], + [ + 319, + 163 + ] + ], + [ + [ + 346, + 176 + ], + [ + 343, + 187 + ], + [ + 329, + 191 + ], + [ + 363, + 162 + ], + [ + 331, + 179 + ], + [ + 367, + 150 + ] + ], + [ + [ + 356, + 127 + ], + [ + 380, + 121 + ], + [ + 358, + 142 + ], + [ + 354, + 113 + ], + [ + 334, + 148 + ], + [ + 331, + 118 + ] + ] + ] + }, + { + "task_id": "0004966", + "id": "RH20T_cfg1_task_0032_user_0015_scene_0004_cfg_0001_750612070851#143", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004966/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 474, + 244 + ], + [ + 451, + 219 + ], + [ + 444, + 248 + ], + [ + 507, + 239 + ], + [ + 465, + 270 + ], + [ + 529, + 264 + ] + ], + [ + [ + 477, + 280 + ], + [ + 493, + 261 + ], + [ + 443, + 293 + ], + [ + 508, + 267 + ], + [ + 431, + 309 + ], + [ + 493, + 284 + ] + ], + [ + [ + 535, + 293 + ], + [ + 561, + 270 + ], + [ + 565, + 287 + ], + [ + 509, + 298 + ], + [ + 535, + 314 + ], + [ + 482, + 321 + ] + ], + [ + [ + 461, + 239 + ], + [ + 490, + 269 + ], + [ + 445, + 253 + ], + [ + 475, + 228 + ], + [ + 415, + 222 + ], + [ + 448, + 201 + ] + ] + ] + }, + { + "task_id": "0004967", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0006_cfg_0001_038522063145#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004967/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 445, + 116 + ], + [ + 435, + 128 + ], + [ + 424, + 117 + ], + [ + 467, + 115 + ], + [ + 434, + 107 + ], + [ + 476, + 104 + ] + ], + [ + [ + 471, + 123 + ], + [ + 452, + 109 + ], + [ + 469, + 129 + ], + [ + 473, + 118 + ], + [ + 489, + 144 + ], + [ + 491, + 132 + ] + ], + [ + [ + 454, + 117 + ], + [ + 443, + 98 + ], + [ + 471, + 112 + ], + [ + 435, + 122 + ], + [ + 482, + 131 + ], + [ + 446, + 142 + ] + ], + [ + [ + 458, + 121 + ], + [ + 458, + 99 + ], + [ + 479, + 127 + ], + [ + 439, + 117 + ], + [ + 480, + 148 + ], + [ + 440, + 137 + ] + ] + ] + }, + { + "task_id": "0004969", + "id": "RH20T_cfg4_task_0074_user_0015_scene_0006_cfg_0004_104422070011#95", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button horizontally'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004969/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 125, + 193 + ], + [ + 92, + 202 + ], + [ + 117, + 206 + ], + [ + 134, + 179 + ], + [ + 146, + 199 + ], + [ + 165, + 172 + ] + ], + [ + [ + 146, + 209 + ], + [ + 154, + 193 + ], + [ + 120, + 229 + ], + [ + 170, + 190 + ], + [ + 114, + 243 + ], + [ + 163, + 205 + ] + ], + [ + [ + 150, + 232 + ], + [ + 183, + 238 + ], + [ + 156, + 202 + ], + [ + 143, + 265 + ], + [ + 122, + 195 + ], + [ + 106, + 260 + ] + ], + [ + [ + 112, + 222 + ], + [ + 82, + 246 + ], + [ + 122, + 240 + ], + [ + 99, + 201 + ], + [ + 147, + 218 + ], + [ + 128, + 178 + ] + ] + ] + }, + { + "task_id": "0004971", + "id": "RH20T_cfg6_task_0091_user_0020_scene_0003_cfg_0006_037522061512#385", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004971/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 260, + 120 + ], + [ + 274, + 116 + ], + [ + 261, + 98 + ], + [ + 259, + 141 + ], + [ + 246, + 102 + ], + [ + 244, + 147 + ] + ], + [ + [ + 244, + 108 + ], + [ + 225, + 105 + ], + [ + 252, + 91 + ], + [ + 236, + 125 + ], + [ + 271, + 94 + ], + [ + 256, + 129 + ] + ], + [ + [ + 247, + 121 + ], + [ + 243, + 106 + ], + [ + 258, + 105 + ], + [ + 236, + 137 + ], + [ + 262, + 119 + ], + [ + 240, + 150 + ] + ], + [ + [ + 269, + 131 + ], + [ + 261, + 110 + ], + [ + 248, + 139 + ], + [ + 289, + 123 + ], + [ + 256, + 159 + ], + [ + 296, + 143 + ] + ] + ] + }, + { + "task_id": "0004974", + "id": "RH20T_cfg5_task_0101_user_0007_scene_0010_cfg_0005_037522062165#169", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the table with a cloth'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004974/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 195, + 222 + ], + [ + 193, + 206 + ], + [ + 180, + 205 + ], + [ + 211, + 241 + ], + [ + 182, + 221 + ], + [ + 212, + 256 + ] + ], + [ + [ + 187, + 206 + ], + [ + 197, + 200 + ], + [ + 160, + 199 + ], + [ + 212, + 213 + ], + [ + 151, + 205 + ], + [ + 202, + 218 + ] + ], + [ + [ + 191, + 219 + ], + [ + 166, + 211 + ], + [ + 198, + 223 + ], + [ + 183, + 215 + ], + [ + 222, + 230 + ], + [ + 210, + 223 + ] + ], + [ + [ + 200, + 224 + ], + [ + 176, + 239 + ], + [ + 194, + 222 + ], + [ + 206, + 228 + ], + [ + 216, + 208 + ], + [ + 230, + 212 + ] + ] + ] + }, + { + "task_id": "0004977", + "id": "RH20T_cfg5_task_0042_user_0007_scene_0003_cfg_0005_105422061350#304", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004977/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 183, + 91 + ], + [ + 184, + 73 + ], + [ + 198, + 75 + ], + [ + 167, + 109 + ], + [ + 196, + 92 + ], + [ + 167, + 125 + ] + ], + [ + [ + 200, + 90 + ], + [ + 226, + 79 + ], + [ + 195, + 74 + ], + [ + 204, + 105 + ], + [ + 166, + 86 + ], + [ + 179, + 116 + ] + ], + [ + [ + 203, + 106 + ], + [ + 203, + 108 + ], + [ + 226, + 90 + ], + [ + 181, + 121 + ], + [ + 224, + 89 + ], + [ + 182, + 118 + ] + ], + [ + [ + 149, + 78 + ], + [ + 119, + 72 + ], + [ + 147, + 55 + ], + [ + 151, + 99 + ], + [ + 178, + 61 + ], + [ + 179, + 104 + ] + ] + ] + }, + { + "task_id": "0004978", + "id": "RH20T_cfg5_task_0042_user_0007_scene_0003_cfg_0005_105422061350#754", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004978/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 262, + 62 + ], + [ + 255, + 66 + ], + [ + 261, + 32 + ], + [ + 263, + 91 + ], + [ + 269, + 25 + ], + [ + 271, + 88 + ] + ], + [ + [ + 272, + 106 + ], + [ + 252, + 87 + ], + [ + 277, + 92 + ], + [ + 267, + 120 + ], + [ + 295, + 110 + ], + [ + 288, + 139 + ] + ], + [ + [ + 260, + 95 + ], + [ + 267, + 76 + ], + [ + 275, + 80 + ], + [ + 244, + 112 + ], + [ + 268, + 97 + ], + [ + 239, + 129 + ] + ], + [ + [ + 251, + 84 + ], + [ + 237, + 68 + ], + [ + 273, + 82 + ], + [ + 227, + 87 + ], + [ + 288, + 98 + ], + [ + 242, + 105 + ] + ] + ] + }, + { + "task_id": "0004981", + "id": "RH20T_cfg3_task_0070_user_0010_scene_0004_cfg_0003_038522062288#164", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a sliding window'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004981/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 243, + 288 + ], + [ + 256, + 259 + ], + [ + 264, + 298 + ], + [ + 223, + 280 + ], + [ + 250, + 329 + ], + [ + 210, + 308 + ] + ], + [ + [ + 243, + 277 + ], + [ + 267, + 295 + ], + [ + 236, + 289 + ], + [ + 250, + 264 + ], + [ + 213, + 272 + ], + [ + 224, + 245 + ] + ], + [ + [ + 250, + 258 + ], + [ + 236, + 254 + ], + [ + 229, + 242 + ], + [ + 272, + 275 + ], + [ + 243, + 246 + ], + [ + 284, + 277 + ] + ], + [ + [ + 279, + 251 + ], + [ + 308, + 241 + ], + [ + 286, + 275 + ], + [ + 272, + 224 + ], + [ + 257, + 285 + ], + [ + 242, + 234 + ] + ] + ] + }, + { + "task_id": "0004982", + "id": "RH20T_cfg4_task_0012_user_0010_scene_0005_cfg_0004_104122063550#117", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push the soccer ball into the goal'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004982/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 331, + 67 + ], + [ + 321, + 43 + ], + [ + 318, + 77 + ], + [ + 344, + 57 + ], + [ + 328, + 101 + ], + [ + 353, + 80 + ] + ], + [ + [ + 372, + 68 + ], + [ + 361, + 52 + ], + [ + 388, + 49 + ], + [ + 356, + 86 + ], + [ + 400, + 65 + ], + [ + 366, + 102 + ] + ], + [ + [ + 363, + 54 + ], + [ + 348, + 44 + ], + [ + 369, + 30 + ], + [ + 358, + 77 + ], + [ + 386, + 41 + ], + [ + 375, + 88 + ] + ], + [ + [ + 343, + 74 + ], + [ + 345, + 66 + ], + [ + 360, + 91 + ], + [ + 327, + 56 + ], + [ + 358, + 98 + ], + [ + 326, + 65 + ] + ] + ] + }, + { + "task_id": "0004986", + "id": "RH20T_cfg3_task_0038_user_0010_scene_0005_cfg_0003_038522062288#201", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004986/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 292, + 282 + ], + [ + 310, + 251 + ], + [ + 295, + 288 + ], + [ + 290, + 275 + ], + [ + 278, + 318 + ], + [ + 269, + 309 + ] + ], + [ + [ + 281, + 240 + ], + [ + 273, + 252 + ], + [ + 300, + 269 + ], + [ + 263, + 211 + ], + [ + 311, + 257 + ], + [ + 272, + 196 + ] + ], + [ + [ + 274, + 209 + ], + [ + 277, + 193 + ], + [ + 285, + 181 + ], + [ + 261, + 239 + ], + [ + 282, + 195 + ], + [ + 259, + 251 + ] + ], + [ + [ + 272, + 244 + ], + [ + 268, + 229 + ], + [ + 240, + 237 + ], + [ + 305, + 251 + ], + [ + 246, + 251 + ], + [ + 307, + 265 + ] + ] + ] + }, + { + "task_id": "0004988", + "id": "RH20T_cfg1_task_0029_user_0001_scene_0004_cfg_0001_750612070853#191", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004988/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 373, + 311 + ], + [ + 364, + 244 + ], + [ + 310, + 319 + ], + [ + 442, + 302 + ], + [ + 315, + 360 + ], + [ + 456, + 360 + ] + ], + [ + [ + 308, + 297 + ], + [ + 321, + 238 + ], + [ + 288, + 280 + ], + [ + 334, + 319 + ], + [ + 277, + 333 + ], + [ + 319, + 360 + ] + ], + [ + [ + 343, + 356 + ], + [ + 328, + 329 + ], + [ + 293, + 326 + ], + [ + 401, + 360 + ], + [ + 309, + 350 + ], + [ + 408, + 360 + ] + ], + [ + [ + 384, + 357 + ], + [ + 371, + 358 + ], + [ + 348, + 360 + ], + [ + 419, + 309 + ], + [ + 362, + 360 + ], + [ + 425, + 313 + ] + ] + ] + }, + { + "task_id": "0004991", + "id": "RH20T_cfg3_task_0091_user_0010_scene_0010_cfg_0003_104122062823#190", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004991/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 140, + 52 + ], + [ + 134, + 26 + ], + [ + 165, + 59 + ], + [ + 114, + 45 + ], + [ + 170, + 83 + ], + [ + 120, + 69 + ] + ], + [ + [ + 150, + 29 + ], + [ + 178, + 37 + ], + [ + 153, + 7 + ], + [ + 147, + 51 + ], + [ + 124, + 0 + ], + [ + 116, + 42 + ] + ], + [ + [ + 120, + 66 + ], + [ + 94, + 81 + ], + [ + 102, + 40 + ], + [ + 138, + 92 + ], + [ + 128, + 25 + ], + [ + 163, + 77 + ] + ], + [ + [ + 167, + 53 + ], + [ + 150, + 38 + ], + [ + 184, + 59 + ], + [ + 149, + 47 + ], + [ + 202, + 74 + ], + [ + 168, + 65 + ] + ] + ] + }, + { + "task_id": "0004992", + "id": "RH20T_cfg3_task_0091_user_0010_scene_0010_cfg_0003_104122062823#244", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004992/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 177, + 70 + ], + [ + 184, + 89 + ], + [ + 196, + 52 + ], + [ + 159, + 86 + ], + [ + 190, + 33 + ], + [ + 154, + 67 + ] + ], + [ + [ + 171, + 42 + ], + [ + 146, + 43 + ], + [ + 158, + 31 + ], + [ + 185, + 53 + ], + [ + 182, + 31 + ], + [ + 209, + 52 + ] + ], + [ + [ + 165, + 49 + ], + [ + 155, + 25 + ], + [ + 191, + 51 + ], + [ + 138, + 46 + ], + [ + 198, + 72 + ], + [ + 147, + 67 + ] + ], + [ + [ + 161, + 62 + ], + [ + 154, + 44 + ], + [ + 133, + 58 + ], + [ + 187, + 65 + ], + [ + 139, + 78 + ], + [ + 194, + 84 + ] + ] + ] + }, + { + "task_id": "0004993", + "id": "RH20T_cfg1_task_0001_user_0009_scene_0003_cfg_0001_038522062547#89", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004993/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 378, + 152 + ], + [ + 405, + 161 + ], + [ + 378, + 178 + ], + [ + 379, + 129 + ], + [ + 349, + 170 + ], + [ + 352, + 120 + ] + ], + [ + [ + 410, + 177 + ], + [ + 440, + 175 + ], + [ + 408, + 155 + ], + [ + 413, + 202 + ], + [ + 380, + 157 + ], + [ + 383, + 204 + ] + ], + [ + [ + 380, + 170 + ], + [ + 383, + 191 + ], + [ + 351, + 173 + ], + [ + 410, + 167 + ], + [ + 347, + 150 + ], + [ + 408, + 143 + ] + ], + [ + [ + 395, + 156 + ], + [ + 400, + 142 + ], + [ + 396, + 182 + ], + [ + 393, + 131 + ], + [ + 391, + 192 + ], + [ + 389, + 144 + ] + ] + ] + }, + { + "task_id": "0004994", + "id": "RH20T_cfg1_task_0001_user_0014_scene_0006_cfg_0001_750612070853#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004994/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 545, + 155 + ], + [ + 601, + 111 + ], + [ + 528, + 109 + ], + [ + 566, + 211 + ], + [ + 487, + 146 + ], + [ + 514, + 244 + ] + ], + [ + [ + 520, + 219 + ], + [ + 499, + 258 + ], + [ + 463, + 229 + ], + [ + 591, + 205 + ], + [ + 483, + 198 + ], + [ + 604, + 170 + ] + ], + [ + [ + 512, + 206 + ], + [ + 521, + 227 + ], + [ + 453, + 228 + ], + [ + 576, + 183 + ], + [ + 436, + 207 + ], + [ + 571, + 153 + ] + ], + [ + [ + 471, + 217 + ], + [ + 484, + 166 + ], + [ + 418, + 190 + ], + [ + 532, + 249 + ], + [ + 411, + 232 + ], + [ + 515, + 293 + ] + ] + ] + }, + { + "task_id": "0004997", + "id": "RH20T_cfg4_task_0058_user_0014_scene_0004_cfg_0004_104122062295#114", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004997/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 236, + 229 + ], + [ + 258, + 214 + ], + [ + 242, + 185 + ], + [ + 230, + 271 + ], + [ + 215, + 199 + ], + [ + 204, + 292 + ] + ], + [ + [ + 189, + 193 + ], + [ + 189, + 180 + ], + [ + 144, + 166 + ], + [ + 230, + 218 + ], + [ + 138, + 179 + ], + [ + 234, + 235 + ] + ], + [ + [ + 217, + 205 + ], + [ + 253, + 189 + ], + [ + 248, + 235 + ], + [ + 189, + 179 + ], + [ + 207, + 258 + ], + [ + 147, + 195 + ] + ], + [ + [ + 232, + 211 + ], + [ + 245, + 173 + ], + [ + 193, + 191 + ], + [ + 271, + 232 + ], + [ + 183, + 226 + ], + [ + 258, + 267 + ] + ] + ] + }, + { + "task_id": "0004998", + "id": "RH20T_cfg4_task_0058_user_0014_scene_0004_cfg_0004_104122062295#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004998/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 257, + 215 + ], + [ + 221, + 220 + ], + [ + 232, + 237 + ], + [ + 285, + 191 + ], + [ + 263, + 231 + ], + [ + 314, + 188 + ] + ], + [ + [ + 228, + 200 + ], + [ + 240, + 162 + ], + [ + 190, + 181 + ], + [ + 267, + 219 + ], + [ + 180, + 217 + ], + [ + 254, + 254 + ] + ], + [ + [ + 216, + 237 + ], + [ + 197, + 256 + ], + [ + 218, + 277 + ], + [ + 214, + 195 + ], + [ + 234, + 258 + ], + [ + 231, + 182 + ] + ], + [ + [ + 247, + 196 + ], + [ + 283, + 195 + ], + [ + 277, + 183 + ], + [ + 221, + 207 + ], + [ + 234, + 184 + ], + [ + 180, + 209 + ] + ] + ] + }, + { + "task_id": "0004999", + "id": "RH20T_cfg4_task_0058_user_0014_scene_0004_cfg_0004_104122062295#171", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0004999/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 239, + 235 + ], + [ + 228, + 222 + ], + [ + 284, + 230 + ], + [ + 193, + 239 + ], + [ + 289, + 241 + ], + [ + 205, + 250 + ] + ], + [ + [ + 235, + 235 + ], + [ + 264, + 239 + ], + [ + 252, + 274 + ], + [ + 220, + 200 + ], + [ + 217, + 272 + ], + [ + 186, + 192 + ] + ], + [ + [ + 234, + 189 + ], + [ + 190, + 202 + ], + [ + 218, + 188 + ], + [ + 247, + 189 + ], + [ + 267, + 173 + ], + [ + 288, + 177 + ] + ], + [ + [ + 242, + 206 + ], + [ + 256, + 168 + ], + [ + 205, + 186 + ], + [ + 280, + 226 + ], + [ + 193, + 221 + ], + [ + 265, + 261 + ] + ] + ] + }, + { + "task_id": "0005000", + "id": "RH20T_cfg4_task_0058_user_0014_scene_0004_cfg_0004_104122062295#198", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005000/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 231, + 198 + ], + [ + 216, + 165 + ], + [ + 263, + 217 + ], + [ + 203, + 181 + ], + [ + 274, + 245 + ], + [ + 217, + 209 + ] + ], + [ + [ + 298, + 248 + ], + [ + 322, + 283 + ], + [ + 271, + 266 + ], + [ + 327, + 229 + ], + [ + 248, + 234 + ], + [ + 302, + 194 + ] + ], + [ + [ + 267, + 205 + ], + [ + 281, + 167 + ], + [ + 229, + 186 + ], + [ + 305, + 225 + ], + [ + 217, + 221 + ], + [ + 290, + 261 + ] + ], + [ + [ + 278, + 187 + ], + [ + 247, + 200 + ], + [ + 305, + 217 + ], + [ + 252, + 158 + ], + [ + 330, + 203 + ], + [ + 279, + 149 + ] + ] + ] + }, + { + "task_id": "0005002", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0001_cfg_0005_037522062165#432", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005002/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 234, + 220 + ], + [ + 235, + 202 + ], + [ + 227, + 199 + ], + [ + 242, + 242 + ], + [ + 227, + 216 + ], + [ + 241, + 258 + ] + ], + [ + [ + 253, + 214 + ], + [ + 241, + 189 + ], + [ + 236, + 227 + ], + [ + 269, + 203 + ], + [ + 248, + 253 + ], + [ + 280, + 226 + ] + ], + [ + [ + 234, + 214 + ], + [ + 239, + 235 + ], + [ + 226, + 195 + ], + [ + 242, + 231 + ], + [ + 222, + 176 + ], + [ + 238, + 212 + ] + ], + [ + [ + 257, + 197 + ], + [ + 269, + 184 + ], + [ + 250, + 173 + ], + [ + 264, + 220 + ], + [ + 236, + 187 + ], + [ + 252, + 236 + ] + ] + ] + }, + { + "task_id": "0005005", + "id": "RH20T_cfg1_task_0018_user_0008_scene_0007_cfg_0001_039422060546#139", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005005/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 349, + 137 + ], + [ + 359, + 120 + ], + [ + 346, + 115 + ], + [ + 351, + 157 + ], + [ + 334, + 133 + ], + [ + 340, + 175 + ] + ], + [ + [ + 327, + 119 + ], + [ + 337, + 109 + ], + [ + 305, + 108 + ], + [ + 350, + 130 + ], + [ + 293, + 119 + ], + [ + 341, + 141 + ] + ], + [ + [ + 329, + 123 + ], + [ + 329, + 102 + ], + [ + 356, + 129 + ], + [ + 302, + 117 + ], + [ + 354, + 147 + ], + [ + 303, + 135 + ] + ], + [ + [ + 360, + 101 + ], + [ + 341, + 118 + ], + [ + 338, + 92 + ], + [ + 383, + 110 + ], + [ + 356, + 77 + ], + [ + 401, + 93 + ] + ] + ] + }, + { + "task_id": "0005007", + "id": "RH20T_cfg5_task_0059_user_0007_scene_0003_cfg_0005_104122061850#536", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005007/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 296, + 112 + ], + [ + 291, + 96 + ], + [ + 281, + 130 + ], + [ + 311, + 95 + ], + [ + 286, + 143 + ], + [ + 315, + 110 + ] + ], + [ + [ + 287, + 124 + ], + [ + 301, + 106 + ], + [ + 306, + 140 + ], + [ + 268, + 108 + ], + [ + 293, + 159 + ], + [ + 254, + 126 + ] + ], + [ + [ + 296, + 125 + ], + [ + 282, + 124 + ], + [ + 306, + 148 + ], + [ + 286, + 101 + ], + [ + 322, + 150 + ], + [ + 302, + 100 + ] + ], + [ + [ + 284, + 136 + ], + [ + 262, + 142 + ], + [ + 284, + 115 + ], + [ + 284, + 155 + ], + [ + 306, + 110 + ], + [ + 305, + 149 + ] + ] + ] + }, + { + "task_id": "0005015", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122064161#242", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005015/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 336, + 254 + ], + [ + 337, + 240 + ], + [ + 338, + 280 + ], + [ + 334, + 230 + ], + [ + 338, + 293 + ], + [ + 333, + 244 + ] + ], + [ + [ + 358, + 264 + ], + [ + 379, + 243 + ], + [ + 336, + 243 + ], + [ + 379, + 286 + ], + [ + 315, + 264 + ], + [ + 358, + 308 + ] + ], + [ + [ + 350, + 257 + ], + [ + 363, + 268 + ], + [ + 358, + 230 + ], + [ + 342, + 283 + ], + [ + 346, + 221 + ], + [ + 331, + 272 + ] + ], + [ + [ + 343, + 250 + ], + [ + 345, + 221 + ], + [ + 319, + 252 + ], + [ + 369, + 248 + ], + [ + 317, + 281 + ], + [ + 367, + 280 + ] + ] + ] + }, + { + "task_id": "0005016", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122064161#497", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005016/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 225 + ], + [ + 384, + 231 + ], + [ + 380, + 216 + ], + [ + 344, + 234 + ], + [ + 361, + 210 + ], + [ + 325, + 228 + ] + ], + [ + [ + 323, + 267 + ], + [ + 315, + 249 + ], + [ + 314, + 288 + ], + [ + 333, + 245 + ], + [ + 322, + 309 + ], + [ + 344, + 264 + ] + ], + [ + [ + 343, + 218 + ], + [ + 329, + 198 + ], + [ + 329, + 216 + ], + [ + 356, + 221 + ], + [ + 345, + 239 + ], + [ + 370, + 241 + ] + ], + [ + [ + 338, + 239 + ], + [ + 339, + 225 + ], + [ + 337, + 265 + ], + [ + 338, + 215 + ], + [ + 336, + 277 + ], + [ + 337, + 228 + ] + ] + ] + }, + { + "task_id": "0005017", + "id": "RH20T_cfg6_task_0096_user_0020_scene_0003_cfg_0006_104122064161#621", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005017/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 317, + 236 + ], + [ + 317, + 210 + ], + [ + 338, + 241 + ], + [ + 297, + 232 + ], + [ + 338, + 268 + ], + [ + 297, + 257 + ] + ], + [ + [ + 334, + 263 + ], + [ + 330, + 285 + ], + [ + 328, + 275 + ], + [ + 339, + 250 + ], + [ + 332, + 254 + ], + [ + 343, + 229 + ] + ], + [ + [ + 330, + 242 + ], + [ + 327, + 266 + ], + [ + 327, + 251 + ], + [ + 333, + 232 + ], + [ + 330, + 229 + ], + [ + 336, + 208 + ] + ], + [ + [ + 356, + 223 + ], + [ + 356, + 232 + ], + [ + 338, + 207 + ], + [ + 376, + 239 + ], + [ + 338, + 197 + ], + [ + 378, + 230 + ] + ] + ] + }, + { + "task_id": "0005018", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_036422060909#94", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005018/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 316, + 88 + ], + [ + 331, + 96 + ], + [ + 299, + 87 + ], + [ + 332, + 89 + ], + [ + 286, + 78 + ], + [ + 319, + 81 + ] + ], + [ + [ + 328, + 54 + ], + [ + 333, + 50 + ], + [ + 350, + 61 + ], + [ + 306, + 47 + ], + [ + 345, + 65 + ], + [ + 300, + 50 + ] + ], + [ + [ + 323, + 77 + ], + [ + 325, + 60 + ], + [ + 320, + 61 + ], + [ + 325, + 95 + ], + [ + 318, + 77 + ], + [ + 323, + 110 + ] + ], + [ + [ + 334, + 61 + ], + [ + 357, + 60 + ], + [ + 331, + 40 + ], + [ + 336, + 84 + ], + [ + 308, + 41 + ], + [ + 312, + 85 + ] + ] + ] + }, + { + "task_id": "0005019", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_036422060909#171", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005019/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 293, + 58 + ], + [ + 306, + 48 + ], + [ + 303, + 48 + ], + [ + 283, + 68 + ], + [ + 289, + 59 + ], + [ + 269, + 79 + ] + ], + [ + [ + 292, + 64 + ], + [ + 299, + 78 + ], + [ + 297, + 79 + ], + [ + 287, + 47 + ], + [ + 290, + 66 + ], + [ + 280, + 34 + ] + ], + [ + [ + 314, + 67 + ], + [ + 319, + 49 + ], + [ + 318, + 52 + ], + [ + 309, + 82 + ], + [ + 313, + 68 + ], + [ + 304, + 98 + ] + ], + [ + [ + 300, + 65 + ], + [ + 276, + 68 + ], + [ + 302, + 63 + ], + [ + 298, + 66 + ], + [ + 326, + 59 + ], + [ + 320, + 63 + ] + ] + ] + }, + { + "task_id": "0005020", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0003_cfg_0004_036422060909#302", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005020/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 287, + 80 + ], + [ + 284, + 56 + ], + [ + 276, + 71 + ], + [ + 298, + 87 + ], + [ + 279, + 96 + ], + [ + 301, + 110 + ] + ], + [ + [ + 304, + 94 + ], + [ + 311, + 79 + ], + [ + 323, + 86 + ], + [ + 284, + 103 + ], + [ + 316, + 100 + ], + [ + 279, + 117 + ] + ], + [ + [ + 303, + 116 + ], + [ + 289, + 125 + ], + [ + 308, + 93 + ], + [ + 298, + 136 + ], + [ + 321, + 85 + ], + [ + 311, + 127 + ] + ], + [ + [ + 329, + 92 + ], + [ + 342, + 107 + ], + [ + 314, + 83 + ], + [ + 341, + 99 + ], + [ + 302, + 68 + ], + [ + 329, + 85 + ] + ] + ] + }, + { + "task_id": "0005024", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104422070011#254", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005024/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 279, + 158 + ], + [ + 257, + 162 + ], + [ + 289, + 135 + ], + [ + 270, + 179 + ], + [ + 309, + 132 + ], + [ + 290, + 174 + ] + ], + [ + [ + 274, + 189 + ], + [ + 279, + 175 + ], + [ + 259, + 209 + ], + [ + 288, + 170 + ], + [ + 255, + 222 + ], + [ + 284, + 184 + ] + ], + [ + [ + 286, + 181 + ], + [ + 310, + 184 + ], + [ + 293, + 193 + ], + [ + 278, + 167 + ], + [ + 270, + 190 + ], + [ + 253, + 163 + ] + ], + [ + [ + 278, + 193 + ], + [ + 289, + 218 + ], + [ + 296, + 179 + ], + [ + 258, + 209 + ], + [ + 285, + 154 + ], + [ + 245, + 182 + ] + ] + ] + }, + { + "task_id": "0005026", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104422070011#294", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005026/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 268, + 199 + ], + [ + 277, + 216 + ], + [ + 243, + 209 + ], + [ + 293, + 188 + ], + [ + 233, + 190 + ], + [ + 285, + 169 + ] + ], + [ + [ + 286, + 176 + ], + [ + 290, + 161 + ], + [ + 273, + 197 + ], + [ + 298, + 157 + ], + [ + 269, + 209 + ], + [ + 293, + 171 + ] + ], + [ + [ + 303, + 178 + ], + [ + 315, + 200 + ], + [ + 285, + 187 + ], + [ + 319, + 170 + ], + [ + 272, + 164 + ], + [ + 307, + 149 + ] + ], + [ + [ + 265, + 153 + ], + [ + 238, + 142 + ], + [ + 262, + 161 + ], + [ + 268, + 144 + ], + [ + 287, + 171 + ], + [ + 297, + 155 + ] + ] + ] + }, + { + "task_id": "0005027", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104422070011#356", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005027/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 275, + 188 + ], + [ + 274, + 183 + ], + [ + 244, + 183 + ], + [ + 306, + 194 + ], + [ + 246, + 188 + ], + [ + 304, + 198 + ] + ], + [ + [ + 286, + 189 + ], + [ + 297, + 171 + ], + [ + 285, + 213 + ], + [ + 286, + 167 + ], + [ + 274, + 229 + ], + [ + 276, + 184 + ] + ], + [ + [ + 305, + 166 + ], + [ + 332, + 174 + ], + [ + 312, + 145 + ], + [ + 297, + 189 + ], + [ + 285, + 137 + ], + [ + 269, + 181 + ] + ], + [ + [ + 282, + 170 + ], + [ + 309, + 177 + ], + [ + 290, + 177 + ], + [ + 275, + 165 + ], + [ + 259, + 170 + ], + [ + 248, + 159 + ] + ] + ] + }, + { + "task_id": "0005030", + "id": "RH20T_cfg4_task_0081_user_0015_scene_0002_cfg_0004_104422070011#489", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005030/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 235, + 183 + ], + [ + 234, + 213 + ], + [ + 215, + 186 + ], + [ + 253, + 180 + ], + [ + 216, + 154 + ], + [ + 254, + 151 + ] + ], + [ + [ + 259, + 191 + ], + [ + 266, + 175 + ], + [ + 247, + 213 + ], + [ + 270, + 171 + ], + [ + 241, + 227 + ], + [ + 264, + 186 + ] + ], + [ + [ + 231, + 184 + ], + [ + 241, + 174 + ], + [ + 238, + 158 + ], + [ + 225, + 208 + ], + [ + 228, + 168 + ], + [ + 214, + 221 + ] + ], + [ + [ + 247, + 201 + ], + [ + 262, + 175 + ], + [ + 238, + 199 + ], + [ + 255, + 202 + ], + [ + 222, + 227 + ], + [ + 241, + 226 + ] + ] + ] + }, + { + "task_id": "0005031", + "id": "RH20T_cfg4_task_0016_user_0016_scene_0006_cfg_0004_104122062823#128", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005031/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 188, + 37 + ], + [ + 208, + 16 + ], + [ + 187, + 46 + ], + [ + 189, + 26 + ], + [ + 167, + 66 + ], + [ + 167, + 49 + ] + ], + [ + [ + 241, + 32 + ], + [ + 248, + 43 + ], + [ + 248, + 62 + ], + [ + 232, + 0 + ], + [ + 242, + 51 + ], + [ + 226, + 0 + ] + ], + [ + [ + 234, + 0 + ], + [ + 222, + 0 + ], + [ + 258, + 8 + ], + [ + 210, + 0 + ], + [ + 267, + 10 + ], + [ + 222, + 0 + ] + ], + [ + [ + 207, + 25 + ], + [ + 203, + 0 + ], + [ + 239, + 23 + ], + [ + 174, + 27 + ], + [ + 241, + 50 + ], + [ + 180, + 55 + ] + ] + ] + }, + { + "task_id": "0005032", + "id": "RH20T_cfg4_task_0016_user_0016_scene_0006_cfg_0004_104122062823#223", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005032/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 149, + 0 + ], + [ + 149, + 0 + ], + [ + 133, + 0 + ], + [ + 165, + 0 + ], + [ + 134, + 0 + ], + [ + 164, + 0 + ] + ], + [ + [ + 88, + 0 + ], + [ + 83, + 0 + ], + [ + 56, + 0 + ], + [ + 120, + 0 + ], + [ + 60, + 0 + ], + [ + 125, + 0 + ] + ], + [ + [ + 121, + 0 + ], + [ + 139, + 0 + ], + [ + 133, + 0 + ], + [ + 107, + 0 + ], + [ + 116, + 0 + ], + [ + 89, + 0 + ] + ], + [ + [ + 130, + 0 + ], + [ + 120, + 0 + ], + [ + 153, + 0 + ], + [ + 107, + 0 + ], + [ + 161, + 0 + ], + [ + 118, + 3 + ] + ] + ] + }, + { + "task_id": "0005038", + "id": "RH20T_cfg5_task_0083_user_0007_scene_0010_cfg_0005_104122061850#293", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Swat with a flyswatter'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005038/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 256, + 131 + ], + [ + 266, + 112 + ], + [ + 272, + 145 + ], + [ + 240, + 116 + ], + [ + 262, + 163 + ], + [ + 231, + 134 + ] + ], + [ + [ + 256, + 138 + ], + [ + 262, + 152 + ], + [ + 279, + 138 + ], + [ + 234, + 138 + ], + [ + 274, + 122 + ], + [ + 226, + 123 + ] + ], + [ + [ + 259, + 150 + ], + [ + 253, + 135 + ], + [ + 245, + 168 + ], + [ + 273, + 132 + ], + [ + 252, + 180 + ], + [ + 278, + 146 + ] + ], + [ + [ + 245, + 132 + ], + [ + 262, + 140 + ], + [ + 237, + 154 + ], + [ + 253, + 109 + ], + [ + 220, + 146 + ], + [ + 235, + 102 + ] + ] + ] + }, + { + "task_id": "0005044", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0007_cfg_0004_104122062823#220", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005044/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 470, + 80 + ], + [ + 471, + 50 + ], + [ + 503, + 77 + ], + [ + 439, + 83 + ], + [ + 502, + 108 + ], + [ + 438, + 113 + ] + ], + [ + [ + 496, + 62 + ], + [ + 487, + 66 + ], + [ + 488, + 92 + ], + [ + 504, + 29 + ], + [ + 496, + 87 + ], + [ + 512, + 27 + ] + ], + [ + [ + 494, + 81 + ], + [ + 476, + 100 + ], + [ + 496, + 58 + ], + [ + 492, + 102 + ], + [ + 514, + 39 + ], + [ + 508, + 84 + ] + ], + [ + [ + 467, + 67 + ], + [ + 473, + 44 + ], + [ + 500, + 71 + ], + [ + 436, + 64 + ], + [ + 494, + 92 + ], + [ + 433, + 85 + ] + ] + ] + }, + { + "task_id": "0005047", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0007_cfg_0004_104122062823#493", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005047/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 358, + 24 + ], + [ + 335, + 35 + ], + [ + 349, + 35 + ], + [ + 367, + 15 + ], + [ + 376, + 23 + ], + [ + 391, + 3 + ] + ], + [ + [ + 392, + 42 + ], + [ + 420, + 36 + ], + [ + 398, + 66 + ], + [ + 387, + 20 + ], + [ + 370, + 72 + ], + [ + 360, + 26 + ] + ], + [ + [ + 394, + 47 + ], + [ + 414, + 63 + ], + [ + 385, + 73 + ], + [ + 404, + 18 + ], + [ + 368, + 58 + ], + [ + 385, + 3 + ] + ], + [ + [ + 384, + 38 + ], + [ + 387, + 11 + ], + [ + 407, + 50 + ], + [ + 362, + 27 + ], + [ + 403, + 75 + ], + [ + 360, + 51 + ] + ] + ] + }, + { + "task_id": "0005049", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0007_cfg_0004_104122062823#711", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005049/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 313, + 15 + ], + [ + 309, + 0 + ], + [ + 321, + 0 + ], + [ + 305, + 37 + ], + [ + 324, + 13 + ], + [ + 309, + 55 + ] + ], + [ + [ + 308, + 47 + ], + [ + 285, + 56 + ], + [ + 324, + 67 + ], + [ + 293, + 27 + ], + [ + 346, + 57 + ], + [ + 315, + 19 + ] + ], + [ + [ + 322, + 39 + ], + [ + 323, + 13 + ], + [ + 346, + 50 + ], + [ + 299, + 29 + ], + [ + 343, + 74 + ], + [ + 298, + 52 + ] + ], + [ + [ + 328, + 16 + ], + [ + 299, + 14 + ], + [ + 330, + 0 + ], + [ + 326, + 39 + ], + [ + 358, + 0 + ], + [ + 356, + 40 + ] + ] + ] + }, + { + "task_id": "0005052", + "id": "RH20T_cfg6_task_0002_user_0014_scene_0006_cfg_0006_104122061602#97", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005052/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 477, + 177 + ], + [ + 470, + 172 + ], + [ + 451, + 186 + ], + [ + 504, + 168 + ], + [ + 458, + 190 + ], + [ + 509, + 174 + ] + ], + [ + [ + 475, + 169 + ], + [ + 486, + 152 + ], + [ + 491, + 188 + ], + [ + 460, + 151 + ], + [ + 480, + 203 + ], + [ + 451, + 167 + ] + ], + [ + [ + 482, + 175 + ], + [ + 474, + 189 + ], + [ + 503, + 191 + ], + [ + 463, + 160 + ], + [ + 513, + 175 + ], + [ + 471, + 144 + ] + ], + [ + [ + 471, + 162 + ], + [ + 450, + 169 + ], + [ + 482, + 157 + ], + [ + 460, + 167 + ], + [ + 503, + 150 + ], + [ + 479, + 160 + ] + ] + ] + }, + { + "task_id": "0005056", + "id": "RH20T_cfg6_task_0129_user_0020_scene_0009_cfg_0006_037522061512#222", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder with a bulb to the socket'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005056/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 171, + 182 + ], + [ + 178, + 157 + ], + [ + 190, + 186 + ], + [ + 150, + 179 + ], + [ + 184, + 209 + ], + [ + 143, + 205 + ] + ], + [ + [ + 173, + 173 + ], + [ + 182, + 187 + ], + [ + 171, + 192 + ], + [ + 175, + 152 + ], + [ + 162, + 178 + ], + [ + 165, + 139 + ] + ], + [ + [ + 149, + 163 + ], + [ + 172, + 169 + ], + [ + 137, + 152 + ], + [ + 160, + 172 + ], + [ + 113, + 146 + ], + [ + 139, + 167 + ] + ], + [ + [ + 202, + 175 + ], + [ + 189, + 162 + ], + [ + 180, + 178 + ], + [ + 223, + 172 + ], + [ + 194, + 193 + ], + [ + 237, + 186 + ] + ] + ] + }, + { + "task_id": "0005058", + "id": "RH20T_cfg2_task_0065_user_0007_scene_0010_cfg_0002_105422061350#521", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005058/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 180, + 117 + ], + [ + 177, + 100 + ], + [ + 191, + 101 + ], + [ + 169, + 134 + ], + [ + 193, + 118 + ], + [ + 172, + 150 + ] + ], + [ + [ + 211, + 113 + ], + [ + 193, + 122 + ], + [ + 209, + 92 + ], + [ + 213, + 134 + ], + [ + 226, + 84 + ], + [ + 228, + 125 + ] + ], + [ + [ + 174, + 132 + ], + [ + 178, + 154 + ], + [ + 159, + 127 + ], + [ + 190, + 136 + ], + [ + 154, + 104 + ], + [ + 186, + 112 + ] + ], + [ + [ + 165, + 104 + ], + [ + 180, + 122 + ], + [ + 146, + 119 + ], + [ + 183, + 88 + ], + [ + 130, + 101 + ], + [ + 167, + 70 + ] + ] + ] + }, + { + "task_id": "0005059", + "id": "RH20T_cfg2_task_0065_user_0007_scene_0010_cfg_0002_105422061350#766", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005059/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 232, + 62 + ], + [ + 248, + 59 + ], + [ + 228, + 83 + ], + [ + 236, + 40 + ], + [ + 211, + 87 + ], + [ + 219, + 42 + ] + ], + [ + [ + 198, + 98 + ], + [ + 176, + 107 + ], + [ + 209, + 119 + ], + [ + 187, + 78 + ], + [ + 230, + 110 + ], + [ + 209, + 70 + ] + ], + [ + [ + 209, + 87 + ], + [ + 208, + 68 + ], + [ + 214, + 71 + ], + [ + 204, + 103 + ], + [ + 215, + 88 + ], + [ + 205, + 119 + ] + ], + [ + [ + 199, + 80 + ], + [ + 201, + 101 + ], + [ + 200, + 67 + ], + [ + 198, + 91 + ], + [ + 198, + 46 + ], + [ + 196, + 72 + ] + ] + ] + }, + { + "task_id": "0005061", + "id": "RH20T_cfg4_task_0018_user_0016_scene_0001_cfg_0004_104122062295#150", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play Jenga'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005061/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 154, + 248 + ], + [ + 185, + 264 + ], + [ + 181, + 210 + ], + [ + 125, + 289 + ], + [ + 153, + 198 + ], + [ + 98, + 272 + ] + ], + [ + [ + 143, + 219 + ], + [ + 147, + 184 + ], + [ + 95, + 237 + ], + [ + 185, + 204 + ], + [ + 95, + 270 + ], + [ + 179, + 234 + ] + ], + [ + [ + 163, + 175 + ], + [ + 185, + 171 + ], + [ + 131, + 140 + ], + [ + 193, + 207 + ], + [ + 114, + 146 + ], + [ + 173, + 208 + ] + ], + [ + [ + 92, + 222 + ], + [ + 134, + 195 + ], + [ + 78, + 191 + ], + [ + 107, + 256 + ], + [ + 36, + 216 + ], + [ + 60, + 287 + ] + ] + ] + }, + { + "task_id": "0005062", + "id": "RH20T_cfg1_task_0009_user_0014_scene_0004_cfg_0001_038522063145#206", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005062/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 436, + 94 + ], + [ + 446, + 74 + ], + [ + 426, + 91 + ], + [ + 444, + 97 + ], + [ + 415, + 110 + ], + [ + 434, + 115 + ] + ], + [ + [ + 421, + 106 + ], + [ + 425, + 89 + ], + [ + 442, + 113 + ], + [ + 401, + 99 + ], + [ + 438, + 129 + ], + [ + 399, + 115 + ] + ], + [ + [ + 402, + 102 + ], + [ + 395, + 84 + ], + [ + 416, + 89 + ], + [ + 390, + 113 + ], + [ + 424, + 108 + ], + [ + 397, + 131 + ] + ], + [ + [ + 440, + 106 + ], + [ + 422, + 95 + ], + [ + 426, + 125 + ], + [ + 454, + 87 + ], + [ + 444, + 137 + ], + [ + 472, + 99 + ] + ] + ] + }, + { + "task_id": "0005064", + "id": "RH20T_cfg5_task_0016_user_0010_scene_0001_cfg_0005_036422060215#263", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005064/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 302, + 174 + ], + [ + 294, + 155 + ], + [ + 293, + 186 + ], + [ + 313, + 161 + ], + [ + 301, + 205 + ], + [ + 322, + 181 + ] + ], + [ + [ + 281, + 158 + ], + [ + 276, + 138 + ], + [ + 283, + 169 + ], + [ + 279, + 145 + ], + [ + 288, + 189 + ], + [ + 285, + 167 + ] + ], + [ + [ + 282, + 178 + ], + [ + 271, + 163 + ], + [ + 262, + 175 + ], + [ + 300, + 181 + ], + [ + 274, + 193 + ], + [ + 313, + 197 + ] + ], + [ + [ + 290, + 172 + ], + [ + 293, + 153 + ], + [ + 311, + 167 + ], + [ + 269, + 177 + ], + [ + 308, + 184 + ], + [ + 267, + 195 + ] + ] + ] + }, + { + "task_id": "0005066", + "id": "RH20T_cfg4_task_0001_user_0010_scene_0004_cfg_0004_104122062295#215", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005066/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 105, + 161 + ], + [ + 113, + 119 + ], + [ + 55, + 154 + ], + [ + 152, + 168 + ], + [ + 45, + 200 + ], + [ + 145, + 212 + ] + ], + [ + [ + 164, + 104 + ], + [ + 152, + 134 + ], + [ + 153, + 131 + ], + [ + 174, + 80 + ], + [ + 166, + 97 + ], + [ + 186, + 48 + ] + ], + [ + [ + 86, + 151 + ], + [ + 81, + 107 + ], + [ + 127, + 154 + ], + [ + 38, + 148 + ], + [ + 132, + 196 + ], + [ + 43, + 197 + ] + ], + [ + [ + 119, + 139 + ], + [ + 113, + 110 + ], + [ + 73, + 129 + ], + [ + 164, + 148 + ], + [ + 82, + 155 + ], + [ + 166, + 172 + ] + ] + ] + }, + { + "task_id": "0005068", + "id": "RH20T_cfg4_task_0001_user_0010_scene_0004_cfg_0004_104122062295#360", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005068/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 103, + 162 + ], + [ + 96, + 132 + ], + [ + 58, + 148 + ], + [ + 148, + 175 + ], + [ + 67, + 175 + ], + [ + 151, + 199 + ] + ], + [ + [ + 91, + 165 + ], + [ + 136, + 185 + ], + [ + 106, + 186 + ], + [ + 73, + 139 + ], + [ + 66, + 167 + ], + [ + 25, + 117 + ] + ], + [ + [ + 108, + 157 + ], + [ + 143, + 156 + ], + [ + 108, + 115 + ], + [ + 107, + 199 + ], + [ + 67, + 112 + ], + [ + 66, + 203 + ] + ], + [ + [ + 63, + 184 + ], + [ + 55, + 214 + ], + [ + 29, + 153 + ], + [ + 98, + 215 + ], + [ + 37, + 117 + ], + [ + 110, + 183 + ] + ] + ] + }, + { + "task_id": "0005069", + "id": "RH20T_cfg4_task_0001_user_0010_scene_0004_cfg_0004_104122062295#399", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005069/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 96, + 150 + ], + [ + 87, + 115 + ], + [ + 50, + 137 + ], + [ + 142, + 163 + ], + [ + 61, + 167 + ], + [ + 146, + 191 + ] + ], + [ + [ + 97, + 155 + ], + [ + 112, + 133 + ], + [ + 144, + 148 + ], + [ + 42, + 164 + ], + [ + 128, + 166 + ], + [ + 31, + 185 + ] + ], + [ + [ + 49, + 150 + ], + [ + 75, + 190 + ], + [ + 79, + 152 + ], + [ + 12, + 148 + ], + [ + 56, + 116 + ], + [ + 0, + 103 + ] + ], + [ + [ + 73, + 105 + ], + [ + 89, + 89 + ], + [ + 61, + 147 + ], + [ + 86, + 59 + ], + [ + 42, + 170 + ], + [ + 68, + 74 + ] + ] + ] + }, + { + "task_id": "0005071", + "id": "RH20T_cfg3_task_0062_user_0010_scene_0008_cfg_0003_104122062823#167", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the toilet paper on its holder'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005071/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 262, + 166 + ], + [ + 267, + 196 + ], + [ + 249, + 181 + ], + [ + 278, + 150 + ], + [ + 244, + 154 + ], + [ + 272, + 122 + ] + ], + [ + [ + 234, + 149 + ], + [ + 241, + 132 + ], + [ + 203, + 136 + ], + [ + 265, + 162 + ], + [ + 193, + 154 + ], + [ + 259, + 181 + ] + ], + [ + [ + 254, + 184 + ], + [ + 273, + 164 + ], + [ + 274, + 174 + ], + [ + 231, + 196 + ], + [ + 258, + 192 + ], + [ + 215, + 215 + ] + ], + [ + [ + 239, + 146 + ], + [ + 243, + 114 + ], + [ + 274, + 152 + ], + [ + 204, + 139 + ], + [ + 269, + 180 + ], + [ + 202, + 169 + ] + ] + ] + }, + { + "task_id": "0005074", + "id": "RH20T_cfg3_task_0062_user_0010_scene_0008_cfg_0003_104122062823#223", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the toilet paper on its holder'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005074/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 248, + 178 + ], + [ + 237, + 202 + ], + [ + 216, + 170 + ], + [ + 282, + 186 + ], + [ + 227, + 149 + ], + [ + 290, + 164 + ] + ], + [ + [ + 270, + 151 + ], + [ + 262, + 120 + ], + [ + 264, + 137 + ], + [ + 278, + 167 + ], + [ + 272, + 165 + ], + [ + 286, + 197 + ] + ], + [ + [ + 240, + 176 + ], + [ + 205, + 163 + ], + [ + 233, + 161 + ], + [ + 248, + 193 + ], + [ + 263, + 172 + ], + [ + 281, + 205 + ] + ], + [ + [ + 237, + 161 + ], + [ + 241, + 133 + ], + [ + 271, + 168 + ], + [ + 201, + 155 + ], + [ + 266, + 193 + ], + [ + 199, + 182 + ] + ] + ] + }, + { + "task_id": "0005076", + "id": "RH20T_cfg6_task_0002_user_0014_scene_0006_cfg_0006_104122063633#213", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005076/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 185, + 192 + ], + [ + 185, + 203 + ], + [ + 206, + 204 + ], + [ + 163, + 180 + ], + [ + 205, + 194 + ], + [ + 164, + 170 + ] + ], + [ + [ + 146, + 157 + ], + [ + 122, + 149 + ], + [ + 135, + 178 + ], + [ + 156, + 136 + ], + [ + 160, + 187 + ], + [ + 180, + 145 + ] + ], + [ + [ + 179, + 173 + ], + [ + 156, + 171 + ], + [ + 184, + 151 + ], + [ + 174, + 196 + ], + [ + 208, + 154 + ], + [ + 198, + 199 + ] + ], + [ + [ + 170, + 174 + ], + [ + 168, + 154 + ], + [ + 176, + 160 + ], + [ + 163, + 189 + ], + [ + 178, + 179 + ], + [ + 166, + 208 + ] + ] + ] + }, + { + "task_id": "0005078", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122061602#235", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005078/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 450, + 159 + ], + [ + 475, + 155 + ], + [ + 451, + 140 + ], + [ + 449, + 179 + ], + [ + 429, + 144 + ], + [ + 426, + 182 + ] + ], + [ + [ + 471, + 168 + ], + [ + 477, + 151 + ], + [ + 487, + 184 + ], + [ + 456, + 153 + ], + [ + 480, + 200 + ], + [ + 451, + 169 + ] + ], + [ + [ + 494, + 163 + ], + [ + 514, + 156 + ], + [ + 504, + 185 + ], + [ + 485, + 140 + ], + [ + 483, + 193 + ], + [ + 465, + 147 + ] + ], + [ + [ + 488, + 169 + ], + [ + 515, + 166 + ], + [ + 491, + 194 + ], + [ + 485, + 144 + ], + [ + 466, + 196 + ], + [ + 459, + 147 + ] + ] + ] + }, + { + "task_id": "0005079", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122061602#571", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005079/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 484, + 205 + ], + [ + 506, + 201 + ], + [ + 470, + 198 + ], + [ + 499, + 213 + ], + [ + 448, + 202 + ], + [ + 475, + 218 + ] + ], + [ + [ + 490, + 214 + ], + [ + 509, + 216 + ], + [ + 494, + 239 + ], + [ + 486, + 188 + ], + [ + 478, + 236 + ], + [ + 470, + 187 + ] + ], + [ + [ + 480, + 238 + ], + [ + 477, + 243 + ], + [ + 478, + 265 + ], + [ + 481, + 211 + ], + [ + 481, + 260 + ], + [ + 484, + 209 + ] + ], + [ + [ + 486, + 219 + ], + [ + 494, + 237 + ], + [ + 488, + 237 + ], + [ + 485, + 200 + ], + [ + 481, + 221 + ], + [ + 478, + 184 + ] + ] + ] + }, + { + "task_id": "0005080", + "id": "RH20T_cfg6_task_0105_user_0020_scene_0007_cfg_0006_104122061602#869", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object on the shelf'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005080/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 561, + 232 + ], + [ + 552, + 239 + ], + [ + 550, + 206 + ], + [ + 571, + 257 + ], + [ + 558, + 200 + ], + [ + 578, + 249 + ] + ], + [ + [ + 517, + 198 + ], + [ + 525, + 174 + ], + [ + 494, + 194 + ], + [ + 542, + 204 + ], + [ + 486, + 217 + ], + [ + 534, + 229 + ] + ], + [ + [ + 539, + 214 + ], + [ + 557, + 206 + ], + [ + 552, + 202 + ], + [ + 527, + 225 + ], + [ + 532, + 210 + ], + [ + 508, + 233 + ] + ], + [ + [ + 537, + 206 + ], + [ + 535, + 183 + ], + [ + 556, + 216 + ], + [ + 519, + 197 + ], + [ + 557, + 239 + ], + [ + 520, + 218 + ] + ] + ] + }, + { + "task_id": "0005083", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0002_cfg_0001_750612070853#149", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005083/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 0, + 80 + ], + [ + 0, + 50 + ], + [ + 0, + 43 + ], + [ + 24, + 112 + ], + [ + 0, + 79 + ], + [ + 23, + 145 + ] + ], + [ + [ + 0, + 136 + ], + [ + 0, + 91 + ], + [ + 0, + 165 + ], + [ + 0, + 101 + ], + [ + 0, + 212 + ], + [ + 0, + 156 + ] + ], + [ + [ + 0, + 117 + ], + [ + 0, + 72 + ], + [ + 0, + 100 + ], + [ + 4, + 134 + ], + [ + 0, + 138 + ], + [ + 30, + 172 + ] + ], + [ + [ + 0, + 82 + ], + [ + 0, + 113 + ], + [ + 0, + 45 + ], + [ + 0, + 113 + ], + [ + 0, + 20 + ], + [ + 0, + 87 + ] + ] + ] + }, + { + "task_id": "0005084", + "id": "RH20T_cfg1_task_0040_user_0010_scene_0002_cfg_0001_750612070853#262", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the cups'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005084/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 227, + 233 + ], + [ + 216, + 189 + ], + [ + 174, + 208 + ], + [ + 285, + 259 + ], + [ + 188, + 244 + ], + [ + 288, + 293 + ] + ], + [ + [ + 186, + 242 + ], + [ + 134, + 256 + ], + [ + 172, + 183 + ], + [ + 201, + 302 + ], + [ + 215, + 178 + ], + [ + 241, + 285 + ] + ], + [ + [ + 212, + 208 + ], + [ + 232, + 177 + ], + [ + 267, + 200 + ], + [ + 162, + 216 + ], + [ + 249, + 240 + ], + [ + 133, + 254 + ] + ], + [ + [ + 134, + 241 + ], + [ + 118, + 289 + ], + [ + 196, + 274 + ], + [ + 64, + 203 + ], + [ + 201, + 236 + ], + [ + 87, + 169 + ] + ] + ] + }, + { + "task_id": "0005085", + "id": "RH20T_cfg6_task_0060_user_0014_scene_0006_cfg_0006_104122061330#136", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to decrease the volume of the speaker'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005085/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 168 + ], + [ + 294, + 182 + ], + [ + 307, + 189 + ], + [ + 312, + 147 + ], + [ + 321, + 176 + ], + [ + 326, + 134 + ] + ], + [ + [ + 326, + 146 + ], + [ + 338, + 125 + ], + [ + 350, + 154 + ], + [ + 303, + 138 + ], + [ + 337, + 176 + ], + [ + 291, + 159 + ] + ], + [ + [ + 340, + 170 + ], + [ + 335, + 145 + ], + [ + 314, + 177 + ], + [ + 366, + 163 + ], + [ + 320, + 201 + ], + [ + 371, + 187 + ] + ], + [ + [ + 278, + 186 + ], + [ + 299, + 171 + ], + [ + 288, + 195 + ], + [ + 266, + 176 + ], + [ + 267, + 209 + ], + [ + 244, + 191 + ] + ] + ] + }, + { + "task_id": "0005086", + "id": "RH20T_cfg6_task_0038_user_0014_scene_0010_cfg_0006_104122061018#245", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the cup'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005086/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 353, + 52 + ], + [ + 320, + 41 + ], + [ + 356, + 63 + ], + [ + 350, + 43 + ], + [ + 389, + 74 + ], + [ + 379, + 54 + ] + ], + [ + [ + 369, + 58 + ], + [ + 357, + 74 + ], + [ + 356, + 27 + ], + [ + 382, + 88 + ], + [ + 369, + 14 + ], + [ + 392, + 73 + ] + ], + [ + [ + 384, + 51 + ], + [ + 396, + 22 + ], + [ + 371, + 24 + ], + [ + 398, + 79 + ], + [ + 361, + 50 + ], + [ + 386, + 103 + ] + ], + [ + [ + 407, + 68 + ], + [ + 437, + 67 + ], + [ + 418, + 101 + ], + [ + 398, + 36 + ], + [ + 386, + 103 + ], + [ + 366, + 36 + ] + ] + ] + }, + { + "task_id": "0005088", + "id": "RH20T_cfg5_task_0080_user_0007_scene_0010_cfg_0005_036422060215#238", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Flip over and spread out the paper that is laid flat on the table'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005088/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 181, + 198 + ], + [ + 171, + 214 + ], + [ + 197, + 214 + ], + [ + 164, + 181 + ], + [ + 206, + 198 + ], + [ + 175, + 166 + ] + ], + [ + [ + 205, + 199 + ], + [ + 208, + 183 + ], + [ + 218, + 216 + ], + [ + 193, + 184 + ], + [ + 215, + 231 + ], + [ + 191, + 199 + ] + ], + [ + [ + 176, + 189 + ], + [ + 181, + 203 + ], + [ + 164, + 174 + ], + [ + 189, + 205 + ], + [ + 159, + 159 + ], + [ + 184, + 191 + ] + ], + [ + [ + 196, + 225 + ], + [ + 174, + 236 + ], + [ + 210, + 241 + ], + [ + 182, + 210 + ], + [ + 231, + 229 + ], + [ + 203, + 200 + ] + ] + ] + }, + { + "task_id": "0005090", + "id": "RH20T_cfg6_task_0020_user_0014_scene_0001_cfg_0006_104122064161#135", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the desk lamp by pressing the button'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005090/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 302, + 84 + ], + [ + 302, + 63 + ], + [ + 318, + 84 + ], + [ + 285, + 84 + ], + [ + 319, + 104 + ], + [ + 286, + 105 + ] + ], + [ + [ + 323, + 102 + ], + [ + 322, + 92 + ], + [ + 304, + 98 + ], + [ + 342, + 105 + ], + [ + 305, + 108 + ], + [ + 344, + 115 + ] + ], + [ + [ + 313, + 84 + ], + [ + 316, + 73 + ], + [ + 302, + 98 + ], + [ + 323, + 71 + ], + [ + 300, + 108 + ], + [ + 319, + 81 + ] + ], + [ + [ + 317, + 104 + ], + [ + 328, + 109 + ], + [ + 306, + 92 + ], + [ + 328, + 117 + ], + [ + 296, + 88 + ], + [ + 318, + 112 + ] + ] + ] + }, + { + "task_id": "0005091", + "id": "RH20T_cfg6_task_0029_user_0014_scene_0003_cfg_0006_104122063633#222", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005091/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 174, + 203 + ], + [ + 174, + 174 + ], + [ + 181, + 192 + ], + [ + 166, + 215 + ], + [ + 180, + 217 + ], + [ + 166, + 244 + ] + ], + [ + [ + 158, + 187 + ], + [ + 163, + 186 + ], + [ + 177, + 163 + ], + [ + 140, + 211 + ], + [ + 173, + 162 + ], + [ + 133, + 213 + ] + ], + [ + [ + 202, + 198 + ], + [ + 211, + 206 + ], + [ + 231, + 205 + ], + [ + 172, + 189 + ], + [ + 222, + 198 + ], + [ + 167, + 183 + ] + ], + [ + [ + 179, + 208 + ], + [ + 189, + 223 + ], + [ + 151, + 222 + ], + [ + 206, + 194 + ], + [ + 138, + 205 + ], + [ + 196, + 177 + ] + ] + ] + }, + { + "task_id": "0005092", + "id": "RH20T_cfg4_task_0068_user_0007_scene_0005_cfg_0004_104422070011#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button horizontally with obstacles'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005092/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 129, + 255 + ], + [ + 115, + 226 + ], + [ + 95, + 268 + ], + [ + 160, + 244 + ], + [ + 110, + 300 + ], + [ + 175, + 272 + ] + ], + [ + [ + 152, + 256 + ], + [ + 154, + 243 + ], + [ + 116, + 271 + ], + [ + 184, + 242 + ], + [ + 116, + 282 + ], + [ + 180, + 254 + ] + ], + [ + [ + 127, + 262 + ], + [ + 115, + 227 + ], + [ + 147, + 259 + ], + [ + 103, + 265 + ], + [ + 159, + 292 + ], + [ + 116, + 305 + ] + ], + [ + [ + 161, + 248 + ], + [ + 192, + 255 + ], + [ + 179, + 225 + ], + [ + 143, + 270 + ], + [ + 145, + 216 + ], + [ + 108, + 264 + ] + ] + ] + }, + { + "task_id": "0005093", + "id": "RH20T_cfg5_task_0075_user_0007_scene_0001_cfg_0005_104422070044#141", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Finish setting up the starting position of a chessboard that is almost arranged'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005093/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 373, + 129 + ], + [ + 371, + 117 + ], + [ + 373, + 141 + ], + [ + 373, + 118 + ], + [ + 375, + 153 + ], + [ + 376, + 130 + ] + ], + [ + [ + 388, + 115 + ], + [ + 371, + 113 + ], + [ + 386, + 129 + ], + [ + 390, + 100 + ], + [ + 403, + 132 + ], + [ + 407, + 103 + ] + ], + [ + [ + 374, + 123 + ], + [ + 360, + 120 + ], + [ + 367, + 137 + ], + [ + 380, + 108 + ], + [ + 381, + 140 + ], + [ + 394, + 111 + ] + ], + [ + [ + 359, + 137 + ], + [ + 366, + 152 + ], + [ + 363, + 135 + ], + [ + 355, + 139 + ], + [ + 355, + 120 + ], + [ + 348, + 124 + ] + ] + ] + }, + { + "task_id": "0005095", + "id": "RH20T_cfg1_task_0055_user_0015_scene_0009_cfg_0001_750612070853#69", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Transfer liquid using a dropper'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005095/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 439, + 75 + ], + [ + 420, + 119 + ], + [ + 474, + 91 + ], + [ + 408, + 61 + ], + [ + 497, + 40 + ], + [ + 425, + 16 + ] + ], + [ + [ + 444, + 113 + ], + [ + 444, + 67 + ], + [ + 395, + 113 + ], + [ + 498, + 114 + ], + [ + 398, + 152 + ], + [ + 495, + 157 + ] + ], + [ + [ + 446, + 151 + ], + [ + 498, + 136 + ], + [ + 433, + 104 + ], + [ + 458, + 200 + ], + [ + 386, + 119 + ], + [ + 408, + 212 + ] + ], + [ + [ + 442, + 96 + ], + [ + 437, + 77 + ], + [ + 487, + 104 + ], + [ + 397, + 88 + ], + [ + 496, + 128 + ], + [ + 399, + 110 + ] + ] + ] + }, + { + "task_id": "0005099", + "id": "RH20T_cfg7_task_0047_user_0020_scene_0004_cfg_0007_104122061018#143", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005099/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 299, + 102 + ], + [ + 279, + 105 + ], + [ + 270, + 104 + ], + [ + 325, + 100 + ], + [ + 292, + 101 + ], + [ + 348, + 98 + ] + ], + [ + [ + 271, + 118 + ], + [ + 256, + 143 + ], + [ + 303, + 134 + ], + [ + 237, + 103 + ], + [ + 320, + 107 + ], + [ + 251, + 74 + ] + ], + [ + [ + 276, + 115 + ], + [ + 290, + 143 + ], + [ + 276, + 131 + ], + [ + 276, + 101 + ], + [ + 260, + 99 + ], + [ + 262, + 71 + ] + ], + [ + [ + 279, + 94 + ], + [ + 270, + 65 + ], + [ + 261, + 113 + ], + [ + 299, + 74 + ], + [ + 270, + 142 + ], + [ + 311, + 106 + ] + ] + ] + }, + { + "task_id": "0005103", + "id": "RH20T_cfg5_task_0057_user_0007_scene_0003_cfg_0005_036422060215#435", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them on'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005103/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 275, + 178 + ], + [ + 260, + 194 + ], + [ + 260, + 168 + ], + [ + 290, + 187 + ], + [ + 275, + 153 + ], + [ + 305, + 171 + ] + ], + [ + [ + 274, + 182 + ], + [ + 290, + 171 + ], + [ + 270, + 195 + ], + [ + 279, + 168 + ], + [ + 254, + 208 + ], + [ + 262, + 181 + ] + ], + [ + [ + 276, + 199 + ], + [ + 295, + 198 + ], + [ + 265, + 203 + ], + [ + 287, + 196 + ], + [ + 245, + 204 + ], + [ + 266, + 197 + ] + ], + [ + [ + 262, + 197 + ], + [ + 268, + 181 + ], + [ + 270, + 215 + ], + [ + 255, + 180 + ], + [ + 264, + 229 + ], + [ + 251, + 195 + ] + ] + ] + }, + { + "task_id": "0005108", + "id": "RH20T_cfg4_task_0011_user_0016_scene_0002_cfg_0004_104122062823#154", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005108/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 164, + 0 + ], + [ + 139, + 0 + ], + [ + 137, + 7 + ], + [ + 188, + 0 + ], + [ + 165, + 14 + ], + [ + 214, + 0 + ] + ], + [ + [ + 148, + 0 + ], + [ + 133, + 0 + ], + [ + 172, + 0 + ], + [ + 122, + 15 + ], + [ + 183, + 15 + ], + [ + 138, + 47 + ] + ], + [ + [ + 137, + 0 + ], + [ + 148, + 10 + ], + [ + 156, + 17 + ], + [ + 115, + 0 + ], + [ + 146, + 2 + ], + [ + 105, + 0 + ] + ], + [ + [ + 132, + 22 + ], + [ + 116, + 0 + ], + [ + 141, + 0 + ], + [ + 124, + 48 + ], + [ + 159, + 16 + ], + [ + 140, + 71 + ] + ] + ] + }, + { + "task_id": "0005109", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0009_cfg_0007_037522061512#191", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005109/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 223, + 150 + ], + [ + 237, + 152 + ], + [ + 217, + 173 + ], + [ + 229, + 127 + ], + [ + 202, + 173 + ], + [ + 214, + 125 + ] + ], + [ + [ + 237, + 131 + ], + [ + 222, + 142 + ], + [ + 218, + 116 + ], + [ + 255, + 146 + ], + [ + 232, + 106 + ], + [ + 268, + 134 + ] + ], + [ + [ + 231, + 128 + ], + [ + 244, + 111 + ], + [ + 244, + 126 + ], + [ + 217, + 131 + ], + [ + 232, + 142 + ], + [ + 204, + 149 + ] + ], + [ + [ + 219, + 170 + ], + [ + 234, + 189 + ], + [ + 230, + 167 + ], + [ + 206, + 173 + ], + [ + 216, + 148 + ], + [ + 190, + 153 + ] + ] + ] + }, + { + "task_id": "0005110", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0009_cfg_0007_037522061512#428", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005110/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 225, + 151 + ], + [ + 239, + 152 + ], + [ + 220, + 174 + ], + [ + 231, + 128 + ], + [ + 205, + 174 + ], + [ + 216, + 126 + ] + ], + [ + [ + 228, + 145 + ], + [ + 239, + 152 + ], + [ + 214, + 128 + ], + [ + 242, + 160 + ], + [ + 204, + 122 + ], + [ + 231, + 153 + ] + ], + [ + [ + 203, + 160 + ], + [ + 225, + 150 + ], + [ + 190, + 142 + ], + [ + 216, + 177 + ], + [ + 169, + 153 + ], + [ + 195, + 187 + ] + ], + [ + [ + 234, + 154 + ], + [ + 218, + 144 + ], + [ + 228, + 176 + ], + [ + 242, + 132 + ], + [ + 245, + 187 + ], + [ + 260, + 143 + ] + ] + ] + }, + { + "task_id": "0005111", + "id": "RH20T_cfg1_task_0121_user_0010_scene_0002_cfg_0001_038522063145#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out the light bulb by rotating it'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005111/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 513, + 112 + ], + [ + 539, + 113 + ], + [ + 513, + 97 + ], + [ + 513, + 126 + ], + [ + 488, + 97 + ], + [ + 489, + 125 + ] + ], + [ + [ + 537, + 119 + ], + [ + 548, + 98 + ], + [ + 553, + 115 + ], + [ + 521, + 123 + ], + [ + 543, + 134 + ], + [ + 511, + 142 + ] + ], + [ + [ + 501, + 132 + ], + [ + 522, + 135 + ], + [ + 491, + 118 + ], + [ + 511, + 146 + ], + [ + 472, + 117 + ], + [ + 492, + 143 + ] + ], + [ + [ + 547, + 115 + ], + [ + 545, + 94 + ], + [ + 521, + 116 + ], + [ + 574, + 113 + ], + [ + 523, + 138 + ], + [ + 577, + 135 + ] + ] + ] + }, + { + "task_id": "0005112", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122063633#123", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005112/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 150, + 190 + ], + [ + 127, + 200 + ], + [ + 138, + 175 + ], + [ + 162, + 203 + ], + [ + 163, + 165 + ], + [ + 184, + 194 + ] + ], + [ + [ + 185, + 227 + ], + [ + 186, + 202 + ], + [ + 212, + 228 + ], + [ + 158, + 227 + ], + [ + 211, + 253 + ], + [ + 156, + 254 + ] + ], + [ + [ + 192, + 186 + ], + [ + 186, + 163 + ], + [ + 191, + 179 + ], + [ + 193, + 193 + ], + [ + 197, + 201 + ], + [ + 200, + 217 + ] + ], + [ + [ + 170, + 194 + ], + [ + 169, + 175 + ], + [ + 176, + 177 + ], + [ + 164, + 213 + ], + [ + 177, + 194 + ], + [ + 165, + 231 + ] + ] + ] + }, + { + "task_id": "0005114", + "id": "RH20T_cfg7_task_0061_user_0020_scene_0001_cfg_0007_104122063633#447", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005114/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 186, + 208 + ], + [ + 206, + 221 + ], + [ + 172, + 227 + ], + [ + 201, + 189 + ], + [ + 151, + 213 + ], + [ + 182, + 175 + ] + ], + [ + [ + 194, + 202 + ], + [ + 213, + 210 + ], + [ + 210, + 188 + ], + [ + 178, + 216 + ], + [ + 190, + 178 + ], + [ + 157, + 207 + ] + ], + [ + [ + 185, + 197 + ], + [ + 188, + 175 + ], + [ + 187, + 187 + ], + [ + 182, + 208 + ], + [ + 184, + 207 + ], + [ + 178, + 230 + ] + ], + [ + [ + 166, + 179 + ], + [ + 178, + 158 + ], + [ + 173, + 187 + ], + [ + 158, + 170 + ], + [ + 160, + 208 + ], + [ + 144, + 193 + ] + ] + ] + }, + { + "task_id": "0005117", + "id": "RH20T_cfg4_task_0096_user_0014_scene_0005_cfg_0004_104122062295#282", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005117/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 256, + 117 + ], + [ + 227, + 150 + ], + [ + 220, + 98 + ], + [ + 287, + 132 + ], + [ + 253, + 59 + ], + [ + 318, + 100 + ] + ], + [ + [ + 226, + 63 + ], + [ + 231, + 16 + ], + [ + 176, + 60 + ], + [ + 273, + 66 + ], + [ + 175, + 103 + ], + [ + 267, + 107 + ] + ], + [ + [ + 253, + 58 + ], + [ + 232, + 94 + ], + [ + 254, + 15 + ], + [ + 253, + 94 + ], + [ + 274, + 0 + ], + [ + 270, + 63 + ] + ], + [ + [ + 237, + 30 + ], + [ + 238, + 47 + ], + [ + 224, + 0 + ], + [ + 249, + 71 + ], + [ + 224, + 0 + ], + [ + 248, + 54 + ] + ] + ] + }, + { + "task_id": "0005118", + "id": "RH20T_cfg5_task_0013_user_0007_scene_0001_cfg_0005_104122063678#292", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the block on the scale'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005118/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 261, + 163 + ], + [ + 273, + 143 + ], + [ + 282, + 175 + ], + [ + 240, + 152 + ], + [ + 270, + 195 + ], + [ + 228, + 172 + ] + ], + [ + [ + 245, + 141 + ], + [ + 249, + 124 + ], + [ + 221, + 138 + ], + [ + 268, + 144 + ], + [ + 217, + 154 + ], + [ + 263, + 159 + ] + ], + [ + [ + 253, + 146 + ], + [ + 259, + 129 + ], + [ + 258, + 163 + ], + [ + 247, + 130 + ], + [ + 252, + 179 + ], + [ + 242, + 145 + ] + ], + [ + [ + 272, + 150 + ], + [ + 284, + 132 + ], + [ + 286, + 165 + ], + [ + 258, + 136 + ], + [ + 273, + 183 + ], + [ + 246, + 154 + ] + ] + ] + }, + { + "task_id": "0005119", + "id": "RH20T_cfg4_task_0067_user_0007_scene_0007_cfg_0004_104122062295#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press a button from top to bottom with obstacles'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005119/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 330, + 103 + ], + [ + 297, + 114 + ], + [ + 337, + 138 + ], + [ + 323, + 66 + ], + [ + 366, + 128 + ], + [ + 355, + 57 + ] + ], + [ + [ + 278, + 88 + ], + [ + 299, + 72 + ], + [ + 301, + 116 + ], + [ + 255, + 60 + ], + [ + 280, + 135 + ], + [ + 231, + 76 + ] + ], + [ + [ + 291, + 100 + ], + [ + 296, + 64 + ], + [ + 267, + 85 + ], + [ + 317, + 117 + ], + [ + 264, + 117 + ], + [ + 311, + 150 + ] + ], + [ + [ + 280, + 128 + ], + [ + 258, + 133 + ], + [ + 286, + 95 + ], + [ + 273, + 162 + ], + [ + 310, + 88 + ], + [ + 297, + 159 + ] + ] + ] + }, + { + "task_id": "0005120", + "id": "RH20T_cfg1_task_0016_user_0014_scene_0003_cfg_0001_039422060546#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005120/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 297, + 77 + ], + [ + 293, + 59 + ], + [ + 322, + 83 + ], + [ + 274, + 71 + ], + [ + 325, + 100 + ], + [ + 279, + 88 + ] + ], + [ + [ + 282, + 82 + ], + [ + 284, + 107 + ], + [ + 297, + 82 + ], + [ + 268, + 83 + ], + [ + 295, + 55 + ], + [ + 266, + 58 + ] + ], + [ + [ + 296, + 71 + ], + [ + 304, + 46 + ], + [ + 273, + 65 + ], + [ + 321, + 77 + ], + [ + 265, + 89 + ], + [ + 313, + 102 + ] + ], + [ + [ + 299, + 70 + ], + [ + 289, + 65 + ], + [ + 280, + 59 + ], + [ + 320, + 81 + ], + [ + 290, + 64 + ], + [ + 328, + 85 + ] + ] + ] + }, + { + "task_id": "0005123", + "id": "RH20T_cfg1_task_0090_user_0007_scene_0001_cfg_0001_035622060973#153", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch an object with obstacles'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005123/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 228, + 98 + ], + [ + 225, + 84 + ], + [ + 220, + 84 + ], + [ + 236, + 111 + ], + [ + 223, + 97 + ], + [ + 239, + 123 + ] + ], + [ + [ + 217, + 77 + ], + [ + 202, + 90 + ], + [ + 210, + 70 + ], + [ + 223, + 83 + ], + [ + 225, + 56 + ], + [ + 236, + 70 + ] + ], + [ + [ + 240, + 97 + ], + [ + 248, + 88 + ], + [ + 238, + 81 + ], + [ + 242, + 114 + ], + [ + 232, + 90 + ], + [ + 235, + 122 + ] + ], + [ + [ + 215, + 95 + ], + [ + 230, + 83 + ], + [ + 202, + 80 + ], + [ + 227, + 111 + ], + [ + 187, + 92 + ], + [ + 212, + 123 + ] + ] + ] + }, + { + "task_id": "0005124", + "id": "RH20T_cfg1_task_0016_user_0001_scene_0004_cfg_0001_038522062547#192", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005124/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 390, + 223 + ], + [ + 402, + 205 + ], + [ + 390, + 199 + ], + [ + 391, + 248 + ], + [ + 380, + 216 + ], + [ + 380, + 264 + ] + ], + [ + [ + 375, + 271 + ], + [ + 370, + 246 + ], + [ + 347, + 274 + ], + [ + 403, + 268 + ], + [ + 352, + 301 + ], + [ + 410, + 294 + ] + ], + [ + [ + 379, + 243 + ], + [ + 378, + 234 + ], + [ + 408, + 236 + ], + [ + 350, + 249 + ], + [ + 406, + 244 + ], + [ + 352, + 256 + ] + ], + [ + [ + 402, + 209 + ], + [ + 426, + 192 + ], + [ + 418, + 226 + ], + [ + 387, + 192 + ], + [ + 392, + 243 + ], + [ + 363, + 208 + ] + ] + ] + }, + { + "task_id": "0005126", + "id": "RH20T_cfg5_task_0001_user_0014_scene_0010_cfg_0005_036422060215#197", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the button from top to bottom'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005126/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 239, + 140 + ], + [ + 233, + 156 + ], + [ + 217, + 132 + ], + [ + 262, + 147 + ], + [ + 224, + 118 + ], + [ + 267, + 132 + ] + ], + [ + [ + 236, + 131 + ], + [ + 239, + 114 + ], + [ + 255, + 126 + ], + [ + 216, + 137 + ], + [ + 252, + 141 + ], + [ + 214, + 153 + ] + ], + [ + [ + 233, + 123 + ], + [ + 227, + 139 + ], + [ + 239, + 137 + ], + [ + 228, + 108 + ], + [ + 244, + 122 + ], + [ + 234, + 92 + ] + ], + [ + [ + 244, + 115 + ], + [ + 250, + 129 + ], + [ + 265, + 119 + ], + [ + 222, + 112 + ], + [ + 260, + 107 + ], + [ + 217, + 100 + ] + ] + ] + }, + { + "task_id": "0005127", + "id": "RH20T_cfg1_task_0065_user_0014_scene_0002_cfg_0001_819112070044#230", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005127/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 422, + 188 + ], + [ + 425, + 173 + ], + [ + 431, + 179 + ], + [ + 413, + 198 + ], + [ + 428, + 193 + ], + [ + 410, + 213 + ] + ], + [ + [ + 430, + 203 + ], + [ + 443, + 188 + ], + [ + 417, + 189 + ], + [ + 443, + 216 + ], + [ + 403, + 204 + ], + [ + 430, + 231 + ] + ], + [ + [ + 422, + 190 + ], + [ + 443, + 190 + ], + [ + 421, + 170 + ], + [ + 423, + 210 + ], + [ + 401, + 170 + ], + [ + 404, + 209 + ] + ], + [ + [ + 429, + 200 + ], + [ + 410, + 207 + ], + [ + 424, + 193 + ], + [ + 433, + 208 + ], + [ + 442, + 187 + ], + [ + 453, + 201 + ] + ] + ] + }, + { + "task_id": "0005129", + "id": "RH20T_cfg1_task_0016_user_0014_scene_0003_cfg_0001_750612070851#164", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005129/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 472, + 214 + ], + [ + 456, + 236 + ], + [ + 484, + 235 + ], + [ + 461, + 195 + ], + [ + 503, + 210 + ], + [ + 477, + 172 + ] + ], + [ + [ + 468, + 194 + ], + [ + 472, + 172 + ], + [ + 475, + 219 + ], + [ + 462, + 172 + ], + [ + 472, + 239 + ], + [ + 459, + 193 + ] + ], + [ + [ + 498, + 196 + ], + [ + 532, + 193 + ], + [ + 491, + 208 + ], + [ + 506, + 182 + ], + [ + 460, + 210 + ], + [ + 470, + 185 + ] + ], + [ + [ + 456, + 179 + ], + [ + 423, + 188 + ], + [ + 463, + 207 + ], + [ + 449, + 152 + ], + [ + 499, + 198 + ], + [ + 482, + 143 + ] + ] + ] + }, + { + "task_id": "0005131", + "id": "RH20T_cfg4_task_0111_user_0007_scene_0003_cfg_0004_036422060909#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the object with the corresponding shape into the corresponding hole'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005131/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 213, + 122 + ], + [ + 206, + 100 + ], + [ + 226, + 128 + ], + [ + 202, + 117 + ], + [ + 232, + 150 + ], + [ + 209, + 138 + ] + ], + [ + [ + 229, + 126 + ], + [ + 228, + 108 + ], + [ + 250, + 116 + ], + [ + 207, + 138 + ], + [ + 251, + 132 + ], + [ + 210, + 154 + ] + ], + [ + [ + 236, + 96 + ], + [ + 252, + 79 + ], + [ + 232, + 80 + ], + [ + 240, + 113 + ], + [ + 218, + 96 + ], + [ + 225, + 129 + ] + ], + [ + [ + 208, + 140 + ], + [ + 228, + 156 + ], + [ + 226, + 123 + ], + [ + 191, + 158 + ], + [ + 207, + 107 + ], + [ + 171, + 141 + ] + ] + ] + }, + { + "task_id": "0005133", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0001_cfg_0003_104122062295#148", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005133/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 155, + 86 + ], + [ + 156, + 73 + ], + [ + 192, + 89 + ], + [ + 116, + 83 + ], + [ + 188, + 100 + ], + [ + 117, + 94 + ] + ], + [ + [ + 122, + 75 + ], + [ + 137, + 43 + ], + [ + 136, + 75 + ], + [ + 110, + 74 + ], + [ + 121, + 110 + ], + [ + 96, + 105 + ] + ], + [ + [ + 131, + 119 + ], + [ + 166, + 114 + ], + [ + 132, + 151 + ], + [ + 130, + 85 + ], + [ + 95, + 158 + ], + [ + 89, + 89 + ] + ], + [ + [ + 148, + 102 + ], + [ + 149, + 67 + ], + [ + 133, + 86 + ], + [ + 164, + 118 + ], + [ + 132, + 117 + ], + [ + 162, + 151 + ] + ] + ] + }, + { + "task_id": "0005135", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0001_cfg_0003_104122062295#263", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005135/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 286 + ], + [ + 351, + 246 + ], + [ + 302, + 257 + ], + [ + 378, + 319 + ], + [ + 292, + 293 + ], + [ + 363, + 356 + ] + ], + [ + [ + 379, + 269 + ], + [ + 326, + 259 + ], + [ + 378, + 222 + ], + [ + 380, + 324 + ], + [ + 426, + 231 + ], + [ + 435, + 333 + ] + ], + [ + [ + 388, + 317 + ], + [ + 372, + 351 + ], + [ + 363, + 342 + ], + [ + 417, + 290 + ], + [ + 377, + 313 + ], + [ + 429, + 263 + ] + ], + [ + [ + 343, + 257 + ], + [ + 361, + 294 + ], + [ + 306, + 282 + ], + [ + 382, + 231 + ], + [ + 294, + 248 + ], + [ + 364, + 200 + ] + ] + ] + }, + { + "task_id": "0005138", + "id": "RH20T_cfg6_task_0062_user_0014_scene_0003_cfg_0006_104122060811#106", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the toilet paper on its holder'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005138/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 404, + 120 + ], + [ + 411, + 112 + ], + [ + 414, + 118 + ], + [ + 392, + 121 + ], + [ + 407, + 125 + ], + [ + 385, + 128 + ] + ], + [ + [ + 400, + 129 + ], + [ + 403, + 143 + ], + [ + 397, + 127 + ], + [ + 404, + 132 + ], + [ + 394, + 113 + ], + [ + 401, + 118 + ] + ], + [ + [ + 403, + 114 + ], + [ + 412, + 125 + ], + [ + 414, + 104 + ], + [ + 392, + 123 + ], + [ + 405, + 94 + ], + [ + 383, + 112 + ] + ], + [ + [ + 390, + 115 + ], + [ + 397, + 125 + ], + [ + 381, + 113 + ], + [ + 400, + 117 + ], + [ + 374, + 103 + ], + [ + 394, + 107 + ] + ] + ] + }, + { + "task_id": "0005141", + "id": "RH20T_cfg7_task_0006_user_0020_scene_0006_cfg_0007_104122063633#167", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Use the gripper to push the small block from left to right'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005141/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 176, + 145 + ], + [ + 192, + 143 + ], + [ + 182, + 164 + ], + [ + 170, + 125 + ], + [ + 166, + 167 + ], + [ + 153, + 127 + ] + ], + [ + [ + 185, + 144 + ], + [ + 178, + 160 + ], + [ + 180, + 152 + ], + [ + 189, + 136 + ], + [ + 187, + 134 + ], + [ + 196, + 119 + ] + ], + [ + [ + 204, + 149 + ], + [ + 188, + 161 + ], + [ + 207, + 145 + ], + [ + 202, + 153 + ], + [ + 222, + 134 + ], + [ + 218, + 140 + ] + ], + [ + [ + 196, + 137 + ], + [ + 200, + 124 + ], + [ + 198, + 152 + ], + [ + 194, + 121 + ], + [ + 194, + 165 + ], + [ + 189, + 134 + ] + ] + ] + }, + { + "task_id": "0005145", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_037522061512#316", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005145/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 280, + 97 + ], + [ + 285, + 77 + ], + [ + 263, + 102 + ], + [ + 297, + 92 + ], + [ + 257, + 122 + ], + [ + 291, + 113 + ] + ], + [ + [ + 284, + 87 + ], + [ + 297, + 71 + ], + [ + 279, + 68 + ], + [ + 290, + 106 + ], + [ + 267, + 82 + ], + [ + 277, + 120 + ] + ], + [ + [ + 290, + 116 + ], + [ + 301, + 101 + ], + [ + 294, + 134 + ], + [ + 285, + 97 + ], + [ + 283, + 150 + ], + [ + 273, + 113 + ] + ], + [ + [ + 290, + 90 + ], + [ + 290, + 70 + ], + [ + 315, + 88 + ], + [ + 266, + 91 + ], + [ + 314, + 107 + ], + [ + 267, + 109 + ] + ] + ] + }, + { + "task_id": "0005148", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_037522061512#658", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005148/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 100 + ], + [ + 278, + 87 + ], + [ + 281, + 120 + ], + [ + 250, + 79 + ], + [ + 268, + 130 + ], + [ + 239, + 92 + ] + ], + [ + [ + 284, + 120 + ], + [ + 289, + 96 + ], + [ + 294, + 119 + ], + [ + 273, + 121 + ], + [ + 290, + 142 + ], + [ + 268, + 146 + ] + ], + [ + [ + 275, + 94 + ], + [ + 297, + 89 + ], + [ + 268, + 79 + ], + [ + 282, + 110 + ], + [ + 245, + 84 + ], + [ + 258, + 115 + ] + ], + [ + [ + 281, + 97 + ], + [ + 303, + 102 + ], + [ + 272, + 87 + ], + [ + 289, + 105 + ], + [ + 250, + 82 + ], + [ + 269, + 100 + ] + ] + ] + }, + { + "task_id": "0005149", + "id": "RH20T_cfg2_task_0010_user_0003_scene_0002_cfg_0002_104122061850#381", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005149/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 440, + 157 + ], + [ + 457, + 147 + ], + [ + 455, + 153 + ], + [ + 426, + 160 + ], + [ + 436, + 164 + ], + [ + 408, + 169 + ] + ], + [ + [ + 439, + 118 + ], + [ + 442, + 112 + ], + [ + 413, + 118 + ], + [ + 465, + 118 + ], + [ + 411, + 123 + ], + [ + 461, + 124 + ] + ], + [ + [ + 417, + 142 + ], + [ + 412, + 127 + ], + [ + 409, + 161 + ], + [ + 425, + 123 + ], + [ + 414, + 175 + ], + [ + 429, + 138 + ] + ], + [ + [ + 413, + 161 + ], + [ + 432, + 152 + ], + [ + 421, + 182 + ], + [ + 405, + 139 + ], + [ + 401, + 192 + ], + [ + 384, + 148 + ] + ] + ] + }, + { + "task_id": "0005150", + "id": "RH20T_cfg2_task_0010_user_0003_scene_0002_cfg_0002_104122061850#526", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the handset of the telephone on the corresponding phone cradle'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005150/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 413, + 134 + ], + [ + 416, + 111 + ], + [ + 415, + 139 + ], + [ + 410, + 129 + ], + [ + 412, + 161 + ], + [ + 406, + 154 + ] + ], + [ + [ + 418, + 138 + ], + [ + 393, + 138 + ], + [ + 425, + 126 + ], + [ + 411, + 149 + ], + [ + 451, + 125 + ], + [ + 435, + 149 + ] + ], + [ + [ + 424, + 140 + ], + [ + 422, + 124 + ], + [ + 414, + 159 + ], + [ + 434, + 123 + ], + [ + 416, + 173 + ], + [ + 435, + 138 + ] + ], + [ + [ + 407, + 129 + ], + [ + 426, + 127 + ], + [ + 391, + 128 + ], + [ + 423, + 130 + ], + [ + 374, + 130 + ], + [ + 406, + 132 + ] + ] + ] + }, + { + "task_id": "0005151", + "id": "RH20T_cfg1_task_0064_user_0015_scene_0010_cfg_0001_039422060546#120", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005151/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 363, + 110 + ], + [ + 395, + 120 + ], + [ + 374, + 88 + ], + [ + 354, + 130 + ], + [ + 340, + 77 + ], + [ + 324, + 120 + ] + ], + [ + [ + 348, + 103 + ], + [ + 351, + 88 + ], + [ + 360, + 131 + ], + [ + 336, + 77 + ], + [ + 356, + 142 + ], + [ + 334, + 91 + ] + ], + [ + [ + 384, + 99 + ], + [ + 379, + 131 + ], + [ + 355, + 100 + ], + [ + 415, + 99 + ], + [ + 360, + 70 + ], + [ + 420, + 67 + ] + ], + [ + [ + 320, + 101 + ], + [ + 347, + 90 + ], + [ + 306, + 80 + ], + [ + 336, + 123 + ], + [ + 279, + 91 + ], + [ + 308, + 135 + ] + ] + ] + }, + { + "task_id": "0005153", + "id": "RH20T_cfg6_task_0104_user_0020_scene_0006_cfg_0006_104122060811#203", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate back after holding it down'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005153/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 303, + 155 + ], + [ + 316, + 149 + ], + [ + 308, + 166 + ], + [ + 298, + 144 + ], + [ + 295, + 172 + ], + [ + 284, + 151 + ] + ], + [ + [ + 307, + 172 + ], + [ + 304, + 163 + ], + [ + 305, + 161 + ], + [ + 310, + 183 + ], + [ + 309, + 170 + ], + [ + 314, + 192 + ] + ], + [ + [ + 313, + 154 + ], + [ + 305, + 167 + ], + [ + 300, + 146 + ], + [ + 326, + 162 + ], + [ + 308, + 134 + ], + [ + 334, + 150 + ] + ], + [ + [ + 318, + 160 + ], + [ + 328, + 153 + ], + [ + 320, + 172 + ], + [ + 317, + 149 + ], + [ + 310, + 179 + ], + [ + 307, + 156 + ] + ] + ] + }, + { + "task_id": "0005154", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_035622060973#91", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005154/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 178, + 107 + ], + [ + 186, + 91 + ], + [ + 163, + 96 + ], + [ + 193, + 118 + ], + [ + 157, + 112 + ], + [ + 186, + 134 + ] + ], + [ + [ + 194, + 110 + ], + [ + 207, + 118 + ], + [ + 210, + 110 + ], + [ + 179, + 109 + ], + [ + 196, + 101 + ], + [ + 165, + 99 + ] + ], + [ + [ + 168, + 100 + ], + [ + 170, + 111 + ], + [ + 180, + 113 + ], + [ + 157, + 87 + ], + [ + 178, + 102 + ], + [ + 154, + 75 + ] + ], + [ + [ + 181, + 106 + ], + [ + 178, + 94 + ], + [ + 168, + 95 + ], + [ + 195, + 118 + ], + [ + 172, + 107 + ], + [ + 198, + 129 + ] + ] + ] + }, + { + "task_id": "0005156", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_035622060973#305", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005156/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 193, + 146 + ], + [ + 185, + 128 + ], + [ + 208, + 143 + ], + [ + 178, + 150 + ], + [ + 216, + 162 + ], + [ + 186, + 167 + ] + ], + [ + [ + 204, + 127 + ], + [ + 184, + 130 + ], + [ + 210, + 146 + ], + [ + 199, + 108 + ], + [ + 230, + 143 + ], + [ + 218, + 105 + ] + ], + [ + [ + 178, + 154 + ], + [ + 178, + 164 + ], + [ + 186, + 138 + ], + [ + 169, + 169 + ], + [ + 186, + 127 + ], + [ + 168, + 158 + ] + ], + [ + [ + 193, + 146 + ], + [ + 190, + 135 + ], + [ + 187, + 131 + ], + [ + 199, + 163 + ], + [ + 189, + 141 + ], + [ + 202, + 173 + ] + ] + ] + }, + { + "task_id": "0005157", + "id": "RH20T_cfg1_task_0066_user_0014_scene_0008_cfg_0001_035622060973#414", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005157/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 215, + 165 + ], + [ + 202, + 181 + ], + [ + 233, + 177 + ], + [ + 197, + 153 + ], + [ + 246, + 161 + ], + [ + 210, + 136 + ] + ], + [ + [ + 198, + 176 + ], + [ + 177, + 182 + ], + [ + 195, + 163 + ], + [ + 200, + 190 + ], + [ + 215, + 158 + ], + [ + 221, + 185 + ] + ], + [ + [ + 201, + 162 + ], + [ + 198, + 152 + ], + [ + 194, + 146 + ], + [ + 208, + 179 + ], + [ + 196, + 157 + ], + [ + 210, + 189 + ] + ], + [ + [ + 196, + 183 + ], + [ + 212, + 190 + ], + [ + 187, + 202 + ], + [ + 205, + 164 + ], + [ + 170, + 195 + ], + [ + 189, + 156 + ] + ] + ] + }, + { + "task_id": "0005159", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0006_cfg_0001_819112071037#157", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005159/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 241, + 153 + ], + [ + 264, + 124 + ], + [ + 214, + 91 + ], + [ + 265, + 207 + ], + [ + 179, + 122 + ], + [ + 241, + 248 + ] + ], + [ + [ + 273, + 153 + ], + [ + 307, + 82 + ], + [ + 198, + 122 + ], + [ + 342, + 182 + ], + [ + 171, + 190 + ], + [ + 310, + 242 + ] + ], + [ + [ + 221, + 67 + ], + [ + 260, + 128 + ], + [ + 128, + 85 + ], + [ + 298, + 52 + ], + [ + 101, + 27 + ], + [ + 262, + 6 + ] + ], + [ + [ + 324, + 147 + ], + [ + 314, + 115 + ], + [ + 305, + 208 + ], + [ + 347, + 76 + ], + [ + 315, + 256 + ], + [ + 366, + 110 + ] + ] + ] + }, + { + "task_id": "0005160", + "id": "RH20T_cfg1_task_0061_user_0001_scene_0006_cfg_0001_819112071037#342", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005160/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 328, + 168 + ], + [ + 332, + 101 + ], + [ + 285, + 189 + ], + [ + 362, + 151 + ], + [ + 283, + 254 + ], + [ + 356, + 205 + ] + ], + [ + [ + 321, + 179 + ], + [ + 367, + 208 + ], + [ + 349, + 188 + ], + [ + 286, + 167 + ], + [ + 311, + 164 + ], + [ + 244, + 139 + ] + ], + [ + [ + 358, + 129 + ], + [ + 301, + 112 + ], + [ + 307, + 178 + ], + [ + 402, + 88 + ], + [ + 378, + 208 + ], + [ + 470, + 104 + ] + ], + [ + [ + 344, + 195 + ], + [ + 310, + 265 + ], + [ + 276, + 158 + ], + [ + 410, + 231 + ], + [ + 310, + 96 + ], + [ + 436, + 168 + ] + ] + ] + }, + { + "task_id": "0005169", + "id": "RH20T_cfg3_task_0008_user_0016_scene_0005_cfg_0003_104122063550#200", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grab the block and place it at the designated location'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005169/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 356, + 302 + ], + [ + 385, + 297 + ], + [ + 353, + 282 + ], + [ + 359, + 324 + ], + [ + 326, + 286 + ], + [ + 329, + 328 + ] + ], + [ + [ + 409, + 310 + ], + [ + 407, + 287 + ], + [ + 390, + 321 + ], + [ + 430, + 299 + ], + [ + 392, + 346 + ], + [ + 433, + 325 + ] + ], + [ + [ + 384, + 323 + ], + [ + 361, + 333 + ], + [ + 369, + 309 + ], + [ + 401, + 338 + ], + [ + 391, + 299 + ], + [ + 425, + 327 + ] + ], + [ + [ + 389, + 324 + ], + [ + 392, + 319 + ], + [ + 396, + 354 + ], + [ + 382, + 295 + ], + [ + 394, + 357 + ], + [ + 381, + 301 + ] + ] + ] + }, + { + "task_id": "0005171", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0001_cfg_0007_104122064161#172", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005171/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 389, + 197 + ], + [ + 385, + 222 + ], + [ + 384, + 194 + ], + [ + 394, + 200 + ], + [ + 388, + 170 + ], + [ + 398, + 174 + ] + ], + [ + [ + 377, + 203 + ], + [ + 363, + 198 + ], + [ + 392, + 216 + ], + [ + 362, + 189 + ], + [ + 408, + 221 + ], + [ + 377, + 193 + ] + ], + [ + [ + 372, + 208 + ], + [ + 376, + 190 + ], + [ + 388, + 198 + ], + [ + 355, + 218 + ], + [ + 384, + 215 + ], + [ + 352, + 235 + ] + ], + [ + [ + 395, + 216 + ], + [ + 388, + 190 + ], + [ + 411, + 209 + ], + [ + 382, + 222 + ], + [ + 419, + 236 + ], + [ + 389, + 247 + ] + ] + ] + }, + { + "task_id": "0005172", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0007_cfg_0004_038522062288#223", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005172/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 185, + 300 + ], + [ + 201, + 321 + ], + [ + 186, + 271 + ], + [ + 184, + 328 + ], + [ + 170, + 253 + ], + [ + 170, + 309 + ] + ], + [ + [ + 244, + 297 + ], + [ + 222, + 327 + ], + [ + 215, + 276 + ], + [ + 273, + 318 + ], + [ + 237, + 247 + ], + [ + 294, + 289 + ] + ], + [ + [ + 225, + 294 + ], + [ + 220, + 281 + ], + [ + 207, + 324 + ], + [ + 242, + 266 + ], + [ + 212, + 334 + ], + [ + 245, + 278 + ] + ], + [ + [ + 244, + 283 + ], + [ + 223, + 289 + ], + [ + 226, + 304 + ], + [ + 260, + 264 + ], + [ + 250, + 297 + ], + [ + 282, + 256 + ] + ] + ] + }, + { + "task_id": "0005174", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0007_cfg_0004_038522062288#1023", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005174/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 215, + 195 + ], + [ + 211, + 178 + ], + [ + 191, + 221 + ], + [ + 237, + 170 + ], + [ + 195, + 235 + ], + [ + 238, + 186 + ] + ], + [ + [ + 222, + 223 + ], + [ + 242, + 195 + ], + [ + 239, + 224 + ], + [ + 203, + 222 + ], + [ + 221, + 250 + ], + [ + 182, + 252 + ] + ], + [ + [ + 208, + 165 + ], + [ + 220, + 144 + ], + [ + 241, + 172 + ], + [ + 172, + 158 + ], + [ + 229, + 190 + ], + [ + 163, + 178 + ] + ], + [ + [ + 242, + 213 + ], + [ + 235, + 187 + ], + [ + 216, + 230 + ], + [ + 268, + 196 + ], + [ + 222, + 258 + ], + [ + 277, + 224 + ] + ] + ] + }, + { + "task_id": "0005176", + "id": "RH20T_cfg3_task_0091_user_0010_scene_0006_cfg_0003_104122062295#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005176/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 247, + 115 + ], + [ + 230, + 129 + ], + [ + 273, + 128 + ], + [ + 221, + 102 + ], + [ + 288, + 115 + ], + [ + 238, + 90 + ] + ], + [ + [ + 276, + 118 + ], + [ + 282, + 95 + ], + [ + 254, + 101 + ], + [ + 299, + 135 + ], + [ + 250, + 122 + ], + [ + 292, + 156 + ] + ], + [ + [ + 317, + 86 + ], + [ + 328, + 73 + ], + [ + 332, + 62 + ], + [ + 303, + 107 + ], + [ + 320, + 75 + ], + [ + 290, + 121 + ] + ], + [ + [ + 281, + 98 + ], + [ + 287, + 71 + ], + [ + 253, + 86 + ], + [ + 310, + 111 + ], + [ + 249, + 112 + ], + [ + 303, + 135 + ] + ] + ] + }, + { + "task_id": "0005177", + "id": "RH20T_cfg3_task_0091_user_0010_scene_0006_cfg_0003_104122062295#233", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005177/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 259, + 100 + ], + [ + 258, + 78 + ], + [ + 234, + 91 + ], + [ + 285, + 110 + ], + [ + 236, + 111 + ], + [ + 285, + 130 + ] + ], + [ + [ + 249, + 101 + ], + [ + 225, + 109 + ], + [ + 266, + 121 + ], + [ + 233, + 82 + ], + [ + 289, + 113 + ], + [ + 255, + 76 + ] + ], + [ + [ + 265, + 71 + ], + [ + 293, + 80 + ], + [ + 262, + 77 + ], + [ + 268, + 66 + ], + [ + 232, + 68 + ], + [ + 241, + 57 + ] + ], + [ + [ + 273, + 108 + ], + [ + 248, + 122 + ], + [ + 256, + 84 + ], + [ + 291, + 131 + ], + [ + 281, + 71 + ], + [ + 315, + 117 + ] + ] + ] + }, + { + "task_id": "0005178", + "id": "RH20T_cfg3_task_0091_user_0010_scene_0006_cfg_0003_104122062295#321", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Move an object from one box to another'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005178/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 251, + 114 + ], + [ + 261, + 141 + ], + [ + 243, + 105 + ], + [ + 257, + 122 + ], + [ + 233, + 77 + ], + [ + 248, + 97 + ] + ], + [ + [ + 289, + 89 + ], + [ + 304, + 75 + ], + [ + 299, + 71 + ], + [ + 280, + 106 + ], + [ + 282, + 87 + ], + [ + 263, + 122 + ] + ], + [ + [ + 283, + 118 + ], + [ + 286, + 123 + ], + [ + 267, + 95 + ], + [ + 298, + 139 + ], + [ + 264, + 91 + ], + [ + 294, + 133 + ] + ], + [ + [ + 280, + 109 + ], + [ + 278, + 87 + ], + [ + 256, + 100 + ], + [ + 306, + 118 + ], + [ + 259, + 120 + ], + [ + 307, + 138 + ] + ] + ] + }, + { + "task_id": "0005179", + "id": "RH20T_cfg4_task_0011_user_0016_scene_0006_cfg_0004_104122062295#140", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005179/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 212, + 111 + ], + [ + 213, + 126 + ], + [ + 247, + 85 + ], + [ + 175, + 138 + ], + [ + 250, + 65 + ], + [ + 170, + 123 + ] + ], + [ + [ + 244, + 99 + ], + [ + 251, + 52 + ], + [ + 193, + 96 + ], + [ + 290, + 102 + ], + [ + 189, + 139 + ], + [ + 281, + 141 + ] + ], + [ + [ + 226, + 109 + ], + [ + 183, + 81 + ], + [ + 194, + 126 + ], + [ + 251, + 94 + ], + [ + 242, + 158 + ], + [ + 291, + 120 + ] + ], + [ + [ + 309, + 92 + ], + [ + 331, + 45 + ], + [ + 334, + 126 + ], + [ + 287, + 62 + ], + [ + 312, + 167 + ], + [ + 270, + 102 + ] + ] + ] + }, + { + "task_id": "0005180", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0002_cfg_0001_038522063145#124", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005180/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 340, + 126 + ], + [ + 343, + 109 + ], + [ + 355, + 139 + ], + [ + 325, + 113 + ], + [ + 351, + 154 + ], + [ + 323, + 129 + ] + ], + [ + [ + 338, + 132 + ], + [ + 322, + 118 + ], + [ + 354, + 118 + ], + [ + 324, + 144 + ], + [ + 370, + 132 + ], + [ + 340, + 157 + ] + ], + [ + [ + 323, + 141 + ], + [ + 319, + 154 + ], + [ + 304, + 132 + ], + [ + 343, + 149 + ], + [ + 308, + 117 + ], + [ + 348, + 135 + ] + ], + [ + [ + 341, + 123 + ], + [ + 344, + 129 + ], + [ + 357, + 136 + ], + [ + 325, + 109 + ], + [ + 353, + 130 + ], + [ + 322, + 104 + ] + ] + ] + }, + { + "task_id": "0005181", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0002_cfg_0001_038522063145#323", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005181/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 447, + 127 + ], + [ + 457, + 109 + ], + [ + 465, + 140 + ], + [ + 429, + 114 + ], + [ + 455, + 157 + ], + [ + 421, + 131 + ] + ], + [ + [ + 437, + 111 + ], + [ + 448, + 96 + ], + [ + 456, + 124 + ], + [ + 418, + 98 + ], + [ + 446, + 140 + ], + [ + 406, + 114 + ] + ], + [ + [ + 462, + 130 + ], + [ + 439, + 128 + ], + [ + 461, + 149 + ], + [ + 464, + 110 + ], + [ + 484, + 151 + ], + [ + 489, + 112 + ] + ], + [ + [ + 449, + 128 + ], + [ + 435, + 136 + ], + [ + 447, + 108 + ], + [ + 451, + 148 + ], + [ + 460, + 100 + ], + [ + 464, + 140 + ] + ] + ] + }, + { + "task_id": "0005182", + "id": "RH20T_cfg1_task_0037_user_0015_scene_0002_cfg_0001_038522063145#538", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005182/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 100 + ], + [ + 411, + 82 + ], + [ + 426, + 108 + ], + [ + 387, + 93 + ], + [ + 420, + 125 + ], + [ + 383, + 109 + ] + ], + [ + [ + 427, + 105 + ], + [ + 446, + 103 + ], + [ + 415, + 110 + ], + [ + 439, + 101 + ], + [ + 397, + 111 + ], + [ + 421, + 103 + ] + ], + [ + [ + 420, + 76 + ], + [ + 427, + 75 + ], + [ + 404, + 61 + ], + [ + 435, + 90 + ], + [ + 398, + 62 + ], + [ + 428, + 90 + ] + ], + [ + [ + 392, + 101 + ], + [ + 384, + 114 + ], + [ + 388, + 83 + ], + [ + 395, + 119 + ], + [ + 395, + 71 + ], + [ + 402, + 106 + ] + ] + ] + }, + { + "task_id": "0005185", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0007_cfg_0004_036422060909#247", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005185/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 313, + 49 + ], + [ + 311, + 56 + ], + [ + 295, + 37 + ], + [ + 331, + 61 + ], + [ + 296, + 29 + ], + [ + 334, + 55 + ] + ], + [ + [ + 319, + 51 + ], + [ + 327, + 35 + ], + [ + 332, + 40 + ], + [ + 306, + 62 + ], + [ + 325, + 55 + ], + [ + 300, + 76 + ] + ], + [ + [ + 318, + 33 + ], + [ + 334, + 17 + ], + [ + 317, + 34 + ], + [ + 320, + 33 + ], + [ + 301, + 51 + ], + [ + 305, + 49 + ] + ], + [ + [ + 311, + 40 + ], + [ + 332, + 44 + ], + [ + 312, + 19 + ], + [ + 310, + 61 + ], + [ + 291, + 16 + ], + [ + 290, + 57 + ] + ] + ] + }, + { + "task_id": "0005187", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0007_cfg_0004_036422060909#441", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005187/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 272, + 84 + ], + [ + 273, + 105 + ], + [ + 252, + 79 + ], + [ + 292, + 89 + ], + [ + 250, + 57 + ], + [ + 292, + 66 + ] + ], + [ + [ + 293, + 60 + ], + [ + 312, + 66 + ], + [ + 283, + 44 + ], + [ + 303, + 74 + ], + [ + 264, + 39 + ], + [ + 285, + 68 + ] + ], + [ + [ + 283, + 72 + ], + [ + 288, + 54 + ], + [ + 304, + 69 + ], + [ + 262, + 76 + ], + [ + 299, + 86 + ], + [ + 258, + 93 + ] + ], + [ + [ + 264, + 67 + ], + [ + 274, + 83 + ], + [ + 245, + 80 + ], + [ + 283, + 53 + ], + [ + 236, + 65 + ], + [ + 273, + 38 + ] + ] + ] + }, + { + "task_id": "0005188", + "id": "RH20T_cfg4_task_0066_user_0016_scene_0007_cfg_0004_036422060909#525", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005188/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 263, + 70 + ], + [ + 266, + 50 + ], + [ + 285, + 67 + ], + [ + 240, + 72 + ], + [ + 281, + 85 + ], + [ + 238, + 90 + ] + ], + [ + [ + 275, + 80 + ], + [ + 259, + 89 + ], + [ + 279, + 65 + ], + [ + 270, + 95 + ], + [ + 296, + 55 + ], + [ + 288, + 85 + ] + ], + [ + [ + 254, + 76 + ], + [ + 265, + 63 + ], + [ + 249, + 92 + ], + [ + 259, + 59 + ], + [ + 237, + 106 + ], + [ + 247, + 73 + ] + ], + [ + [ + 293, + 70 + ], + [ + 274, + 81 + ], + [ + 279, + 57 + ], + [ + 306, + 81 + ], + [ + 299, + 44 + ], + [ + 326, + 69 + ] + ] + ] + }, + { + "task_id": "0005194", + "id": "RH20T_cfg1_task_0066_user_0015_scene_0006_cfg_0001_039422060546#906", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with large Megabloks for the nth time with n blocks'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005194/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 346, + 19 + ], + [ + 324, + 19 + ], + [ + 346, + 45 + ], + [ + 346, + 0 + ], + [ + 370, + 45 + ], + [ + 370, + 0 + ] + ], + [ + [ + 331, + 43 + ], + [ + 330, + 22 + ], + [ + 322, + 59 + ], + [ + 339, + 28 + ], + [ + 324, + 78 + ], + [ + 339, + 47 + ] + ], + [ + [ + 315, + 43 + ], + [ + 322, + 54 + ], + [ + 306, + 24 + ], + [ + 324, + 63 + ], + [ + 298, + 12 + ], + [ + 316, + 53 + ] + ], + [ + [ + 343, + 40 + ], + [ + 354, + 61 + ], + [ + 326, + 56 + ], + [ + 362, + 23 + ], + [ + 316, + 37 + ], + [ + 351, + 2 + ] + ] + ] + }, + { + "task_id": "0005201", + "id": "RH20T_cfg4_task_0035_user_0017_scene_0005_cfg_0004_104422070011#113", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up one small block'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005201/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 324, + 127 + ], + [ + 300, + 120 + ], + [ + 315, + 139 + ], + [ + 335, + 113 + ], + [ + 338, + 145 + ], + [ + 359, + 120 + ] + ], + [ + [ + 317, + 139 + ], + [ + 328, + 127 + ], + [ + 303, + 155 + ], + [ + 331, + 124 + ], + [ + 293, + 166 + ], + [ + 320, + 135 + ] + ], + [ + [ + 300, + 156 + ], + [ + 284, + 168 + ], + [ + 287, + 134 + ], + [ + 312, + 178 + ], + [ + 302, + 123 + ], + [ + 327, + 166 + ] + ], + [ + [ + 302, + 143 + ], + [ + 313, + 129 + ], + [ + 295, + 164 + ], + [ + 309, + 124 + ], + [ + 286, + 177 + ], + [ + 300, + 138 + ] + ] + ] + }, + { + "task_id": "0005203", + "id": "RH20T_cfg5_task_0011_user_0007_scene_0004_cfg_0005_104422070044#158", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005203/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 372, + 115 + ], + [ + 385, + 116 + ], + [ + 362, + 129 + ], + [ + 381, + 102 + ], + [ + 350, + 129 + ], + [ + 368, + 102 + ] + ], + [ + [ + 373, + 133 + ], + [ + 371, + 121 + ], + [ + 371, + 147 + ], + [ + 375, + 120 + ], + [ + 373, + 159 + ], + [ + 377, + 132 + ] + ], + [ + [ + 370, + 131 + ], + [ + 389, + 133 + ], + [ + 367, + 127 + ], + [ + 371, + 135 + ], + [ + 348, + 125 + ], + [ + 353, + 133 + ] + ], + [ + [ + 380, + 140 + ], + [ + 398, + 141 + ], + [ + 377, + 128 + ], + [ + 383, + 151 + ], + [ + 359, + 127 + ], + [ + 366, + 151 + ] + ] + ] + }, + { + "task_id": "0005206", + "id": "RH20T_cfg1_task_0059_user_0001_scene_0001_cfg_0001_750612070851#179", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005206/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 250, + 170 + ], + [ + 250, + 145 + ], + [ + 242, + 197 + ], + [ + 257, + 145 + ], + [ + 242, + 219 + ], + [ + 256, + 167 + ] + ], + [ + [ + 248, + 175 + ], + [ + 229, + 169 + ], + [ + 276, + 159 + ], + [ + 217, + 193 + ], + [ + 298, + 165 + ], + [ + 237, + 203 + ] + ], + [ + [ + 254, + 159 + ], + [ + 286, + 168 + ], + [ + 267, + 145 + ], + [ + 240, + 175 + ], + [ + 238, + 137 + ], + [ + 207, + 166 + ] + ], + [ + [ + 256, + 177 + ], + [ + 285, + 163 + ], + [ + 270, + 206 + ], + [ + 241, + 148 + ], + [ + 241, + 222 + ], + [ + 211, + 162 + ] + ] + ] + }, + { + "task_id": "0005208", + "id": "RH20T_cfg1_task_0059_user_0001_scene_0001_cfg_0001_750612070851#555", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005208/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 250, + 168 + ], + [ + 248, + 144 + ], + [ + 255, + 196 + ], + [ + 245, + 141 + ], + [ + 257, + 216 + ], + [ + 247, + 163 + ] + ], + [ + [ + 275, + 136 + ], + [ + 307, + 137 + ], + [ + 261, + 161 + ], + [ + 288, + 112 + ], + [ + 230, + 159 + ], + [ + 259, + 112 + ] + ], + [ + [ + 236, + 146 + ], + [ + 229, + 173 + ], + [ + 225, + 170 + ], + [ + 250, + 121 + ], + [ + 232, + 146 + ], + [ + 256, + 97 + ] + ], + [ + [ + 282, + 164 + ], + [ + 244, + 163 + ], + [ + 281, + 202 + ], + [ + 283, + 125 + ], + [ + 320, + 203 + ], + [ + 322, + 126 + ] + ] + ] + }, + { + "task_id": "0005209", + "id": "RH20T_cfg1_task_0059_user_0001_scene_0001_cfg_0001_750612070851#714", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to increase the volume of the speaker'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005209/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 242, + 173 + ], + [ + 248, + 142 + ], + [ + 261, + 181 + ], + [ + 221, + 164 + ], + [ + 256, + 212 + ], + [ + 214, + 199 + ] + ], + [ + [ + 246, + 166 + ], + [ + 206, + 167 + ], + [ + 242, + 128 + ], + [ + 249, + 204 + ], + [ + 280, + 128 + ], + [ + 287, + 201 + ] + ], + [ + [ + 253, + 168 + ], + [ + 253, + 146 + ], + [ + 250, + 198 + ], + [ + 255, + 140 + ], + [ + 250, + 216 + ], + [ + 254, + 160 + ] + ], + [ + [ + 260, + 206 + ], + [ + 293, + 213 + ], + [ + 252, + 240 + ], + [ + 267, + 172 + ], + [ + 217, + 232 + ], + [ + 234, + 164 + ] + ] + ] + }, + { + "task_id": "0005211", + "id": "RH20T_cfg1_task_0042_user_0016_scene_0007_cfg_0001_038522063145#123", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005211/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 333, + 47 + ], + [ + 354, + 49 + ], + [ + 332, + 62 + ], + [ + 334, + 31 + ], + [ + 312, + 60 + ], + [ + 311, + 29 + ] + ], + [ + [ + 362, + 33 + ], + [ + 370, + 25 + ], + [ + 383, + 35 + ], + [ + 343, + 32 + ], + [ + 375, + 44 + ], + [ + 334, + 40 + ] + ], + [ + [ + 369, + 28 + ], + [ + 368, + 43 + ], + [ + 359, + 10 + ], + [ + 378, + 45 + ], + [ + 359, + 0 + ], + [ + 378, + 31 + ] + ], + [ + [ + 358, + 38 + ], + [ + 358, + 19 + ], + [ + 375, + 47 + ], + [ + 341, + 29 + ], + [ + 374, + 64 + ], + [ + 341, + 47 + ] + ] + ] + }, + { + "task_id": "0005212", + "id": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_819112070044#81", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005212/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 334, + 228 + ], + [ + 334, + 207 + ], + [ + 339, + 238 + ], + [ + 329, + 217 + ], + [ + 339, + 260 + ], + [ + 329, + 241 + ] + ], + [ + [ + 329, + 200 + ], + [ + 312, + 195 + ], + [ + 339, + 219 + ], + [ + 320, + 183 + ], + [ + 355, + 223 + ], + [ + 336, + 188 + ] + ], + [ + [ + 314, + 213 + ], + [ + 315, + 195 + ], + [ + 295, + 202 + ], + [ + 334, + 225 + ], + [ + 295, + 219 + ], + [ + 333, + 243 + ] + ], + [ + [ + 311, + 192 + ], + [ + 305, + 201 + ], + [ + 314, + 215 + ], + [ + 308, + 170 + ], + [ + 321, + 207 + ], + [ + 314, + 160 + ] + ] + ] + }, + { + "task_id": "0005216", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0002_cfg_0006_037522061512#988", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005216/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 177, + 166 + ], + [ + 160, + 159 + ], + [ + 160, + 175 + ], + [ + 192, + 158 + ], + [ + 178, + 183 + ], + [ + 210, + 165 + ] + ], + [ + [ + 184, + 139 + ], + [ + 188, + 161 + ], + [ + 166, + 135 + ], + [ + 202, + 144 + ], + [ + 163, + 113 + ], + [ + 198, + 124 + ] + ], + [ + [ + 197, + 146 + ], + [ + 209, + 134 + ], + [ + 215, + 156 + ], + [ + 178, + 135 + ], + [ + 203, + 167 + ], + [ + 167, + 147 + ] + ], + [ + [ + 186, + 144 + ], + [ + 201, + 132 + ], + [ + 191, + 162 + ], + [ + 180, + 126 + ], + [ + 176, + 175 + ], + [ + 164, + 139 + ] + ] + ] + }, + { + "task_id": "0005217", + "id": "RH20T_cfg5_task_0043_user_0007_scene_0002_cfg_0005_105422061350#363", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005217/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 266, + 90 + ], + [ + 297, + 89 + ], + [ + 262, + 86 + ], + [ + 270, + 94 + ], + [ + 230, + 88 + ], + [ + 242, + 96 + ] + ], + [ + [ + 308, + 98 + ], + [ + 305, + 74 + ], + [ + 333, + 86 + ], + [ + 282, + 111 + ], + [ + 335, + 108 + ], + [ + 286, + 133 + ] + ], + [ + [ + 285, + 106 + ], + [ + 287, + 87 + ], + [ + 314, + 110 + ], + [ + 255, + 103 + ], + [ + 311, + 126 + ], + [ + 255, + 121 + ] + ], + [ + [ + 262, + 85 + ], + [ + 265, + 84 + ], + [ + 289, + 71 + ], + [ + 235, + 99 + ], + [ + 284, + 72 + ], + [ + 234, + 99 + ] + ] + ] + }, + { + "task_id": "0005218", + "id": "RH20T_cfg5_task_0043_user_0007_scene_0002_cfg_0005_105422061350#554", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005218/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 290, + 111 + ], + [ + 294, + 91 + ], + [ + 316, + 124 + ], + [ + 264, + 98 + ], + [ + 311, + 141 + ], + [ + 261, + 116 + ] + ], + [ + [ + 283, + 106 + ], + [ + 283, + 121 + ], + [ + 311, + 111 + ], + [ + 253, + 100 + ], + [ + 310, + 97 + ], + [ + 255, + 87 + ] + ], + [ + [ + 264, + 129 + ], + [ + 243, + 115 + ], + [ + 249, + 125 + ], + [ + 278, + 132 + ], + [ + 272, + 140 + ], + [ + 300, + 146 + ] + ], + [ + [ + 298, + 114 + ], + [ + 314, + 92 + ], + [ + 313, + 114 + ], + [ + 285, + 115 + ], + [ + 296, + 139 + ], + [ + 269, + 137 + ] + ] + ] + }, + { + "task_id": "0005221", + "id": "RH20T_cfg2_task_0011_user_0013_scene_0001_cfg_0002_104122061850#312", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Water the plant'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005221/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 448, + 31 + ], + [ + 435, + 25 + ], + [ + 469, + 20 + ], + [ + 426, + 43 + ], + [ + 484, + 27 + ], + [ + 440, + 51 + ] + ], + [ + [ + 456, + 18 + ], + [ + 440, + 42 + ], + [ + 437, + 3 + ], + [ + 475, + 33 + ], + [ + 453, + 0 + ], + [ + 491, + 9 + ] + ], + [ + [ + 429, + 33 + ], + [ + 429, + 12 + ], + [ + 417, + 20 + ], + [ + 442, + 47 + ], + [ + 417, + 39 + ], + [ + 442, + 67 + ] + ], + [ + [ + 437, + 5 + ], + [ + 453, + 0 + ], + [ + 423, + 0 + ], + [ + 450, + 26 + ], + [ + 406, + 0 + ], + [ + 434, + 41 + ] + ] + ] + }, + { + "task_id": "0005222", + "id": "RH20T_cfg6_task_0017_user_0014_scene_0002_cfg_0006_104122061602#152", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the pen into the pen holder'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005222/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 410, + 186 + ], + [ + 423, + 170 + ], + [ + 425, + 208 + ], + [ + 395, + 165 + ], + [ + 412, + 222 + ], + [ + 384, + 181 + ] + ], + [ + [ + 391, + 190 + ], + [ + 404, + 171 + ], + [ + 388, + 174 + ], + [ + 394, + 208 + ], + [ + 376, + 192 + ], + [ + 381, + 226 + ] + ], + [ + [ + 438, + 164 + ], + [ + 452, + 149 + ], + [ + 433, + 142 + ], + [ + 443, + 185 + ], + [ + 417, + 158 + ], + [ + 428, + 201 + ] + ], + [ + [ + 453, + 185 + ], + [ + 467, + 180 + ], + [ + 429, + 174 + ], + [ + 478, + 198 + ], + [ + 414, + 179 + ], + [ + 463, + 205 + ] + ] + ] + }, + { + "task_id": "0005224", + "id": "RH20T_cfg4_task_0024_user_0014_scene_0007_cfg_0004_104422070011#122", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn off the power strip by pressing the button'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005224/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 284, + 142 + ], + [ + 298, + 126 + ], + [ + 276, + 160 + ], + [ + 291, + 126 + ], + [ + 263, + 175 + ], + [ + 278, + 141 + ] + ], + [ + [ + 265, + 147 + ], + [ + 254, + 161 + ], + [ + 242, + 135 + ], + [ + 288, + 158 + ], + [ + 253, + 119 + ], + [ + 300, + 144 + ] + ], + [ + [ + 281, + 101 + ], + [ + 257, + 88 + ], + [ + 295, + 77 + ], + [ + 268, + 125 + ], + [ + 318, + 91 + ], + [ + 291, + 137 + ] + ], + [ + [ + 283, + 163 + ], + [ + 294, + 183 + ], + [ + 260, + 168 + ], + [ + 307, + 158 + ], + [ + 247, + 146 + ], + [ + 296, + 135 + ] + ] + ] + }, + { + "task_id": "0005228", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_105422061350#236", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005228/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 259, + 112 + ], + [ + 262, + 93 + ], + [ + 274, + 94 + ], + [ + 243, + 132 + ], + [ + 271, + 112 + ], + [ + 242, + 149 + ] + ], + [ + [ + 283, + 116 + ], + [ + 269, + 90 + ], + [ + 277, + 118 + ], + [ + 290, + 112 + ], + [ + 289, + 142 + ], + [ + 304, + 139 + ] + ], + [ + [ + 272, + 135 + ], + [ + 293, + 127 + ], + [ + 252, + 119 + ], + [ + 294, + 152 + ], + [ + 228, + 126 + ], + [ + 271, + 161 + ] + ], + [ + [ + 271, + 127 + ], + [ + 263, + 102 + ], + [ + 299, + 125 + ], + [ + 240, + 129 + ], + [ + 308, + 150 + ], + [ + 248, + 156 + ] + ] + ] + }, + { + "task_id": "0005229", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0010_cfg_0005_105422061350#442", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005229/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 261, + 96 + ], + [ + 269, + 77 + ], + [ + 236, + 100 + ], + [ + 287, + 92 + ], + [ + 227, + 120 + ], + [ + 279, + 113 + ] + ], + [ + [ + 258, + 130 + ], + [ + 261, + 112 + ], + [ + 276, + 114 + ], + [ + 239, + 148 + ], + [ + 272, + 130 + ], + [ + 237, + 163 + ] + ], + [ + [ + 236, + 105 + ], + [ + 259, + 121 + ], + [ + 253, + 106 + ], + [ + 218, + 103 + ], + [ + 232, + 90 + ], + [ + 195, + 86 + ] + ], + [ + [ + 231, + 114 + ], + [ + 238, + 142 + ], + [ + 241, + 103 + ], + [ + 221, + 123 + ], + [ + 234, + 73 + ], + [ + 215, + 97 + ] + ] + ] + }, + { + "task_id": "0005231", + "id": "RH20T_cfg4_task_0047_user_0014_scene_0005_cfg_0004_104122063550#152", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the cup off the cup rack'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005231/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 325, + 124 + ], + [ + 317, + 149 + ], + [ + 344, + 121 + ], + [ + 308, + 126 + ], + [ + 351, + 95 + ], + [ + 315, + 102 + ] + ], + [ + [ + 315, + 141 + ], + [ + 327, + 121 + ], + [ + 310, + 123 + ], + [ + 319, + 158 + ], + [ + 296, + 144 + ], + [ + 307, + 178 + ] + ], + [ + [ + 314, + 128 + ], + [ + 305, + 131 + ], + [ + 324, + 152 + ], + [ + 304, + 103 + ], + [ + 331, + 147 + ], + [ + 312, + 101 + ] + ], + [ + [ + 321, + 131 + ], + [ + 320, + 120 + ], + [ + 344, + 145 + ], + [ + 299, + 117 + ], + [ + 344, + 155 + ], + [ + 301, + 128 + ] + ] + ] + }, + { + "task_id": "0005232", + "id": "RH20T_cfg1_task_0045_user_0015_scene_0002_cfg_0001_038522062547#146", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Place the brush on the pen rack'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005232/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 424, + 105 + ], + [ + 429, + 92 + ], + [ + 445, + 118 + ], + [ + 403, + 93 + ], + [ + 439, + 130 + ], + [ + 400, + 106 + ] + ], + [ + [ + 444, + 79 + ], + [ + 431, + 80 + ], + [ + 453, + 102 + ], + [ + 436, + 56 + ], + [ + 465, + 101 + ], + [ + 448, + 56 + ] + ], + [ + [ + 431, + 104 + ], + [ + 418, + 119 + ], + [ + 439, + 121 + ], + [ + 422, + 86 + ], + [ + 451, + 106 + ], + [ + 435, + 70 + ] + ], + [ + [ + 435, + 93 + ], + [ + 441, + 101 + ], + [ + 422, + 114 + ], + [ + 447, + 70 + ], + [ + 417, + 106 + ], + [ + 441, + 63 + ] + ] + ] + }, + { + "task_id": "0005236", + "id": "RH20T_cfg4_task_0032_user_0016_scene_0002_cfg_0004_104122062823#211", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Slide the outer casing onto the gift box'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005236/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 22 + ], + [ + 299, + 0 + ], + [ + 340, + 2 + ], + [ + 289, + 42 + ], + [ + 353, + 24 + ], + [ + 304, + 62 + ] + ], + [ + [ + 333, + 35 + ], + [ + 306, + 23 + ], + [ + 320, + 64 + ], + [ + 346, + 6 + ], + [ + 347, + 77 + ], + [ + 375, + 18 + ] + ], + [ + [ + 319, + 30 + ], + [ + 336, + 55 + ], + [ + 295, + 48 + ], + [ + 342, + 12 + ], + [ + 277, + 22 + ], + [ + 326, + 0 + ] + ], + [ + [ + 298, + 51 + ], + [ + 293, + 18 + ], + [ + 264, + 54 + ], + [ + 331, + 48 + ], + [ + 268, + 88 + ], + [ + 335, + 80 + ] + ] + ] + }, + { + "task_id": "0005240", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0003_cfg_0006_104122061330#183", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005240/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 327, + 77 + ], + [ + 350, + 76 + ], + [ + 313, + 93 + ], + [ + 340, + 61 + ], + [ + 292, + 93 + ], + [ + 319, + 62 + ] + ], + [ + [ + 294, + 64 + ], + [ + 293, + 64 + ], + [ + 266, + 63 + ], + [ + 321, + 66 + ], + [ + 265, + 64 + ], + [ + 323, + 66 + ] + ], + [ + [ + 333, + 35 + ], + [ + 309, + 32 + ], + [ + 321, + 54 + ], + [ + 344, + 16 + ], + [ + 347, + 58 + ], + [ + 369, + 19 + ] + ], + [ + [ + 323, + 57 + ], + [ + 325, + 30 + ], + [ + 307, + 68 + ], + [ + 337, + 47 + ], + [ + 305, + 93 + ], + [ + 334, + 72 + ] + ] + ] + }, + { + "task_id": "0005241", + "id": "RH20T_cfg6_task_0073_user_0020_scene_0003_cfg_0006_104122061330#545", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005241/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 268, + 228 + ], + [ + 244, + 218 + ], + [ + 284, + 213 + ], + [ + 251, + 243 + ], + [ + 308, + 223 + ], + [ + 277, + 254 + ] + ], + [ + [ + 296, + 237 + ], + [ + 289, + 224 + ], + [ + 283, + 261 + ], + [ + 308, + 213 + ], + [ + 289, + 271 + ], + [ + 314, + 225 + ] + ], + [ + [ + 268, + 235 + ], + [ + 289, + 249 + ], + [ + 273, + 251 + ], + [ + 263, + 218 + ], + [ + 254, + 238 + ], + [ + 242, + 205 + ] + ], + [ + [ + 302, + 232 + ], + [ + 327, + 224 + ], + [ + 306, + 259 + ], + [ + 299, + 205 + ], + [ + 282, + 266 + ], + [ + 277, + 214 + ] + ] + ] + }, + { + "task_id": "0005243", + "id": "RH20T_cfg1_task_0004_user_0015_scene_0002_cfg_0001_750612070853#165", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005243/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 169 + ], + [ + 326, + 131 + ], + [ + 303, + 152 + ], + [ + 354, + 187 + ], + [ + 304, + 185 + ], + [ + 353, + 223 + ] + ], + [ + [ + 334, + 147 + ], + [ + 298, + 159 + ], + [ + 334, + 112 + ], + [ + 335, + 178 + ], + [ + 370, + 101 + ], + [ + 367, + 166 + ] + ], + [ + [ + 288, + 190 + ], + [ + 302, + 179 + ], + [ + 252, + 180 + ], + [ + 326, + 200 + ], + [ + 233, + 191 + ], + [ + 312, + 214 + ] + ], + [ + [ + 341, + 203 + ], + [ + 372, + 196 + ], + [ + 316, + 175 + ], + [ + 368, + 233 + ], + [ + 281, + 180 + ], + [ + 333, + 243 + ] + ] + ] + }, + { + "task_id": "0005247", + "id": "RH20T_cfg1_task_0004_user_0005_scene_0003_cfg_0001_035622060973#127", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005247/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 248, + 104 + ], + [ + 265, + 94 + ], + [ + 255, + 110 + ], + [ + 241, + 97 + ], + [ + 238, + 119 + ], + [ + 223, + 107 + ] + ], + [ + [ + 221, + 108 + ], + [ + 241, + 101 + ], + [ + 221, + 98 + ], + [ + 222, + 118 + ], + [ + 202, + 104 + ], + [ + 202, + 125 + ] + ], + [ + [ + 246, + 119 + ], + [ + 241, + 107 + ], + [ + 238, + 105 + ], + [ + 254, + 134 + ], + [ + 243, + 116 + ], + [ + 258, + 144 + ] + ], + [ + [ + 240, + 126 + ], + [ + 228, + 138 + ], + [ + 256, + 134 + ], + [ + 225, + 117 + ], + [ + 267, + 123 + ], + [ + 237, + 106 + ] + ] + ] + }, + { + "task_id": "0005250", + "id": "RH20T_cfg6_task_0058_user_0014_scene_0001_cfg_0006_104122061018#138", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005250/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 279, + 106 + ], + [ + 283, + 78 + ], + [ + 306, + 112 + ], + [ + 251, + 99 + ], + [ + 303, + 140 + ], + [ + 246, + 128 + ] + ], + [ + [ + 297, + 112 + ], + [ + 293, + 84 + ], + [ + 288, + 107 + ], + [ + 307, + 118 + ], + [ + 291, + 133 + ], + [ + 311, + 147 + ] + ], + [ + [ + 274, + 133 + ], + [ + 274, + 161 + ], + [ + 290, + 131 + ], + [ + 259, + 135 + ], + [ + 290, + 102 + ], + [ + 259, + 108 + ] + ], + [ + [ + 281, + 100 + ], + [ + 258, + 116 + ], + [ + 296, + 121 + ], + [ + 266, + 78 + ], + [ + 319, + 106 + ], + [ + 291, + 61 + ] + ] + ] + }, + { + "task_id": "0005254", + "id": "RH20T_cfg7_task_0028_user_0020_scene_0003_cfg_0007_104122061602#199", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and close the drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005254/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 444, + 141 + ], + [ + 461, + 143 + ], + [ + 443, + 164 + ], + [ + 445, + 118 + ], + [ + 427, + 161 + ], + [ + 429, + 118 + ] + ], + [ + [ + 419, + 141 + ], + [ + 436, + 138 + ], + [ + 411, + 121 + ], + [ + 428, + 162 + ], + [ + 394, + 124 + ], + [ + 411, + 166 + ] + ], + [ + [ + 435, + 161 + ], + [ + 450, + 173 + ], + [ + 417, + 170 + ], + [ + 454, + 151 + ], + [ + 401, + 159 + ], + [ + 438, + 138 + ] + ], + [ + [ + 463, + 160 + ], + [ + 479, + 160 + ], + [ + 467, + 138 + ], + [ + 458, + 182 + ], + [ + 450, + 138 + ], + [ + 441, + 182 + ] + ] + ] + }, + { + "task_id": "0005255", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0009_cfg_0004_104122062295#475", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005255/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 239, + 334 + ], + [ + 273, + 340 + ], + [ + 262, + 360 + ], + [ + 214, + 292 + ], + [ + 232, + 360 + ], + [ + 186, + 290 + ] + ], + [ + [ + 239, + 330 + ], + [ + 249, + 293 + ], + [ + 208, + 295 + ], + [ + 274, + 360 + ], + [ + 201, + 328 + ], + [ + 262, + 360 + ] + ], + [ + [ + 249, + 317 + ], + [ + 230, + 308 + ], + [ + 268, + 360 + ], + [ + 233, + 275 + ], + [ + 282, + 360 + ], + [ + 249, + 286 + ] + ], + [ + [ + 258, + 322 + ], + [ + 236, + 352 + ], + [ + 302, + 334 + ], + [ + 217, + 310 + ], + [ + 317, + 307 + ], + [ + 238, + 286 + ] + ] + ] + }, + { + "task_id": "0005256", + "id": "RH20T_cfg4_task_0065_user_0010_scene_0009_cfg_0004_104122062295#591", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Build with small Lego blocks for the nth time with n blocks'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005256/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 205, + 300 + ], + [ + 213, + 263 + ], + [ + 178, + 266 + ], + [ + 234, + 338 + ], + [ + 173, + 299 + ], + [ + 225, + 360 + ] + ], + [ + [ + 175, + 298 + ], + [ + 125, + 305 + ], + [ + 151, + 254 + ], + [ + 202, + 348 + ], + [ + 192, + 254 + ], + [ + 242, + 337 + ] + ], + [ + [ + 243, + 327 + ], + [ + 211, + 360 + ], + [ + 241, + 351 + ], + [ + 246, + 299 + ], + [ + 267, + 319 + ], + [ + 275, + 266 + ] + ], + [ + [ + 231, + 338 + ], + [ + 243, + 293 + ], + [ + 208, + 308 + ], + [ + 251, + 360 + ], + [ + 189, + 360 + ], + [ + 238, + 360 + ] + ] + ] + }, + { + "task_id": "0005264", + "id": "RH20T_cfg5_task_0225_user_0010_scene_0010_cfg_0005_104422070044#519", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put the bowl into the microwave'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005264/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 431, + 205 + ], + [ + 435, + 184 + ], + [ + 453, + 214 + ], + [ + 410, + 197 + ], + [ + 447, + 235 + ], + [ + 406, + 217 + ] + ], + [ + [ + 421, + 206 + ], + [ + 410, + 226 + ], + [ + 404, + 201 + ], + [ + 438, + 211 + ], + [ + 414, + 182 + ], + [ + 449, + 190 + ] + ], + [ + [ + 427, + 193 + ], + [ + 426, + 173 + ], + [ + 402, + 194 + ], + [ + 452, + 193 + ], + [ + 403, + 215 + ], + [ + 453, + 214 + ] + ], + [ + [ + 451, + 192 + ], + [ + 453, + 172 + ], + [ + 461, + 180 + ], + [ + 440, + 204 + ], + [ + 459, + 199 + ], + [ + 438, + 223 + ] + ] + ] + }, + { + "task_id": "0005266", + "id": "RH20T_cfg6_task_0089_user_0020_scene_0009_cfg_0006_104122063633#160", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Push an object with obstacles'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005266/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 202, + 229 + ], + [ + 227, + 224 + ], + [ + 201, + 244 + ], + [ + 203, + 216 + ], + [ + 174, + 249 + ], + [ + 179, + 221 + ] + ], + [ + [ + 220, + 243 + ], + [ + 219, + 230 + ], + [ + 222, + 219 + ], + [ + 217, + 268 + ], + [ + 223, + 232 + ], + [ + 218, + 278 + ] + ], + [ + [ + 227, + 232 + ], + [ + 245, + 241 + ], + [ + 214, + 224 + ], + [ + 240, + 240 + ], + [ + 196, + 216 + ], + [ + 223, + 231 + ] + ], + [ + [ + 220, + 233 + ], + [ + 231, + 230 + ], + [ + 203, + 251 + ], + [ + 237, + 217 + ], + [ + 194, + 253 + ], + [ + 227, + 220 + ] + ] + ] + }, + { + "task_id": "0005267", + "id": "RH20T_cfg6_task_0071_user_0020_scene_0002_cfg_0006_104122061018#195", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Close a sliding window'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005267/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 302, + 103 + ], + [ + 322, + 85 + ], + [ + 291, + 89 + ], + [ + 313, + 118 + ], + [ + 273, + 106 + ], + [ + 292, + 136 + ] + ], + [ + [ + 277, + 127 + ], + [ + 260, + 119 + ], + [ + 283, + 151 + ], + [ + 271, + 101 + ], + [ + 302, + 160 + ], + [ + 290, + 109 + ] + ], + [ + [ + 286, + 83 + ], + [ + 261, + 73 + ], + [ + 285, + 71 + ], + [ + 288, + 94 + ], + [ + 312, + 82 + ], + [ + 312, + 104 + ] + ], + [ + [ + 280, + 99 + ], + [ + 274, + 72 + ], + [ + 269, + 90 + ], + [ + 291, + 108 + ], + [ + 275, + 114 + ], + [ + 296, + 134 + ] + ] + ] + }, + { + "task_id": "0005270", + "id": "RH20T_cfg5_task_0096_user_0007_scene_0008_cfg_0005_036422060215#207", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005270/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 336, + 104 + ], + [ + 340, + 118 + ], + [ + 337, + 121 + ], + [ + 336, + 87 + ], + [ + 334, + 108 + ], + [ + 333, + 74 + ] + ], + [ + [ + 324, + 107 + ], + [ + 327, + 88 + ], + [ + 344, + 107 + ], + [ + 304, + 108 + ], + [ + 341, + 124 + ], + [ + 302, + 126 + ] + ], + [ + [ + 307, + 116 + ], + [ + 312, + 136 + ], + [ + 320, + 110 + ], + [ + 293, + 122 + ], + [ + 314, + 90 + ], + [ + 286, + 101 + ] + ], + [ + [ + 344, + 114 + ], + [ + 348, + 100 + ], + [ + 349, + 99 + ], + [ + 339, + 130 + ], + [ + 346, + 113 + ], + [ + 336, + 143 + ] + ] + ] + }, + { + "task_id": "0005272", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0002_cfg_0004_104122063550#175", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005272/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 319, + 105 + ], + [ + 321, + 87 + ], + [ + 345, + 101 + ], + [ + 292, + 108 + ], + [ + 343, + 117 + ], + [ + 292, + 124 + ] + ], + [ + [ + 339, + 77 + ], + [ + 326, + 77 + ], + [ + 330, + 103 + ], + [ + 349, + 51 + ], + [ + 342, + 101 + ], + [ + 360, + 52 + ] + ], + [ + [ + 326, + 94 + ], + [ + 339, + 117 + ], + [ + 303, + 109 + ], + [ + 350, + 78 + ], + [ + 291, + 86 + ], + [ + 337, + 54 + ] + ], + [ + [ + 326, + 121 + ], + [ + 315, + 98 + ], + [ + 302, + 131 + ], + [ + 349, + 112 + ], + [ + 313, + 155 + ], + [ + 360, + 135 + ] + ] + ] + }, + { + "task_id": "0005274", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0002_cfg_0004_104122063550#333", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005274/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 301, + 232 + ], + [ + 297, + 213 + ], + [ + 329, + 225 + ], + [ + 272, + 239 + ], + [ + 331, + 242 + ], + [ + 277, + 255 + ] + ], + [ + [ + 291, + 266 + ], + [ + 271, + 244 + ], + [ + 268, + 286 + ], + [ + 312, + 246 + ], + [ + 289, + 309 + ], + [ + 333, + 267 + ] + ], + [ + [ + 327, + 243 + ], + [ + 351, + 242 + ], + [ + 326, + 214 + ], + [ + 328, + 272 + ], + [ + 304, + 215 + ], + [ + 306, + 272 + ] + ], + [ + [ + 284, + 241 + ], + [ + 311, + 255 + ], + [ + 289, + 243 + ], + [ + 279, + 239 + ], + [ + 265, + 230 + ], + [ + 251, + 225 + ] + ] + ] + }, + { + "task_id": "0005276", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0002_cfg_0004_104122063550#432", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005276/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 309, + 246 + ], + [ + 304, + 280 + ], + [ + 276, + 243 + ], + [ + 343, + 249 + ], + [ + 281, + 212 + ], + [ + 347, + 217 + ] + ], + [ + [ + 305, + 243 + ], + [ + 336, + 255 + ], + [ + 314, + 235 + ], + [ + 293, + 251 + ], + [ + 286, + 224 + ], + [ + 261, + 239 + ] + ], + [ + [ + 330, + 219 + ], + [ + 344, + 244 + ], + [ + 320, + 236 + ], + [ + 339, + 205 + ], + [ + 304, + 209 + ], + [ + 325, + 181 + ] + ], + [ + [ + 305, + 226 + ], + [ + 306, + 211 + ], + [ + 318, + 201 + ], + [ + 292, + 253 + ], + [ + 317, + 215 + ], + [ + 292, + 266 + ] + ] + ] + }, + { + "task_id": "0005278", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0002_cfg_0004_104122063550#536", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005278/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 339, + 247 + ], + [ + 338, + 222 + ], + [ + 334, + 229 + ], + [ + 344, + 267 + ], + [ + 336, + 251 + ], + [ + 346, + 290 + ] + ], + [ + [ + 284, + 206 + ], + [ + 261, + 198 + ], + [ + 273, + 189 + ], + [ + 297, + 225 + ], + [ + 294, + 197 + ], + [ + 319, + 232 + ] + ], + [ + [ + 303, + 227 + ], + [ + 304, + 213 + ], + [ + 329, + 213 + ], + [ + 276, + 243 + ], + [ + 326, + 226 + ], + [ + 276, + 255 + ] + ], + [ + [ + 278, + 208 + ], + [ + 283, + 240 + ], + [ + 297, + 210 + ], + [ + 257, + 207 + ], + [ + 292, + 180 + ], + [ + 253, + 173 + ] + ] + ] + }, + { + "task_id": "0005280", + "id": "RH20T_cfg4_task_0218_user_0010_scene_0002_cfg_0004_104122063550#632", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the bulb holder to the socket, install the bulb, turn on the socket to light up the bulb'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005280/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 218, + 267 + ], + [ + 198, + 286 + ], + [ + 197, + 268 + ], + [ + 240, + 267 + ], + [ + 215, + 252 + ], + [ + 257, + 249 + ] + ], + [ + [ + 199, + 338 + ], + [ + 221, + 314 + ], + [ + 222, + 357 + ], + [ + 177, + 320 + ], + [ + 199, + 360 + ], + [ + 154, + 345 + ] + ], + [ + [ + 209, + 293 + ], + [ + 200, + 275 + ], + [ + 239, + 286 + ], + [ + 177, + 300 + ], + [ + 246, + 302 + ], + [ + 186, + 316 + ] + ], + [ + [ + 218, + 295 + ], + [ + 229, + 300 + ], + [ + 219, + 261 + ], + [ + 217, + 329 + ], + [ + 210, + 258 + ], + [ + 207, + 322 + ] + ] + ] + }, + { + "task_id": "0005282", + "id": "RH20T_cfg4_task_0082_user_0015_scene_0010_cfg_0004_036422060909#104", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Scoop water with a large spoon from one bowl to another'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005282/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 110 + ], + [ + 331, + 91 + ], + [ + 298, + 97 + ], + [ + 329, + 122 + ], + [ + 280, + 118 + ], + [ + 313, + 141 + ] + ], + [ + [ + 319, + 92 + ], + [ + 324, + 73 + ], + [ + 318, + 75 + ], + [ + 321, + 111 + ], + [ + 314, + 92 + ], + [ + 316, + 128 + ] + ], + [ + [ + 305, + 74 + ], + [ + 298, + 59 + ], + [ + 281, + 70 + ], + [ + 329, + 78 + ], + [ + 288, + 84 + ], + [ + 334, + 91 + ] + ], + [ + [ + 300, + 88 + ], + [ + 323, + 85 + ], + [ + 306, + 83 + ], + [ + 293, + 92 + ], + [ + 281, + 85 + ], + [ + 270, + 94 + ] + ] + ] + }, + { + "task_id": "0005283", + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_750612070851#142", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005283/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 90, + 205 + ], + [ + 84, + 237 + ], + [ + 80, + 189 + ], + [ + 99, + 220 + ], + [ + 87, + 159 + ], + [ + 104, + 192 + ] + ], + [ + [ + 143, + 205 + ], + [ + 120, + 196 + ], + [ + 135, + 179 + ], + [ + 151, + 230 + ], + [ + 160, + 188 + ], + [ + 174, + 239 + ] + ], + [ + [ + 116, + 171 + ], + [ + 151, + 161 + ], + [ + 108, + 173 + ], + [ + 125, + 170 + ], + [ + 71, + 183 + ], + [ + 84, + 181 + ] + ], + [ + [ + 120, + 185 + ], + [ + 108, + 163 + ], + [ + 92, + 208 + ], + [ + 145, + 163 + ], + [ + 105, + 226 + ], + [ + 154, + 183 + ] + ] + ] + }, + { + "task_id": "0005285", + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_750612070851#236", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005285/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 198, + 186 + ], + [ + 173, + 170 + ], + [ + 207, + 207 + ], + [ + 187, + 163 + ], + [ + 232, + 223 + ], + [ + 214, + 181 + ] + ], + [ + [ + 175, + 248 + ], + [ + 203, + 232 + ], + [ + 174, + 222 + ], + [ + 176, + 277 + ], + [ + 148, + 237 + ], + [ + 148, + 292 + ] + ], + [ + [ + 160, + 214 + ], + [ + 152, + 193 + ], + [ + 131, + 238 + ], + [ + 186, + 192 + ], + [ + 141, + 255 + ], + [ + 192, + 211 + ] + ], + [ + [ + 130, + 218 + ], + [ + 146, + 247 + ], + [ + 136, + 236 + ], + [ + 123, + 196 + ], + [ + 122, + 210 + ], + [ + 108, + 168 + ] + ] + ] + }, + { + "task_id": "0005286", + "id": "RH20T_cfg1_task_0058_user_0015_scene_0003_cfg_0001_750612070851#277", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press the buttons on a row of four power strips to turn them off'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005286/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 160, + 211 + ], + [ + 178, + 242 + ], + [ + 130, + 225 + ], + [ + 192, + 197 + ], + [ + 112, + 194 + ], + [ + 174, + 164 + ] + ], + [ + [ + 164, + 234 + ], + [ + 146, + 243 + ], + [ + 133, + 227 + ], + [ + 195, + 242 + ], + [ + 150, + 220 + ], + [ + 208, + 233 + ] + ], + [ + [ + 203, + 237 + ], + [ + 226, + 261 + ], + [ + 170, + 252 + ], + [ + 232, + 223 + ], + [ + 148, + 226 + ], + [ + 210, + 201 + ] + ], + [ + [ + 183, + 228 + ], + [ + 176, + 208 + ], + [ + 155, + 253 + ], + [ + 209, + 205 + ], + [ + 163, + 270 + ], + [ + 214, + 224 + ] + ] + ] + }, + { + "task_id": "0005287", + "id": "RH20T_cfg1_task_0015_user_0001_scene_0007_cfg_0001_038522063145#244", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005287/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 437, + 107 + ], + [ + 437, + 90 + ], + [ + 452, + 116 + ], + [ + 422, + 98 + ], + [ + 452, + 132 + ], + [ + 423, + 114 + ] + ], + [ + [ + 451, + 122 + ], + [ + 439, + 128 + ], + [ + 437, + 131 + ], + [ + 466, + 112 + ], + [ + 448, + 125 + ], + [ + 478, + 105 + ] + ], + [ + [ + 419, + 136 + ], + [ + 401, + 143 + ], + [ + 424, + 124 + ], + [ + 415, + 148 + ], + [ + 442, + 118 + ], + [ + 432, + 141 + ] + ], + [ + [ + 440, + 119 + ], + [ + 445, + 103 + ], + [ + 452, + 109 + ], + [ + 427, + 128 + ], + [ + 446, + 124 + ], + [ + 422, + 143 + ] + ] + ] + }, + { + "task_id": "0005288", + "id": "RH20T_cfg1_task_0016_user_0014_scene_0001_cfg_0001_819112071037#430", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005288/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 640, + 227 + ], + [ + 585, + 228 + ], + [ + 635, + 278 + ], + [ + 640, + 171 + ], + [ + 640, + 277 + ], + [ + 640, + 169 + ] + ], + [ + [ + 605, + 184 + ], + [ + 618, + 130 + ], + [ + 640, + 208 + ], + [ + 567, + 165 + ], + [ + 638, + 263 + ], + [ + 558, + 211 + ] + ], + [ + [ + 530, + 187 + ], + [ + 558, + 145 + ], + [ + 523, + 159 + ], + [ + 538, + 222 + ], + [ + 501, + 194 + ], + [ + 511, + 260 + ] + ], + [ + [ + 625, + 218 + ], + [ + 640, + 242 + ], + [ + 640, + 260 + ], + [ + 595, + 184 + ], + [ + 613, + 233 + ], + [ + 554, + 160 + ] + ] + ] + }, + { + "task_id": "0005289", + "id": "RH20T_cfg1_task_0009_user_0014_scene_0004_cfg_0001_038522062547#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005289/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 447, + 124 + ], + [ + 451, + 109 + ], + [ + 458, + 145 + ], + [ + 437, + 104 + ], + [ + 454, + 157 + ], + [ + 434, + 118 + ] + ], + [ + [ + 445, + 120 + ], + [ + 422, + 119 + ], + [ + 454, + 140 + ], + [ + 435, + 100 + ], + [ + 478, + 142 + ], + [ + 460, + 101 + ] + ], + [ + [ + 465, + 141 + ], + [ + 478, + 118 + ], + [ + 455, + 130 + ], + [ + 475, + 154 + ], + [ + 443, + 151 + ], + [ + 462, + 176 + ] + ], + [ + [ + 462, + 141 + ], + [ + 444, + 141 + ], + [ + 478, + 158 + ], + [ + 448, + 124 + ], + [ + 495, + 157 + ], + [ + 465, + 124 + ] + ] + ] + }, + { + "task_id": "0005291", + "id": "RH20T_cfg5_task_0075_user_0007_scene_0001_cfg_0005_036422060215#451", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Finish setting up the starting position of a chessboard that is almost arranged'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005291/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 157, + 160 + ], + [ + 136, + 170 + ], + [ + 144, + 141 + ], + [ + 170, + 178 + ], + [ + 165, + 133 + ], + [ + 190, + 168 + ] + ], + [ + [ + 165, + 153 + ], + [ + 165, + 137 + ], + [ + 183, + 164 + ], + [ + 147, + 143 + ], + [ + 182, + 179 + ], + [ + 147, + 159 + ] + ], + [ + [ + 170, + 139 + ], + [ + 175, + 124 + ], + [ + 189, + 150 + ], + [ + 151, + 126 + ], + [ + 185, + 166 + ], + [ + 145, + 142 + ] + ], + [ + [ + 155, + 151 + ], + [ + 165, + 141 + ], + [ + 137, + 153 + ], + [ + 173, + 149 + ], + [ + 125, + 165 + ], + [ + 162, + 161 + ] + ] + ] + }, + { + "task_id": "0005292", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0010_cfg_0007_104122060811#210", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005292/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 340, + 118 + ], + [ + 337, + 104 + ], + [ + 337, + 121 + ], + [ + 343, + 116 + ], + [ + 340, + 135 + ], + [ + 346, + 131 + ] + ], + [ + [ + 349, + 129 + ], + [ + 361, + 130 + ], + [ + 342, + 138 + ], + [ + 357, + 120 + ], + [ + 330, + 137 + ], + [ + 344, + 118 + ] + ], + [ + [ + 340, + 109 + ], + [ + 342, + 123 + ], + [ + 325, + 111 + ], + [ + 354, + 106 + ], + [ + 323, + 97 + ], + [ + 352, + 92 + ] + ], + [ + [ + 336, + 122 + ], + [ + 342, + 129 + ], + [ + 326, + 132 + ], + [ + 347, + 112 + ], + [ + 321, + 125 + ], + [ + 341, + 105 + ] + ] + ] + }, + { + "task_id": "0005294", + "id": "RH20T_cfg1_task_0096_user_0014_scene_0001_cfg_0001_039422060546#173", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take something out of a drawer'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005294/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 260, + 18 + ], + [ + 267, + 8 + ], + [ + 250, + 45 + ], + [ + 272, + 0 + ], + [ + 243, + 57 + ], + [ + 264, + 0 + ] + ], + [ + [ + 287, + 44 + ], + [ + 284, + 17 + ], + [ + 318, + 48 + ], + [ + 257, + 41 + ], + [ + 319, + 72 + ], + [ + 261, + 65 + ] + ], + [ + [ + 287, + 51 + ], + [ + 270, + 73 + ], + [ + 269, + 54 + ], + [ + 306, + 48 + ], + [ + 284, + 35 + ], + [ + 322, + 26 + ] + ], + [ + [ + 278, + 8 + ], + [ + 261, + 17 + ], + [ + 255, + 0 + ], + [ + 301, + 32 + ], + [ + 272, + 0 + ], + [ + 316, + 22 + ] + ] + ] + }, + { + "task_id": "0005298", + "id": "RH20T_cfg1_task_0061_user_0014_scene_0008_cfg_0001_819112071037#361", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005298/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 462, + 116 + ], + [ + 452, + 94 + ], + [ + 514, + 91 + ], + [ + 413, + 140 + ], + [ + 533, + 116 + ], + [ + 420, + 168 + ] + ], + [ + [ + 463, + 101 + ], + [ + 439, + 89 + ], + [ + 485, + 52 + ], + [ + 442, + 147 + ], + [ + 517, + 62 + ], + [ + 468, + 166 + ] + ], + [ + [ + 518, + 97 + ], + [ + 559, + 94 + ], + [ + 553, + 80 + ], + [ + 480, + 115 + ], + [ + 517, + 84 + ], + [ + 449, + 115 + ] + ], + [ + [ + 502, + 126 + ], + [ + 510, + 74 + ], + [ + 459, + 133 + ], + [ + 543, + 120 + ], + [ + 453, + 182 + ], + [ + 534, + 166 + ] + ] + ] + }, + { + "task_id": "0005299", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0010_cfg_0001_750612070853#92", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005299/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 384, + 141 + ], + [ + 411, + 123 + ], + [ + 402, + 119 + ], + [ + 369, + 160 + ], + [ + 370, + 140 + ], + [ + 341, + 179 + ] + ], + [ + [ + 326, + 126 + ], + [ + 324, + 85 + ], + [ + 301, + 115 + ], + [ + 355, + 139 + ], + [ + 304, + 151 + ], + [ + 356, + 178 + ] + ], + [ + [ + 318, + 160 + ], + [ + 287, + 136 + ], + [ + 297, + 194 + ], + [ + 337, + 128 + ], + [ + 327, + 216 + ], + [ + 365, + 152 + ] + ], + [ + [ + 298, + 143 + ], + [ + 302, + 149 + ], + [ + 324, + 173 + ], + [ + 272, + 113 + ], + [ + 322, + 168 + ], + [ + 265, + 102 + ] + ] + ] + }, + { + "task_id": "0005300", + "id": "RH20T_cfg1_task_0003_user_0015_scene_0010_cfg_0001_750612070853#248", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005300/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 469, + 153 + ], + [ + 502, + 132 + ], + [ + 439, + 104 + ], + [ + 499, + 203 + ], + [ + 401, + 125 + ], + [ + 464, + 231 + ] + ], + [ + [ + 428, + 166 + ], + [ + 430, + 122 + ], + [ + 386, + 152 + ], + [ + 475, + 183 + ], + [ + 387, + 189 + ], + [ + 470, + 224 + ] + ], + [ + [ + 422, + 188 + ], + [ + 422, + 166 + ], + [ + 390, + 221 + ], + [ + 452, + 155 + ], + [ + 392, + 237 + ], + [ + 450, + 176 + ] + ], + [ + [ + 398, + 187 + ], + [ + 434, + 203 + ], + [ + 370, + 230 + ], + [ + 424, + 145 + ], + [ + 342, + 213 + ], + [ + 392, + 135 + ] + ] + ] + }, + { + "task_id": "0005302", + "id": "RH20T_cfg4_task_0103_user_0014_scene_0012_cfg_0004_104122062823#130", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005302/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 339, + 124 + ], + [ + 360, + 140 + ], + [ + 363, + 95 + ], + [ + 315, + 152 + ], + [ + 343, + 81 + ], + [ + 298, + 136 + ] + ], + [ + [ + 337, + 124 + ], + [ + 337, + 97 + ], + [ + 370, + 118 + ], + [ + 304, + 131 + ], + [ + 369, + 143 + ], + [ + 306, + 155 + ] + ], + [ + [ + 360, + 97 + ], + [ + 344, + 81 + ], + [ + 343, + 80 + ], + [ + 375, + 112 + ], + [ + 360, + 98 + ], + [ + 392, + 130 + ] + ], + [ + [ + 318, + 101 + ], + [ + 345, + 95 + ], + [ + 326, + 135 + ], + [ + 309, + 66 + ], + [ + 302, + 138 + ], + [ + 285, + 72 + ] + ] + ] + }, + { + "task_id": "0005303", + "id": "RH20T_cfg4_task_0103_user_0014_scene_0012_cfg_0004_104122062823#177", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate to the goal post after holding it down'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005303/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 281, + 131 + ], + [ + 278, + 103 + ], + [ + 315, + 132 + ], + [ + 247, + 131 + ], + [ + 316, + 158 + ], + [ + 252, + 157 + ] + ], + [ + [ + 261, + 119 + ], + [ + 231, + 132 + ], + [ + 240, + 93 + ], + [ + 283, + 146 + ], + [ + 269, + 84 + ], + [ + 311, + 134 + ] + ], + [ + [ + 284, + 115 + ], + [ + 288, + 97 + ], + [ + 275, + 88 + ], + [ + 291, + 140 + ], + [ + 269, + 107 + ], + [ + 287, + 161 + ] + ], + [ + [ + 252, + 103 + ], + [ + 218, + 109 + ], + [ + 244, + 72 + ], + [ + 259, + 131 + ], + [ + 279, + 65 + ], + [ + 291, + 125 + ] + ] + ] + }, + { + "task_id": "0005305", + "id": "RH20T_cfg5_task_0101_user_0007_scene_0010_cfg_0005_105422061350#204", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Clean the table with a cloth'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005305/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 253, + 72 + ], + [ + 257, + 54 + ], + [ + 269, + 59 + ], + [ + 236, + 86 + ], + [ + 265, + 76 + ], + [ + 233, + 102 + ] + ], + [ + [ + 257, + 73 + ], + [ + 241, + 51 + ], + [ + 266, + 72 + ], + [ + 246, + 75 + ], + [ + 281, + 93 + ], + [ + 262, + 98 + ] + ], + [ + [ + 242, + 79 + ], + [ + 227, + 89 + ], + [ + 225, + 89 + ], + [ + 260, + 67 + ], + [ + 238, + 80 + ], + [ + 273, + 58 + ] + ], + [ + [ + 257, + 76 + ], + [ + 269, + 55 + ], + [ + 252, + 85 + ], + [ + 262, + 68 + ], + [ + 240, + 106 + ], + [ + 251, + 88 + ] + ] + ] + }, + { + "task_id": "0005309", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0004_cfg_0006_104122064161#512", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005309/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 413, + 287 + ], + [ + 414, + 303 + ], + [ + 412, + 311 + ], + [ + 414, + 264 + ], + [ + 412, + 296 + ], + [ + 414, + 251 + ] + ], + [ + [ + 426, + 259 + ], + [ + 428, + 283 + ], + [ + 418, + 272 + ], + [ + 434, + 245 + ], + [ + 416, + 250 + ], + [ + 432, + 222 + ] + ], + [ + [ + 448, + 270 + ], + [ + 422, + 277 + ], + [ + 457, + 293 + ], + [ + 439, + 246 + ], + [ + 483, + 286 + ], + [ + 466, + 238 + ] + ], + [ + [ + 404, + 245 + ], + [ + 400, + 231 + ], + [ + 431, + 249 + ], + [ + 379, + 242 + ], + [ + 433, + 262 + ], + [ + 384, + 255 + ] + ] + ] + }, + { + "task_id": "0005310", + "id": "RH20T_cfg6_task_0132_user_0020_scene_0004_cfg_0006_104122064161#814", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks into a cross shape'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005310/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 390, + 221 + ], + [ + 372, + 227 + ], + [ + 405, + 238 + ], + [ + 375, + 204 + ], + [ + 424, + 232 + ], + [ + 394, + 197 + ] + ], + [ + [ + 375, + 211 + ], + [ + 371, + 236 + ], + [ + 370, + 215 + ], + [ + 380, + 207 + ], + [ + 373, + 192 + ], + [ + 383, + 182 + ] + ], + [ + [ + 381, + 206 + ], + [ + 405, + 204 + ], + [ + 383, + 230 + ], + [ + 378, + 182 + ], + [ + 359, + 231 + ], + [ + 355, + 184 + ] + ], + [ + [ + 352, + 208 + ], + [ + 334, + 217 + ], + [ + 335, + 192 + ], + [ + 368, + 223 + ], + [ + 354, + 181 + ], + [ + 387, + 214 + ] + ] + ] + }, + { + "task_id": "0005311", + "id": "RH20T_cfg7_task_0033_user_0020_scene_0001_cfg_0007_104122064161#231", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp one block to sweep the other block onto the mark'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005311/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 406, + 110 + ], + [ + 390, + 102 + ], + [ + 412, + 90 + ], + [ + 401, + 129 + ], + [ + 428, + 98 + ], + [ + 417, + 136 + ] + ], + [ + [ + 406, + 135 + ], + [ + 384, + 134 + ], + [ + 407, + 147 + ], + [ + 404, + 123 + ], + [ + 429, + 148 + ], + [ + 425, + 124 + ] + ], + [ + [ + 395, + 118 + ], + [ + 395, + 133 + ], + [ + 392, + 133 + ], + [ + 398, + 103 + ], + [ + 392, + 119 + ], + [ + 398, + 89 + ] + ], + [ + [ + 390, + 139 + ], + [ + 406, + 151 + ], + [ + 381, + 147 + ], + [ + 398, + 133 + ], + [ + 364, + 135 + ], + [ + 382, + 122 + ] + ] + ] + }, + { + "task_id": "0005312", + "id": "RH20T_cfg3_task_0023_user_0010_scene_0006_cfg_0003_104122062823#142", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn on the power strip by pressing the button'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005312/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 298, + 142 + ], + [ + 311, + 115 + ], + [ + 287, + 151 + ], + [ + 310, + 131 + ], + [ + 272, + 178 + ], + [ + 295, + 161 + ] + ], + [ + [ + 296, + 133 + ], + [ + 298, + 103 + ], + [ + 313, + 118 + ], + [ + 276, + 149 + ], + [ + 310, + 145 + ], + [ + 275, + 177 + ] + ], + [ + [ + 289, + 152 + ], + [ + 278, + 176 + ], + [ + 308, + 144 + ], + [ + 267, + 161 + ], + [ + 320, + 119 + ], + [ + 279, + 134 + ] + ], + [ + [ + 331, + 139 + ], + [ + 337, + 108 + ], + [ + 357, + 137 + ], + [ + 308, + 142 + ], + [ + 351, + 171 + ], + [ + 302, + 172 + ] + ] + ] + }, + { + "task_id": "0005320", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0005_cfg_0003_104122063550#313", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005320/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 327, + 179 + ], + [ + 308, + 184 + ], + [ + 345, + 196 + ], + [ + 309, + 160 + ], + [ + 366, + 192 + ], + [ + 329, + 153 + ] + ], + [ + [ + 298, + 213 + ], + [ + 315, + 193 + ], + [ + 322, + 223 + ], + [ + 276, + 204 + ], + [ + 304, + 246 + ], + [ + 258, + 224 + ] + ], + [ + [ + 311, + 207 + ], + [ + 313, + 194 + ], + [ + 319, + 234 + ], + [ + 304, + 182 + ], + [ + 316, + 245 + ], + [ + 303, + 195 + ] + ], + [ + [ + 320, + 187 + ], + [ + 319, + 160 + ], + [ + 333, + 202 + ], + [ + 308, + 173 + ], + [ + 333, + 228 + ], + [ + 309, + 197 + ] + ] + ] + }, + { + "task_id": "0005321", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0005_cfg_0003_104122063550#362", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005321/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 273, + 152 + ], + [ + 300, + 147 + ], + [ + 275, + 163 + ], + [ + 272, + 141 + ], + [ + 250, + 168 + ], + [ + 243, + 147 + ] + ], + [ + [ + 287, + 148 + ], + [ + 285, + 135 + ], + [ + 297, + 171 + ], + [ + 278, + 127 + ], + [ + 298, + 182 + ], + [ + 279, + 140 + ] + ], + [ + [ + 280, + 154 + ], + [ + 304, + 140 + ], + [ + 266, + 132 + ], + [ + 295, + 177 + ], + [ + 241, + 145 + ], + [ + 269, + 191 + ] + ], + [ + [ + 298, + 151 + ], + [ + 305, + 124 + ], + [ + 323, + 158 + ], + [ + 274, + 144 + ], + [ + 315, + 185 + ], + [ + 267, + 170 + ] + ] + ] + }, + { + "task_id": "0005322", + "id": "RH20T_cfg6_task_0127_user_0020_scene_0006_cfg_0006_104122064161#233", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Plug in the power cord of the desk lamp, turn on the socket, and light up the desk lamp'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005322/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 423, + 149 + ], + [ + 425, + 129 + ], + [ + 429, + 159 + ], + [ + 418, + 139 + ], + [ + 427, + 179 + ], + [ + 416, + 158 + ] + ], + [ + [ + 405, + 164 + ], + [ + 420, + 154 + ], + [ + 420, + 181 + ], + [ + 391, + 147 + ], + [ + 406, + 189 + ], + [ + 378, + 157 + ] + ], + [ + [ + 407, + 147 + ], + [ + 428, + 144 + ], + [ + 402, + 144 + ], + [ + 412, + 151 + ], + [ + 381, + 148 + ], + [ + 390, + 154 + ] + ], + [ + [ + 414, + 130 + ], + [ + 428, + 136 + ], + [ + 424, + 110 + ], + [ + 404, + 149 + ], + [ + 409, + 104 + ], + [ + 389, + 144 + ] + ] + ] + }, + { + "task_id": "0005326", + "id": "RH20T_cfg4_task_0005_user_0010_scene_0004_cfg_0004_104122062295#95", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Approach and touch the side of the small block'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005326/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 145, + 180 + ], + [ + 105, + 194 + ], + [ + 121, + 181 + ], + [ + 171, + 180 + ], + [ + 156, + 169 + ], + [ + 205, + 168 + ] + ], + [ + [ + 126, + 143 + ], + [ + 130, + 106 + ], + [ + 104, + 121 + ], + [ + 149, + 166 + ], + [ + 101, + 154 + ], + [ + 144, + 200 + ] + ], + [ + [ + 167, + 175 + ], + [ + 207, + 162 + ], + [ + 161, + 151 + ], + [ + 174, + 202 + ], + [ + 121, + 164 + ], + [ + 129, + 217 + ] + ], + [ + [ + 122, + 142 + ], + [ + 135, + 173 + ], + [ + 154, + 123 + ], + [ + 86, + 162 + ], + [ + 142, + 89 + ], + [ + 69, + 126 + ] + ] + ] + }, + { + "task_id": "0005327", + "id": "RH20T_cfg1_task_0004_user_0014_scene_0004_cfg_0001_750612070853#168", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005327/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 342, + 162 + ], + [ + 334, + 127 + ], + [ + 364, + 178 + ], + [ + 317, + 144 + ], + [ + 374, + 214 + ], + [ + 325, + 184 + ] + ], + [ + [ + 306, + 194 + ], + [ + 308, + 157 + ], + [ + 262, + 193 + ], + [ + 350, + 196 + ], + [ + 263, + 226 + ], + [ + 347, + 229 + ] + ], + [ + [ + 343, + 165 + ], + [ + 324, + 161 + ], + [ + 302, + 156 + ], + [ + 387, + 174 + ], + [ + 320, + 160 + ], + [ + 400, + 176 + ] + ], + [ + [ + 336, + 182 + ], + [ + 330, + 148 + ], + [ + 341, + 204 + ], + [ + 330, + 156 + ], + [ + 348, + 240 + ], + [ + 337, + 195 + ] + ] + ] + }, + { + "task_id": "0005329", + "id": "RH20T_cfg6_task_0042_user_0014_scene_0004_cfg_0006_104122060811#118", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Screw the lid onto the jar'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005329/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 394, + 126 + ], + [ + 388, + 118 + ], + [ + 388, + 139 + ], + [ + 401, + 113 + ], + [ + 394, + 148 + ], + [ + 408, + 122 + ] + ], + [ + [ + 402, + 126 + ], + [ + 405, + 117 + ], + [ + 416, + 121 + ], + [ + 388, + 131 + ], + [ + 412, + 130 + ], + [ + 385, + 139 + ] + ], + [ + [ + 398, + 143 + ], + [ + 399, + 132 + ], + [ + 400, + 131 + ], + [ + 396, + 155 + ], + [ + 399, + 140 + ], + [ + 395, + 164 + ] + ], + [ + [ + 407, + 135 + ], + [ + 415, + 142 + ], + [ + 401, + 149 + ], + [ + 413, + 121 + ], + [ + 392, + 142 + ], + [ + 405, + 113 + ] + ] + ] + }, + { + "task_id": "0005331", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122061330#224", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005331/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 255, + 163 + ], + [ + 240, + 168 + ], + [ + 250, + 188 + ], + [ + 261, + 136 + ], + [ + 263, + 182 + ], + [ + 275, + 133 + ] + ], + [ + [ + 226, + 170 + ], + [ + 209, + 154 + ], + [ + 248, + 158 + ], + [ + 202, + 184 + ], + [ + 266, + 174 + ], + [ + 219, + 202 + ] + ], + [ + [ + 274, + 190 + ], + [ + 261, + 206 + ], + [ + 250, + 180 + ], + [ + 298, + 199 + ], + [ + 263, + 166 + ], + [ + 309, + 184 + ] + ], + [ + [ + 256, + 180 + ], + [ + 257, + 200 + ], + [ + 229, + 180 + ], + [ + 283, + 181 + ], + [ + 227, + 159 + ], + [ + 284, + 160 + ] + ] + ] + }, + { + "task_id": "0005332", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122061330#329", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005332/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 257, + 162 + ], + [ + 243, + 184 + ], + [ + 262, + 177 + ], + [ + 252, + 145 + ], + [ + 275, + 157 + ], + [ + 266, + 123 + ] + ], + [ + [ + 256, + 130 + ], + [ + 243, + 146 + ], + [ + 280, + 146 + ], + [ + 232, + 113 + ], + [ + 291, + 131 + ], + [ + 246, + 99 + ] + ], + [ + [ + 248, + 168 + ], + [ + 274, + 176 + ], + [ + 262, + 144 + ], + [ + 234, + 191 + ], + [ + 234, + 135 + ], + [ + 207, + 183 + ] + ], + [ + [ + 264, + 159 + ], + [ + 259, + 175 + ], + [ + 245, + 177 + ], + [ + 284, + 139 + ], + [ + 250, + 161 + ], + [ + 287, + 125 + ] + ] + ] + }, + { + "task_id": "0005333", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122061330#411", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005333/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 286, + 151 + ], + [ + 314, + 148 + ], + [ + 288, + 179 + ], + [ + 284, + 122 + ], + [ + 260, + 181 + ], + [ + 256, + 125 + ] + ], + [ + [ + 266, + 169 + ], + [ + 295, + 166 + ], + [ + 271, + 194 + ], + [ + 262, + 144 + ], + [ + 241, + 198 + ], + [ + 233, + 147 + ] + ], + [ + [ + 286, + 157 + ], + [ + 283, + 131 + ], + [ + 301, + 148 + ], + [ + 273, + 165 + ], + [ + 305, + 176 + ], + [ + 276, + 190 + ] + ], + [ + [ + 261, + 161 + ], + [ + 247, + 182 + ], + [ + 266, + 180 + ], + [ + 256, + 141 + ], + [ + 278, + 162 + ], + [ + 270, + 122 + ] + ] + ] + }, + { + "task_id": "0005334", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122061330#535", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005334/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 245, + 183 + ], + [ + 270, + 171 + ], + [ + 242, + 181 + ], + [ + 247, + 185 + ], + [ + 218, + 192 + ], + [ + 220, + 198 + ] + ], + [ + [ + 261, + 156 + ], + [ + 246, + 169 + ], + [ + 262, + 178 + ], + [ + 260, + 131 + ], + [ + 276, + 165 + ], + [ + 275, + 119 + ] + ], + [ + [ + 282, + 154 + ], + [ + 263, + 142 + ], + [ + 271, + 180 + ], + [ + 293, + 127 + ], + [ + 291, + 193 + ], + [ + 315, + 139 + ] + ], + [ + [ + 267, + 137 + ], + [ + 274, + 114 + ], + [ + 259, + 150 + ], + [ + 276, + 122 + ], + [ + 251, + 173 + ], + [ + 268, + 146 + ] + ] + ] + }, + { + "task_id": "0005335", + "id": "RH20T_cfg6_task_0043_user_0014_scene_0004_cfg_0006_104122061330#683", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005335/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 267, + 152 + ], + [ + 257, + 145 + ], + [ + 258, + 179 + ], + [ + 276, + 126 + ], + [ + 268, + 184 + ], + [ + 285, + 133 + ] + ], + [ + [ + 284, + 170 + ], + [ + 259, + 160 + ], + [ + 295, + 146 + ], + [ + 273, + 195 + ], + [ + 320, + 157 + ], + [ + 297, + 205 + ] + ], + [ + [ + 259, + 130 + ], + [ + 253, + 116 + ], + [ + 278, + 111 + ], + [ + 240, + 147 + ], + [ + 285, + 125 + ], + [ + 245, + 162 + ] + ], + [ + [ + 283, + 133 + ], + [ + 295, + 154 + ], + [ + 281, + 148 + ], + [ + 286, + 116 + ], + [ + 270, + 129 + ], + [ + 275, + 95 + ] + ] + ] + }, + { + "task_id": "0005336", + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104422070011#157", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005336/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 270, + 181 + ], + [ + 281, + 168 + ], + [ + 239, + 193 + ], + [ + 299, + 169 + ], + [ + 231, + 203 + ], + [ + 288, + 180 + ] + ], + [ + [ + 289, + 185 + ], + [ + 297, + 214 + ], + [ + 270, + 176 + ], + [ + 306, + 193 + ], + [ + 263, + 147 + ], + [ + 299, + 166 + ] + ], + [ + [ + 258, + 180 + ], + [ + 241, + 151 + ], + [ + 250, + 190 + ], + [ + 269, + 169 + ], + [ + 266, + 218 + ], + [ + 288, + 201 + ] + ], + [ + [ + 240, + 206 + ], + [ + 269, + 190 + ], + [ + 242, + 226 + ], + [ + 237, + 184 + ], + [ + 212, + 243 + ], + [ + 204, + 201 + ] + ] + ] + }, + { + "task_id": "0005337", + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104422070011#211", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005337/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 285, + 149 + ], + [ + 298, + 179 + ], + [ + 306, + 147 + ], + [ + 267, + 151 + ], + [ + 292, + 113 + ], + [ + 254, + 122 + ] + ], + [ + [ + 316, + 173 + ], + [ + 302, + 140 + ], + [ + 337, + 158 + ], + [ + 293, + 189 + ], + [ + 350, + 188 + ], + [ + 308, + 222 + ] + ], + [ + [ + 292, + 175 + ], + [ + 300, + 167 + ], + [ + 263, + 193 + ], + [ + 319, + 158 + ], + [ + 257, + 200 + ], + [ + 310, + 166 + ] + ], + [ + [ + 314, + 160 + ], + [ + 319, + 133 + ], + [ + 283, + 166 + ], + [ + 343, + 155 + ], + [ + 280, + 192 + ], + [ + 337, + 179 + ] + ] + ] + }, + { + "task_id": "0005338", + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104422070011#251", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005338/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 262, + 178 + ], + [ + 241, + 208 + ], + [ + 272, + 182 + ], + [ + 250, + 173 + ], + [ + 291, + 155 + ], + [ + 272, + 141 + ] + ], + [ + [ + 289, + 197 + ], + [ + 301, + 187 + ], + [ + 259, + 214 + ], + [ + 316, + 182 + ], + [ + 250, + 223 + ], + [ + 304, + 192 + ] + ], + [ + [ + 283, + 157 + ], + [ + 285, + 139 + ], + [ + 320, + 163 + ], + [ + 245, + 151 + ], + [ + 316, + 178 + ], + [ + 246, + 167 + ] + ], + [ + [ + 254, + 191 + ], + [ + 284, + 190 + ], + [ + 248, + 229 + ], + [ + 261, + 154 + ], + [ + 221, + 228 + ], + [ + 234, + 156 + ] + ] + ] + }, + { + "task_id": "0005339", + "id": "RH20T_cfg4_task_0094_user_0014_scene_0003_cfg_0004_104422070011#412", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Open a box'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005339/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 281, + 112 + ], + [ + 285, + 108 + ], + [ + 254, + 131 + ], + [ + 309, + 93 + ], + [ + 253, + 133 + ], + [ + 304, + 97 + ] + ], + [ + [ + 306, + 120 + ], + [ + 297, + 107 + ], + [ + 316, + 91 + ], + [ + 297, + 147 + ], + [ + 327, + 105 + ], + [ + 307, + 163 + ] + ], + [ + [ + 265, + 104 + ], + [ + 293, + 86 + ], + [ + 249, + 113 + ], + [ + 278, + 96 + ], + [ + 222, + 130 + ], + [ + 254, + 112 + ] + ], + [ + [ + 270, + 113 + ], + [ + 237, + 116 + ], + [ + 266, + 84 + ], + [ + 274, + 139 + ], + [ + 300, + 82 + ], + [ + 305, + 137 + ] + ] + ] + }, + { + "task_id": "0005345", + "id": "RH20T_cfg2_task_0022_user_0003_scene_0004_cfg_0002_104422070044#88", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Wave the flag'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005345/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 292, + 160 + ], + [ + 308, + 163 + ], + [ + 293, + 146 + ], + [ + 290, + 175 + ], + [ + 278, + 144 + ], + [ + 274, + 173 + ] + ], + [ + [ + 312, + 165 + ], + [ + 295, + 165 + ], + [ + 314, + 169 + ], + [ + 310, + 162 + ], + [ + 330, + 170 + ], + [ + 325, + 162 + ] + ], + [ + [ + 291, + 160 + ], + [ + 289, + 147 + ], + [ + 285, + 171 + ], + [ + 298, + 150 + ], + [ + 288, + 184 + ], + [ + 300, + 163 + ] + ], + [ + [ + 274, + 150 + ], + [ + 266, + 165 + ], + [ + 269, + 149 + ], + [ + 278, + 152 + ], + [ + 277, + 133 + ], + [ + 286, + 138 + ] + ] + ] + }, + { + "task_id": "0005349", + "id": "RH20T_cfg6_task_0022_user_0014_scene_0007_cfg_0006_104122060811#126", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Wave the flag'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005349/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 338, + 152 + ], + [ + 337, + 139 + ], + [ + 346, + 145 + ], + [ + 331, + 158 + ], + [ + 348, + 159 + ], + [ + 332, + 171 + ] + ], + [ + [ + 328, + 159 + ], + [ + 323, + 171 + ], + [ + 340, + 166 + ], + [ + 317, + 152 + ], + [ + 345, + 154 + ], + [ + 322, + 140 + ] + ], + [ + [ + 342, + 159 + ], + [ + 349, + 146 + ], + [ + 334, + 157 + ], + [ + 350, + 160 + ], + [ + 327, + 169 + ], + [ + 343, + 173 + ] + ], + [ + [ + 333, + 150 + ], + [ + 334, + 142 + ], + [ + 348, + 148 + ], + [ + 319, + 152 + ], + [ + 346, + 156 + ], + [ + 319, + 160 + ] + ] + ] + }, + { + "task_id": "0005350", + "id": "RH20T_cfg5_task_0002_user_0014_scene_0010_cfg_0005_036422060215#186", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005350/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 275, + 146 + ], + [ + 295, + 148 + ], + [ + 273, + 130 + ], + [ + 277, + 163 + ], + [ + 254, + 127 + ], + [ + 257, + 160 + ] + ], + [ + [ + 258, + 140 + ], + [ + 258, + 143 + ], + [ + 238, + 147 + ], + [ + 278, + 133 + ], + [ + 237, + 144 + ], + [ + 279, + 130 + ] + ], + [ + [ + 255, + 144 + ], + [ + 244, + 127 + ], + [ + 255, + 141 + ], + [ + 256, + 146 + ], + [ + 267, + 159 + ], + [ + 266, + 162 + ] + ], + [ + [ + 250, + 139 + ], + [ + 252, + 122 + ], + [ + 269, + 134 + ], + [ + 231, + 143 + ], + [ + 267, + 150 + ], + [ + 230, + 159 + ] + ] + ] + }, + { + "task_id": "0005357", + "id": "RH20T_cfg7_task_0009_user_0014_scene_0001_cfg_0007_104122061602#93", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005357/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 438, + 154 + ], + [ + 444, + 132 + ], + [ + 456, + 167 + ], + [ + 422, + 143 + ], + [ + 450, + 188 + ], + [ + 418, + 163 + ] + ], + [ + [ + 435, + 129 + ], + [ + 442, + 125 + ], + [ + 414, + 142 + ], + [ + 455, + 115 + ], + [ + 408, + 144 + ], + [ + 447, + 119 + ] + ], + [ + [ + 450, + 147 + ], + [ + 441, + 135 + ], + [ + 429, + 139 + ], + [ + 470, + 155 + ], + [ + 439, + 152 + ], + [ + 482, + 169 + ] + ], + [ + [ + 452, + 157 + ], + [ + 472, + 151 + ], + [ + 435, + 148 + ], + [ + 471, + 168 + ], + [ + 414, + 155 + ], + [ + 449, + 175 + ] + ] + ] + }, + { + "task_id": "0005364", + "id": "RH20T_cfg7_task_0036_user_0020_scene_0001_cfg_0007_104122061018#143", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Shake the test tube'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005364/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 325, + 107 + ], + [ + 305, + 85 + ], + [ + 305, + 126 + ], + [ + 345, + 88 + ], + [ + 325, + 148 + ], + [ + 364, + 110 + ] + ], + [ + [ + 303, + 97 + ], + [ + 291, + 87 + ], + [ + 306, + 125 + ], + [ + 299, + 67 + ], + [ + 319, + 137 + ], + [ + 312, + 77 + ] + ], + [ + [ + 314, + 71 + ], + [ + 319, + 46 + ], + [ + 339, + 60 + ], + [ + 292, + 80 + ], + [ + 335, + 88 + ], + [ + 288, + 105 + ] + ], + [ + [ + 291, + 127 + ], + [ + 265, + 130 + ], + [ + 295, + 98 + ], + [ + 286, + 156 + ], + [ + 322, + 93 + ], + [ + 314, + 153 + ] + ] + ] + }, + { + "task_id": "0005365", + "id": "RH20T_cfg2_task_0104_user_0007_scene_0007_cfg_0002_036422060215#205", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Drag the plate back after holding it down'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005365/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 306, + 215 + ], + [ + 297, + 200 + ], + [ + 289, + 226 + ], + [ + 324, + 203 + ], + [ + 299, + 242 + ], + [ + 335, + 218 + ] + ], + [ + [ + 313, + 190 + ], + [ + 315, + 177 + ], + [ + 331, + 184 + ], + [ + 295, + 197 + ], + [ + 329, + 196 + ], + [ + 294, + 210 + ] + ], + [ + [ + 331, + 209 + ], + [ + 321, + 198 + ], + [ + 348, + 199 + ], + [ + 314, + 219 + ], + [ + 358, + 211 + ], + [ + 324, + 231 + ] + ], + [ + [ + 313, + 184 + ], + [ + 322, + 195 + ], + [ + 296, + 188 + ], + [ + 330, + 180 + ], + [ + 286, + 177 + ], + [ + 321, + 168 + ] + ] + ] + }, + { + "task_id": "0005370", + "id": "RH20T_cfg1_task_0009_user_0014_scene_0007_cfg_0001_038522062547#327", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take out one Hanoi block and throw it aside'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005370/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 393, + 88 + ], + [ + 396, + 76 + ], + [ + 417, + 86 + ], + [ + 370, + 90 + ], + [ + 416, + 99 + ], + [ + 367, + 103 + ] + ], + [ + [ + 396, + 60 + ], + [ + 383, + 49 + ], + [ + 405, + 36 + ], + [ + 387, + 82 + ], + [ + 419, + 47 + ], + [ + 399, + 94 + ] + ], + [ + [ + 405, + 86 + ], + [ + 410, + 88 + ], + [ + 417, + 106 + ], + [ + 394, + 66 + ], + [ + 412, + 106 + ], + [ + 388, + 64 + ] + ], + [ + [ + 399, + 73 + ], + [ + 402, + 58 + ], + [ + 415, + 89 + ], + [ + 383, + 58 + ], + [ + 412, + 102 + ], + [ + 381, + 72 + ] + ] + ] + }, + { + "task_id": "0005372", + "id": "RH20T_cfg1_task_0015_user_0001_scene_0007_cfg_0001_039422060546#202", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Play the drum'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005372/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 352, + 188 + ], + [ + 352, + 195 + ], + [ + 385, + 179 + ], + [ + 319, + 197 + ], + [ + 388, + 170 + ], + [ + 316, + 189 + ] + ], + [ + [ + 371, + 187 + ], + [ + 370, + 168 + ], + [ + 403, + 192 + ], + [ + 341, + 182 + ], + [ + 403, + 209 + ], + [ + 344, + 199 + ] + ], + [ + [ + 353, + 180 + ], + [ + 368, + 188 + ], + [ + 336, + 203 + ], + [ + 370, + 157 + ], + [ + 318, + 195 + ], + [ + 353, + 147 + ] + ], + [ + [ + 381, + 187 + ], + [ + 382, + 177 + ], + [ + 350, + 188 + ], + [ + 413, + 185 + ], + [ + 347, + 199 + ], + [ + 415, + 196 + ] + ] + ] + }, + { + "task_id": "0005373", + "id": "RH20T_cfg5_task_0016_user_0010_scene_0010_cfg_0005_104422070044#80", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hit the pool ball'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005373/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 376, + 188 + ], + [ + 367, + 199 + ], + [ + 391, + 186 + ], + [ + 363, + 191 + ], + [ + 399, + 176 + ], + [ + 372, + 180 + ] + ], + [ + [ + 389, + 181 + ], + [ + 389, + 167 + ], + [ + 403, + 191 + ], + [ + 376, + 171 + ], + [ + 402, + 204 + ], + [ + 377, + 184 + ] + ], + [ + [ + 377, + 159 + ], + [ + 368, + 152 + ], + [ + 390, + 166 + ], + [ + 363, + 152 + ], + [ + 399, + 174 + ], + [ + 372, + 160 + ] + ], + [ + [ + 411, + 182 + ], + [ + 430, + 180 + ], + [ + 406, + 189 + ], + [ + 417, + 175 + ], + [ + 388, + 190 + ], + [ + 397, + 177 + ] + ] + ] + }, + { + "task_id": "0005374", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122064161#99", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We present some images with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper requires to pick up an object\u2014so which image below contains the correct grasping pose? The available options are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005374/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 316, + 180 + ], + [ + 337, + 192 + ], + [ + 318, + 167 + ], + [ + 315, + 192 + ], + [ + 296, + 155 + ], + [ + 295, + 181 + ] + ], + [ + [ + 327, + 191 + ], + [ + 326, + 213 + ], + [ + 305, + 191 + ], + [ + 349, + 192 + ], + [ + 307, + 168 + ], + [ + 350, + 170 + ] + ], + [ + [ + 319, + 175 + ], + [ + 321, + 157 + ], + [ + 318, + 191 + ], + [ + 319, + 161 + ], + [ + 316, + 208 + ], + [ + 316, + 178 + ] + ], + [ + [ + 341, + 175 + ], + [ + 365, + 182 + ], + [ + 338, + 186 + ], + [ + 345, + 162 + ], + [ + 315, + 180 + ], + [ + 320, + 155 + ] + ] + ] + }, + { + "task_id": "0005375", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122064161#162", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005375/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 351, + 145 + ], + [ + 328, + 146 + ], + [ + 352, + 167 + ], + [ + 350, + 122 + ], + [ + 375, + 167 + ], + [ + 374, + 121 + ] + ], + [ + [ + 371, + 146 + ], + [ + 349, + 151 + ], + [ + 376, + 162 + ], + [ + 367, + 131 + ], + [ + 399, + 156 + ], + [ + 389, + 126 + ] + ], + [ + [ + 360, + 168 + ], + [ + 366, + 150 + ], + [ + 362, + 183 + ], + [ + 359, + 154 + ], + [ + 356, + 201 + ], + [ + 355, + 171 + ] + ], + [ + [ + 357, + 191 + ], + [ + 334, + 197 + ], + [ + 363, + 211 + ], + [ + 350, + 169 + ], + [ + 386, + 205 + ], + [ + 374, + 162 + ] + ] + ] + }, + { + "task_id": "0005376", + "id": "RH20T_cfg7_task_0037_user_0014_scene_0002_cfg_0007_104122064161#288", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005376/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 395, + 167 + ], + [ + 401, + 150 + ], + [ + 397, + 183 + ], + [ + 393, + 153 + ], + [ + 392, + 200 + ], + [ + 388, + 169 + ] + ], + [ + [ + 378, + 176 + ], + [ + 401, + 178 + ], + [ + 375, + 200 + ], + [ + 380, + 152 + ], + [ + 352, + 197 + ], + [ + 357, + 150 + ] + ], + [ + [ + 416, + 161 + ], + [ + 439, + 164 + ], + [ + 414, + 144 + ], + [ + 418, + 179 + ], + [ + 391, + 141 + ], + [ + 395, + 176 + ] + ], + [ + [ + 412, + 178 + ], + [ + 398, + 194 + ], + [ + 431, + 193 + ], + [ + 394, + 163 + ], + [ + 447, + 177 + ], + [ + 408, + 146 + ] + ] + ] + }, + { + "task_id": "0005378", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_105422061350#247", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005378/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 326, + 143 + ], + [ + 320, + 137 + ], + [ + 346, + 129 + ], + [ + 306, + 157 + ], + [ + 354, + 134 + ], + [ + 312, + 164 + ] + ], + [ + [ + 339, + 161 + ], + [ + 341, + 141 + ], + [ + 357, + 170 + ], + [ + 320, + 151 + ], + [ + 355, + 190 + ], + [ + 317, + 172 + ] + ], + [ + [ + 320, + 151 + ], + [ + 298, + 158 + ], + [ + 324, + 171 + ], + [ + 316, + 129 + ], + [ + 345, + 164 + ], + [ + 338, + 122 + ] + ], + [ + [ + 329, + 141 + ], + [ + 335, + 126 + ], + [ + 342, + 130 + ], + [ + 314, + 153 + ], + [ + 336, + 144 + ], + [ + 308, + 167 + ] + ] + ] + }, + { + "task_id": "0005380", + "id": "RH20T_cfg2_task_0037_user_0007_scene_0006_cfg_0002_105422061350#629", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the blocks in a vertical line of five'. The images we supply contain orange fork-like gripper patterns: these patterns represent the potential poses the gripper may rely on to grab an object. Which of the images below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005380/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 292, + 179 + ], + [ + 294, + 166 + ], + [ + 318, + 183 + ], + [ + 266, + 175 + ], + [ + 314, + 195 + ], + [ + 265, + 187 + ] + ], + [ + [ + 308, + 187 + ], + [ + 322, + 208 + ], + [ + 312, + 183 + ], + [ + 303, + 191 + ], + [ + 298, + 162 + ], + [ + 288, + 168 + ] + ], + [ + [ + 291, + 180 + ], + [ + 310, + 167 + ], + [ + 275, + 162 + ], + [ + 307, + 199 + ], + [ + 255, + 176 + ], + [ + 287, + 213 + ] + ], + [ + [ + 288, + 186 + ], + [ + 306, + 167 + ], + [ + 308, + 204 + ], + [ + 269, + 169 + ], + [ + 291, + 224 + ], + [ + 252, + 189 + ] + ] + ] + }, + { + "task_id": "0005385", + "id": "RH20T_cfg1_task_0004_user_0015_scene_0002_cfg_0001_819112071037#175", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up the small block on the left and move it to the right'. Among the images we offer, there are orange fork-like gripper patterns\u2014they represent the possible poses the gripper may use to grab an object. Which of the following images has the correct grasping pose? The available choice is [A, B, C, D].Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005385/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 206, + 169 + ], + [ + 236, + 216 + ], + [ + 142, + 187 + ], + [ + 262, + 154 + ], + [ + 118, + 141 + ], + [ + 234, + 115 + ] + ], + [ + [ + 183, + 181 + ], + [ + 243, + 157 + ], + [ + 194, + 235 + ], + [ + 174, + 135 + ], + [ + 129, + 261 + ], + [ + 118, + 157 + ] + ], + [ + [ + 233, + 145 + ], + [ + 232, + 76 + ], + [ + 163, + 143 + ], + [ + 300, + 147 + ], + [ + 166, + 207 + ], + [ + 298, + 209 + ] + ], + [ + [ + 268, + 143 + ], + [ + 277, + 127 + ], + [ + 247, + 204 + ], + [ + 291, + 76 + ], + [ + 232, + 235 + ], + [ + 282, + 88 + ] + ] + ] + }, + { + "task_id": "0005386", + "id": "RH20T_cfg7_task_0043_user_0020_scene_0001_cfg_0007_104122060811#163", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unscrew the lid from the jar'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005386/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 375, + 121 + ], + [ + 373, + 118 + ], + [ + 367, + 133 + ], + [ + 383, + 110 + ], + [ + 369, + 136 + ], + [ + 385, + 113 + ] + ], + [ + [ + 373, + 124 + ], + [ + 375, + 112 + ], + [ + 386, + 128 + ], + [ + 359, + 121 + ], + [ + 383, + 141 + ], + [ + 357, + 133 + ] + ], + [ + [ + 380, + 121 + ], + [ + 387, + 126 + ], + [ + 369, + 130 + ], + [ + 392, + 113 + ], + [ + 361, + 125 + ], + [ + 385, + 107 + ] + ], + [ + [ + 385, + 130 + ], + [ + 372, + 122 + ], + [ + 392, + 121 + ], + [ + 376, + 140 + ], + [ + 405, + 128 + ], + [ + 389, + 148 + ] + ] + ] + }, + { + "task_id": "0005387", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_037522062165#166", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005387/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 209, + 205 + ], + [ + 204, + 190 + ], + [ + 186, + 191 + ], + [ + 232, + 220 + ], + [ + 191, + 205 + ], + [ + 235, + 233 + ] + ], + [ + [ + 202, + 209 + ], + [ + 210, + 182 + ], + [ + 215, + 207 + ], + [ + 186, + 212 + ], + [ + 208, + 232 + ], + [ + 177, + 241 + ] + ], + [ + [ + 204, + 184 + ], + [ + 187, + 185 + ], + [ + 214, + 157 + ], + [ + 193, + 212 + ], + [ + 232, + 155 + ], + [ + 211, + 213 + ] + ], + [ + [ + 170, + 215 + ], + [ + 150, + 192 + ], + [ + 152, + 226 + ], + [ + 190, + 202 + ], + [ + 171, + 247 + ], + [ + 210, + 225 + ] + ] + ] + }, + { + "task_id": "0005388", + "id": "RH20T_cfg5_task_0003_user_0010_scene_0001_cfg_0005_037522062165#247", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005388/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 234, + 235 + ], + [ + 258, + 215 + ], + [ + 250, + 239 + ], + [ + 220, + 231 + ], + [ + 223, + 261 + ], + [ + 196, + 251 + ] + ], + [ + [ + 245, + 220 + ], + [ + 239, + 246 + ], + [ + 241, + 204 + ], + [ + 249, + 239 + ], + [ + 246, + 178 + ], + [ + 255, + 211 + ] + ], + [ + [ + 259, + 224 + ], + [ + 257, + 209 + ], + [ + 235, + 210 + ], + [ + 284, + 239 + ], + [ + 238, + 224 + ], + [ + 285, + 252 + ] + ], + [ + [ + 273, + 262 + ], + [ + 251, + 267 + ], + [ + 285, + 291 + ], + [ + 262, + 235 + ], + [ + 305, + 285 + ], + [ + 282, + 232 + ] + ] + ] + }, + { + "task_id": "0005394", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0001_cfg_0003_104122062823#152", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005394/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 430, + 101 + ], + [ + 422, + 59 + ], + [ + 441, + 94 + ], + [ + 418, + 108 + ], + [ + 448, + 131 + ], + [ + 427, + 151 + ] + ], + [ + [ + 405, + 193 + ], + [ + 428, + 218 + ], + [ + 376, + 186 + ], + [ + 432, + 199 + ], + [ + 356, + 164 + ], + [ + 411, + 178 + ] + ], + [ + [ + 393, + 130 + ], + [ + 401, + 99 + ], + [ + 429, + 147 + ], + [ + 358, + 115 + ], + [ + 420, + 174 + ], + [ + 354, + 142 + ] + ], + [ + [ + 401, + 127 + ], + [ + 373, + 93 + ], + [ + 366, + 155 + ], + [ + 436, + 98 + ], + [ + 394, + 189 + ], + [ + 465, + 133 + ] + ] + ] + }, + { + "task_id": "0005396", + "id": "RH20T_cfg3_task_0003_user_0016_scene_0001_cfg_0003_104122062823#263", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Press three buttons from left to right in sequence'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "C" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005396/concat.jpg" + ], + "gt": "C", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 157, + 87 + ], + [ + 155, + 114 + ], + [ + 176, + 102 + ], + [ + 139, + 72 + ], + [ + 180, + 72 + ], + [ + 141, + 43 + ] + ], + [ + [ + 158, + 55 + ], + [ + 188, + 60 + ], + [ + 145, + 68 + ], + [ + 172, + 40 + ], + [ + 114, + 64 + ], + [ + 138, + 33 + ] + ], + [ + [ + 169, + 51 + ], + [ + 163, + 22 + ], + [ + 195, + 61 + ], + [ + 145, + 40 + ], + [ + 200, + 87 + ], + [ + 152, + 65 + ] + ], + [ + [ + 153, + 58 + ], + [ + 147, + 31 + ], + [ + 147, + 72 + ], + [ + 160, + 44 + ], + [ + 152, + 99 + ], + [ + 166, + 73 + ] + ] + ] + }, + { + "task_id": "0005397", + "id": "RH20T_cfg5_task_0081_user_0007_scene_0003_cfg_0005_104422070044#100", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005397/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 385, + 152 + ], + [ + 392, + 141 + ], + [ + 369, + 149 + ], + [ + 402, + 155 + ], + [ + 362, + 160 + ], + [ + 396, + 166 + ] + ], + [ + [ + 385, + 160 + ], + [ + 388, + 147 + ], + [ + 375, + 168 + ], + [ + 395, + 152 + ], + [ + 373, + 181 + ], + [ + 393, + 165 + ] + ], + [ + [ + 392, + 165 + ], + [ + 402, + 155 + ], + [ + 405, + 168 + ], + [ + 379, + 163 + ], + [ + 393, + 179 + ], + [ + 369, + 173 + ] + ], + [ + [ + 386, + 161 + ], + [ + 387, + 147 + ], + [ + 401, + 169 + ], + [ + 372, + 154 + ], + [ + 400, + 182 + ], + [ + 371, + 167 + ] + ] + ] + }, + { + "task_id": "0005398", + "id": "RH20T_cfg5_task_0081_user_0007_scene_0003_cfg_0005_104422070044#180", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Unfold the leg of the glasses (with one hand)'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005398/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 380, + 163 + ], + [ + 394, + 163 + ], + [ + 368, + 158 + ], + [ + 391, + 168 + ], + [ + 355, + 158 + ], + [ + 378, + 168 + ] + ], + [ + [ + 378, + 163 + ], + [ + 374, + 151 + ], + [ + 386, + 173 + ], + [ + 370, + 153 + ], + [ + 389, + 185 + ], + [ + 374, + 164 + ] + ], + [ + [ + 369, + 170 + ], + [ + 357, + 166 + ], + [ + 363, + 158 + ], + [ + 376, + 182 + ], + [ + 376, + 161 + ], + [ + 388, + 185 + ] + ], + [ + [ + 363, + 166 + ], + [ + 374, + 153 + ], + [ + 358, + 167 + ], + [ + 368, + 165 + ], + [ + 346, + 180 + ], + [ + 357, + 177 + ] + ] + ] + }, + { + "task_id": "0005402", + "id": "RH20T_cfg4_task_0002_user_0010_scene_0007_cfg_0004_036422060909#150", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pull out a napkin'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005402/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 240, + 200 + ], + [ + 250, + 225 + ], + [ + 268, + 193 + ], + [ + 211, + 208 + ], + [ + 259, + 169 + ], + [ + 203, + 183 + ] + ], + [ + [ + 253, + 205 + ], + [ + 260, + 176 + ], + [ + 271, + 199 + ], + [ + 233, + 211 + ], + [ + 266, + 226 + ], + [ + 228, + 241 + ] + ], + [ + [ + 262, + 219 + ], + [ + 292, + 223 + ], + [ + 274, + 201 + ], + [ + 249, + 239 + ], + [ + 246, + 198 + ], + [ + 219, + 234 + ] + ], + [ + [ + 212, + 207 + ], + [ + 219, + 204 + ], + [ + 241, + 194 + ], + [ + 183, + 221 + ], + [ + 235, + 198 + ], + [ + 173, + 226 + ] + ] + ] + }, + { + "task_id": "0005407", + "id": "RH20T_cfg1_task_0034_user_0015_scene_0008_cfg_0001_750612070851#802", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Stack the squares into a pyramid shape'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005407/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 347, + 288 + ], + [ + 349, + 271 + ], + [ + 318, + 314 + ], + [ + 374, + 264 + ], + [ + 319, + 328 + ], + [ + 371, + 280 + ] + ], + [ + [ + 360, + 323 + ], + [ + 367, + 315 + ], + [ + 332, + 294 + ], + [ + 389, + 353 + ], + [ + 322, + 300 + ], + [ + 383, + 360 + ] + ], + [ + [ + 315, + 295 + ], + [ + 291, + 269 + ], + [ + 342, + 274 + ], + [ + 285, + 317 + ], + [ + 368, + 299 + ], + [ + 311, + 346 + ] + ], + [ + [ + 344, + 293 + ], + [ + 372, + 327 + ], + [ + 361, + 294 + ], + [ + 324, + 292 + ], + [ + 337, + 266 + ], + [ + 297, + 259 + ] + ] + ] + }, + { + "task_id": "0005409", + "id": "RH20T_cfg5_task_0007_user_0010_scene_0007_cfg_0005_105422061350#282", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Hold a small block with the gripper and sweep it from left to right on the table'. We supply some images that include orange fork-like gripper patterns; these patterns represent the poses the gripper might need to take to pick up an object. Which of the images below contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005409/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 297, + 73 + ], + [ + 298, + 54 + ], + [ + 303, + 56 + ], + [ + 291, + 91 + ], + [ + 303, + 73 + ], + [ + 291, + 108 + ] + ], + [ + [ + 305, + 107 + ], + [ + 312, + 84 + ], + [ + 305, + 118 + ], + [ + 306, + 96 + ], + [ + 298, + 140 + ], + [ + 300, + 117 + ] + ], + [ + [ + 302, + 77 + ], + [ + 316, + 61 + ], + [ + 296, + 86 + ], + [ + 307, + 69 + ], + [ + 282, + 102 + ], + [ + 293, + 85 + ] + ], + [ + [ + 281, + 38 + ], + [ + 257, + 29 + ], + [ + 286, + 35 + ], + [ + 276, + 42 + ], + [ + 308, + 43 + ], + [ + 301, + 51 + ] + ] + ] + }, + { + "task_id": "0005411", + "id": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_104122060811#133", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005411/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 333, + 147 + ], + [ + 338, + 139 + ], + [ + 346, + 145 + ], + [ + 320, + 150 + ], + [ + 341, + 153 + ], + [ + 315, + 158 + ] + ], + [ + [ + 347, + 144 + ], + [ + 360, + 136 + ], + [ + 339, + 132 + ], + [ + 356, + 156 + ], + [ + 327, + 140 + ], + [ + 344, + 164 + ] + ], + [ + [ + 335, + 144 + ], + [ + 327, + 132 + ], + [ + 331, + 144 + ], + [ + 340, + 144 + ], + [ + 340, + 156 + ], + [ + 348, + 156 + ] + ], + [ + [ + 349, + 153 + ], + [ + 339, + 143 + ], + [ + 342, + 163 + ], + [ + 357, + 143 + ], + [ + 352, + 174 + ], + [ + 367, + 154 + ] + ] + ] + }, + { + "task_id": "0005412", + "id": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_104122060811#277", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005412/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 318, + 147 + ], + [ + 311, + 150 + ], + [ + 309, + 156 + ], + [ + 327, + 139 + ], + [ + 315, + 153 + ], + [ + 333, + 136 + ] + ], + [ + [ + 308, + 149 + ], + [ + 316, + 141 + ], + [ + 319, + 149 + ], + [ + 296, + 149 + ], + [ + 312, + 156 + ], + [ + 289, + 156 + ] + ], + [ + [ + 312, + 165 + ], + [ + 323, + 173 + ], + [ + 319, + 164 + ], + [ + 306, + 166 + ], + [ + 308, + 156 + ], + [ + 295, + 159 + ] + ], + [ + [ + 321, + 144 + ], + [ + 330, + 154 + ], + [ + 326, + 133 + ], + [ + 317, + 154 + ], + [ + 317, + 123 + ], + [ + 308, + 144 + ] + ] + ] + }, + { + "task_id": "0005413", + "id": "RH20T_cfg7_task_0073_user_0014_scene_0001_cfg_0007_104122060811#411", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005413/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 327, + 114 + ], + [ + 337, + 113 + ], + [ + 317, + 113 + ], + [ + 338, + 115 + ], + [ + 307, + 114 + ], + [ + 327, + 117 + ] + ], + [ + [ + 325, + 125 + ], + [ + 337, + 120 + ], + [ + 330, + 129 + ], + [ + 319, + 121 + ], + [ + 318, + 134 + ], + [ + 306, + 126 + ] + ], + [ + [ + 314, + 134 + ], + [ + 312, + 121 + ], + [ + 326, + 136 + ], + [ + 301, + 133 + ], + [ + 327, + 149 + ], + [ + 302, + 147 + ] + ], + [ + [ + 312, + 111 + ], + [ + 305, + 108 + ], + [ + 324, + 116 + ], + [ + 301, + 106 + ], + [ + 331, + 119 + ], + [ + 309, + 109 + ] + ] + ] + }, + { + "task_id": "0005416", + "id": "RH20T_cfg1_task_0064_user_0015_scene_0008_cfg_0001_038522062547#174", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the roll of paper down from the shelf'. In the images we provide, there are orange fork-like gripper patterns: they represent the possible poses the gripper needs to use to pick up an object. Which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005416/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 325, + 106 + ], + [ + 328, + 131 + ], + [ + 338, + 110 + ], + [ + 312, + 103 + ], + [ + 335, + 83 + ], + [ + 309, + 79 + ] + ], + [ + [ + 318, + 80 + ], + [ + 327, + 66 + ], + [ + 343, + 88 + ], + [ + 293, + 71 + ], + [ + 334, + 100 + ], + [ + 287, + 84 + ] + ], + [ + [ + 294, + 86 + ], + [ + 304, + 84 + ], + [ + 280, + 109 + ], + [ + 309, + 62 + ], + [ + 269, + 112 + ], + [ + 299, + 62 + ] + ], + [ + [ + 324, + 64 + ], + [ + 310, + 61 + ], + [ + 306, + 83 + ], + [ + 343, + 43 + ], + [ + 320, + 85 + ], + [ + 356, + 47 + ] + ] + ] + }, + { + "task_id": "0005418", + "id": "RH20T_cfg7_task_0029_user_0020_scene_0002_cfg_0007_104122061602#215", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Grasp the handle and open the drawer'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005418/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 429, + 84 + ], + [ + 433, + 109 + ], + [ + 412, + 85 + ], + [ + 449, + 83 + ], + [ + 408, + 61 + ], + [ + 446, + 56 + ] + ], + [ + [ + 417, + 93 + ], + [ + 419, + 90 + ], + [ + 442, + 83 + ], + [ + 392, + 102 + ], + [ + 439, + 86 + ], + [ + 392, + 104 + ] + ], + [ + [ + 433, + 100 + ], + [ + 437, + 101 + ], + [ + 445, + 78 + ], + [ + 421, + 121 + ], + [ + 441, + 75 + ], + [ + 416, + 120 + ] + ], + [ + [ + 404, + 81 + ], + [ + 416, + 103 + ], + [ + 394, + 92 + ], + [ + 414, + 70 + ], + [ + 382, + 72 + ], + [ + 401, + 48 + ] + ] + ] + }, + { + "task_id": "0005419", + "id": "RH20T_cfg6_task_0060_user_0014_scene_0006_cfg_0006_104122061018#132", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Turn the knob to decrease the volume of the speaker'. We offer some images featuring orange fork-like gripper patterns. These patterns stand for the poses the gripper might adopt to pick up an object\u2014so which of the following images contains the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005419/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 315, + 117 + ], + [ + 306, + 90 + ], + [ + 308, + 128 + ], + [ + 322, + 105 + ], + [ + 316, + 154 + ], + [ + 332, + 135 + ] + ], + [ + [ + 342, + 97 + ], + [ + 327, + 110 + ], + [ + 319, + 106 + ], + [ + 364, + 89 + ], + [ + 336, + 91 + ], + [ + 382, + 74 + ] + ], + [ + [ + 295, + 109 + ], + [ + 323, + 102 + ], + [ + 288, + 82 + ], + [ + 302, + 135 + ], + [ + 259, + 89 + ], + [ + 275, + 142 + ] + ], + [ + [ + 328, + 93 + ], + [ + 319, + 66 + ], + [ + 319, + 89 + ], + [ + 336, + 96 + ], + [ + 329, + 118 + ], + [ + 344, + 122 + ] + ] + ] + }, + { + "task_id": "0005420", + "id": "RH20T_cfg4_task_0035_user_0016_scene_0001_cfg_0004_104122062295#95", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Pick up one small block'. We present some images annotated with orange fork-like gripper patterns. These patterns correspond to the potential poses the gripper may adopt to grab an object\u2014so which image below has the right grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "D" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005420/concat.jpg" + ], + "gt": "D", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 154, + 191 + ], + [ + 187, + 201 + ], + [ + 176, + 157 + ], + [ + 132, + 224 + ], + [ + 140, + 143 + ], + [ + 94, + 215 + ] + ], + [ + [ + 180, + 207 + ], + [ + 212, + 221 + ], + [ + 216, + 208 + ], + [ + 148, + 206 + ], + [ + 179, + 190 + ], + [ + 109, + 190 + ] + ], + [ + [ + 135, + 212 + ], + [ + 131, + 176 + ], + [ + 118, + 247 + ], + [ + 154, + 174 + ], + [ + 122, + 289 + ], + [ + 161, + 215 + ] + ], + [ + [ + 146, + 205 + ], + [ + 148, + 167 + ], + [ + 114, + 184 + ], + [ + 181, + 227 + ], + [ + 114, + 218 + ], + [ + 178, + 261 + ] + ] + ] + }, + { + "task_id": "0005421", + "id": "RH20T_cfg4_task_0073_user_0015_scene_0007_cfg_0004_038522062288#97", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Put things in the drawer'. We provide some images that include orange fork-like gripper patterns; these patterns stand for the potential postures the gripper may use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "A" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005421/concat.jpg" + ], + "gt": "A", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 191, + 225 + ], + [ + 185, + 206 + ], + [ + 162, + 216 + ], + [ + 221, + 234 + ], + [ + 170, + 234 + ], + [ + 225, + 251 + ] + ], + [ + [ + 189, + 242 + ], + [ + 202, + 218 + ], + [ + 164, + 219 + ], + [ + 212, + 264 + ], + [ + 146, + 245 + ], + [ + 199, + 291 + ] + ], + [ + [ + 221, + 225 + ], + [ + 250, + 218 + ], + [ + 228, + 255 + ], + [ + 214, + 195 + ], + [ + 197, + 264 + ], + [ + 183, + 201 + ] + ], + [ + [ + 163, + 196 + ], + [ + 186, + 172 + ], + [ + 187, + 213 + ], + [ + 141, + 180 + ], + [ + 163, + 239 + ], + [ + 117, + 204 + ] + ] + ] + }, + { + "task_id": "0005422", + "id": "RH20T_cfg1_task_0110_user_0015_scene_0003_cfg_0001_039422060546#109", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take the pencil out from the pencil sharpener'. We provide some images that feature orange fork-like gripper patterns; these patterns stand for the potential poses the gripper needs to adopt to pick up an object. Which of the images below contains the correct grasping pose? The available choice is [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005422/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 314, + 167 + ], + [ + 341, + 179 + ], + [ + 319, + 162 + ], + [ + 309, + 172 + ], + [ + 289, + 149 + ], + [ + 282, + 161 + ] + ], + [ + [ + 311, + 128 + ], + [ + 310, + 107 + ], + [ + 300, + 151 + ], + [ + 322, + 107 + ], + [ + 303, + 170 + ], + [ + 323, + 126 + ] + ], + [ + [ + 306, + 88 + ], + [ + 334, + 96 + ], + [ + 291, + 91 + ], + [ + 318, + 86 + ], + [ + 263, + 83 + ], + [ + 292, + 79 + ] + ], + [ + [ + 322, + 156 + ], + [ + 336, + 166 + ], + [ + 307, + 182 + ], + [ + 337, + 130 + ], + [ + 295, + 172 + ], + [ + 324, + 123 + ] + ] + ] + }, + { + "task_id": "0005426", + "id": "RH20T_cfg5_task_0061_user_0007_scene_0008_cfg_0005_104122063678#240", + "task": "TrajChoice", + "conversations": [ + { + "from": "huamn", + "value": "\nThe robot task is 'Take everything out of the gift box'. Among the images we present, there are orange fork-like gripper patterns; they correspond to the potential poses the gripper needs to use to grab an object. Which image below has the correct grasping pose? The available choices are [A, B, C, D]. Please select one option only." + }, + { + "from": "gpt", + "value": "B" + } + ], + "images": [ + "Understanding/image/val/grasppose_choice/rh20t_0005426/concat.jpg" + ], + "gt": "B", + "source": "rh20t", + "origin_shape:": [ + 360, + 640 + ], + "smart_resize_shape": [ + 364, + 644 + ], + "poses": [ + [ + [ + 305, + 217 + ], + [ + 321, + 236 + ], + [ + 325, + 206 + ], + [ + 284, + 229 + ], + [ + 310, + 189 + ], + [ + 269, + 210 + ] + ], + [ + [ + 317, + 213 + ], + [ + 320, + 198 + ], + [ + 296, + 225 + ], + [ + 337, + 202 + ], + [ + 295, + 239 + ], + [ + 334, + 216 + ] + ], + [ + [ + 311, + 214 + ], + [ + 292, + 212 + ], + [ + 303, + 236 + ], + [ + 319, + 193 + ], + [ + 324, + 238 + ], + [ + 340, + 194 + ] + ], + [ + [ + 318, + 234 + ], + [ + 319, + 246 + ], + [ + 305, + 255 + ], + [ + 331, + 214 + ], + [ + 303, + 242 + ], + [ + 331, + 200 + ] + ] + ] + } +] \ No newline at end of file