{"image": ["media/000212_54326130.jpg"], "conversations": [{"from": "human", "value": "<image>\nGive bounding boxes for each: The two athletes who are closest to the camera and are in confrontation."}, {"from": "gpt", "value": "<ref>The two athletes who are closest to the camera and are in confrontation</ref><box>[<303><115><873><915>]</box><ref>The two athletes who are closest to the camera and are in confrontation</ref><box>[<28><133><441><951>]</box>"}]}
{"image": ["media/006973_235583703.jpg"], "conversations": [{"from": "human", "value": "<image>\nFind and box every occurrence of: the people on both sides of the lady in the red vest in the middle.\nExpected output: <ref>object name</ref><box>[[x1, y1, x2, y2], [x1, y1, x2, y2], ...]</box> for each category."}, {"from": "gpt", "value": "<ref>the people on both sides of the lady in the red vest in the middle</ref><box>[<555><268><847><1000>, <71><106><380><1000>]</box>"}]}
{"image": ["media/007310_322854542.jpg"], "conversations": [{"from": "human", "value": "<image>\nOutline all regions of: the first person wearing gloves to the left of the first person wearing black clothes when counting from left to right."}, {"from": "gpt", "value": "<ref>the first person wearing gloves to the left of the first person wearing black clothes when counting from left to right</ref><box>[<344><134><470><841>]</box>"}]}
{"image": ["media/010016_443816118.jpg"], "conversations": [{"from": "human", "value": "<image>\nReturn box coordinates for all: the person wearing a white suit."}, {"from": "gpt", "value": "<ref>the person wearing a white suit</ref><box>[<466><385><527><733>]</box>"}]}
{"image": ["media/010365_533976280.jpg"], "conversations": [{"from": "human", "value": "<image>\nLocate all regions matching: the person wearing khaki clothes."}, {"from": "gpt", "value": "<ref>the person wearing khaki clothes</ref><box>[<277><275><451><499>, <257><621><438><992>]</box>"}]}
{"image": ["media/010587_592424752.jpg"], "conversations": [{"from": "human", "value": "<image>\nReturn bounding box coordinates for all: the man.\nCoordinates are [x1, y1, x2, y2], where (x1, y1) is top-left and (x2, y2) is bottom-right."}, {"from": "gpt", "value": "<ref>the man</ref><box>[<321><84><500><651>, <754><424><821><562>]</box>"}]}
{"image": ["media/011708_884609126.jpg"], "conversations": [{"from": "human", "value": "<image>\nGive bounding boxes for each: the person sitting with crossed legs."}, {"from": "gpt", "value": "<ref>the person sitting with crossed legs</ref><box>[<85><257><279><999>]</box><ref>the person sitting with crossed legs</ref><box>[<373><504><497><1000>]</box>"}]}
{"image": ["media/013035_1224795184.jpg"], "conversations": [{"from": "human", "value": "<image>\nReturn bounding box coordinates for all: the person wearing a gray flat cap."}, {"from": "gpt", "value": "<ref>the person wearing a gray flat cap</ref><box>[<525><159><970><760>]</box>"}]}
{"image": ["media/017261_2312423734.jpg"], "conversations": [{"from": "human", "value": "<image>\nLocate all regions matching: the person wearing a cap with a visor."}, {"from": "gpt", "value": "<ref>the person wearing a cap with a visor</ref><box>[<0><68><312><894>]</box>"}]}
{"image": ["media/017486_2370828260.jpg"], "conversations": [{"from": "human", "value": "<image>\nDraw boxes around all: the person wearing a blue sweater."}, {"from": "gpt", "value": "<ref>the person wearing a blue sweater</ref><box>[<644><407><864><999>]</box>"}]}
{"image": ["media/019234_2821283901.jpg"], "conversations": [{"from": "human", "value": "<image>\nDraw boxes around all: the person with long golden hair."}, {"from": "gpt", "value": "<ref>the person with long golden hair</ref><box>[<392><356><578><946>]</box>"}]}
{"image": ["media/019558_2904036473.jpg"], "conversations": [{"from": "human", "value": "<image>\nReturn bounding box coordinates for all: the person wearing a blue-purple tie."}, {"from": "gpt", "value": "<ref>the person wearing a blue-purple tie</ref><box>[<758><184><902><837>]</box>"}]}
{"image": ["media/022293_3608925878.jpg"], "conversations": [{"from": "human", "value": "<image>\nDraw boxes around all: the bridegroom."}, {"from": "gpt", "value": "<ref>the bridegroom</ref><box>[<244><160><620><989>]</box>"}]}
{"image": ["media/024504_4181296290.jpg"], "conversations": [{"from": "human", "value": "<image>\nFind and box every: the person wearing a pink short-sleeved shirt."}, {"from": "gpt", "value": "<ref>the person wearing a pink short-sleeved shirt</ref><box>[<172><21><268><195>]</box>"}]}
{"image": ["media/028492_508978812.jpg"], "conversations": [{"from": "human", "value": "<image>\nFind and box every: the person not wearing a blue hoodie."}, {"from": "gpt", "value": "<ref>the person not wearing a blue hoodie</ref><box>[<58><165><301><1000>]</box><ref>the person not wearing a blue hoodie</ref><box>[<454><204><686><1000>]</box><ref>the person not wearing a blue hoodie</ref><box>[<502><266><1000><1000>]</box>"}]}
{"image": ["media/029735_826290922.jpg"], "conversations": [{"from": "human", "value": "<image>\nReturn bounding box coordinates for all: the person not carrying a red rectangular bag."}, {"from": "gpt", "value": "<ref>the person not carrying a red rectangular bag</ref><box>[<209><118><513><972>, <751><234><872><476>, <829><235><987><445>]</box>"}]}
{"image": ["media/032806_182900778.jpg"], "conversations": [{"from": "human", "value": "<image>\nLocate all regions matching: the first person wearing red and white clothes all over from left to right.\nExpected output: <ref>object name</ref><box>[[x1, y1, x2, y2], [x1, y1, x2, y2], ...]</box> for each category."}, {"from": "gpt", "value": "<ref>the first person wearing red and white clothes all over from left to right</ref><box>[<445><265><534><449>]</box>"}]}
{"image": ["media/033270_302093892.jpg"], "conversations": [{"from": "human", "value": "<image>\nDraw bounding boxes around all: the person sitting on the square cabinet."}, {"from": "gpt", "value": "<ref>the person sitting on the square cabinet</ref><box>[<625><198><927><838>]</box>"}]}
{"image": ["media/033271_302784110.jpg"], "conversations": [{"from": "human", "value": "<image>\nDetect all instances of: the person wearing glasses in the middle and return their bounding boxes."}, {"from": "gpt", "value": "<ref>the person wearing glasses in the middle</ref><box>[<282><30><781><1000>]</box>"}]}
{"image": ["media/035348_134595059.jpg"], "conversations": [{"from": "human", "value": "<image>\nDetect all instances of: Miguel Cabrera and return their bounding boxes."}, {"from": "gpt", "value": "<ref>Miguel Cabrera</ref><box>[<302><31><654><1000>]</box>"}]}