file_name
stringlengths
31
31
im_path
stringlengths
46
46
caption
stringlengths
2
104
bbox
stringlengths
21
45
prompt
stringlengths
2.03k
2.52k
texts
stringlengths
2
61
sentence
stringlengths
42
524
outp_org
listlengths
4
4
formatted_sentence
stringlengths
37
71
modified_sentence
stringlengths
38
518
target
listlengths
4
4
COCO_train2014_000000534131.jpg
refcoco_images/COCO_train2014_000000534131.jpg
man farthest right with strange tie
[453.33,,39.75,554.1,,398.68]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Man', 'Tie']
Detected Man at location [372, 47, 467, 396] Detected man farthest right with strange tie at location [453, 40, 555, 398]
[ 453, 40, 555, 398 ]
Detected man at location [453, 40, 555, 398]
Detected Man at location [372, 47, 467, 396] Detected man at location [453, 40, 555, 398]
[ 453, 40, 555, 398 ]
COCO_train2014_000000518509.jpg
refcoco_images/COCO_train2014_000000518509.jpg
plate in front
[296.25,319.38,640.,,,471.88]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Plate']
Detected Plate at location [291, 316, 639, 479] Detected plate in front at location [292, 318, 639, 479]
[ 292, 318, 639, 479 ]
Detected plate at location [292, 318, 639, 479]
Detected Plate at location [291, 316, 639, 479] Detected plate at location [292, 318, 639, 479]
[ 292, 318, 639, 479 ]
COCO_train2014_000000527243.jpg
refcoco_images/COCO_train2014_000000527243.jpg
left piece of cake on the right
[271.82,,16.92,363.62,173.84]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Piece', 'Cake']
Detected Piece at location [270, 2, 489, 177] Detected Cake at location [270, 3, 488, 177] Detected left piece of cake on the right at location [46, 38, 233, 143]
[ 46, 38, 233, 143 ]
Detected piece of cake at location [46, 38, 233, 143]
Detected Piece at location [270, 2, 489, 177] Detected Cake at location [270, 3, 488, 177] Detected piece of cake at location [46, 38, 233, 143]
[ 46, 38, 233, 143 ]
COCO_train2014_000000567036.jpg
refcoco_images/COCO_train2014_000000567036.jpg
left teenage boy white shirt
[162.36,154.72,363.88,350.51]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Teenage boy', 'White shirt']
Detected Teenage boy at location [160, 153, 361, 350] Detected shirt at location [160, 230, 356, 341] Detected left teenage boy white shirt at location [163, 154, 362, 350]
[ 163, 154, 362, 350 ]
Detected teenage boy white shirt at location [163, 154, 362, 350]
Detected Teenage boy at location [160, 153, 361, 350] Detected shirt at location [160, 230, 356, 341] Detected teenage boy white shirt at location [163, 154, 362, 350]
[ 163, 154, 362, 350 ]
COCO_train2014_000000530684.jpg
refcoco_images/COCO_train2014_000000530684.jpg
A glass filled with liquid
[271.19,,,0.,,,409.25,178.97]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Glass', 'Liquid']
Detected Glass at location [269, 1, 409, 176] Detected liquid at location [270, 1, 408, 175] Detected A glass filled with liquid at location [272, 0, 410, 178]
[ 272, 0, 410, 178 ]
Detected glass at location [272, 0, 410, 178]
Detected Glass at location [269, 1, 409, 176] Detected liquid at location [270, 1, 408, 175] Detected glass at location [272, 0, 410, 178]
[ 272, 0, 410, 178 ]
COCO_train2014_000000573357.jpg
refcoco_images/COCO_train2014_000000573357.jpg
red coat behind center
[329.88,231.18,427.74,465.93]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Red coat']
Detected Red coat at location [108, 180, 370, 368] Detected red coat behind center at location [329, 231, 427, 466]
[ 329, 231, 427, 466 ]
Detected red coat at location [329, 231, 427, 466]
Detected Red coat at location [108, 180, 370, 368] Detected red coat at location [329, 231, 427, 466]
[ 329, 231, 427, 466 ]
COCO_train2014_000000529470.jpg
refcoco_images/COCO_train2014_000000529470.jpg
guy in white closest
[,,0.,,,,92.48,124.12,361.,,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Guy', 'White']
Detected Guy at location [420, 172, 528, 360] Detected guy in white closest at location [0, 93, 123, 360]
[ 0, 93, 123, 360 ]
Detected guy in white at location [0, 93, 123, 360]
Detected Guy at location [420, 172, 528, 360] Detected guy in white at location [0, 93, 123, 360]
[ 0, 93, 123, 360 ]
COCO_train2014_000000562124.jpg
refcoco_images/COCO_train2014_000000562124.jpg
tennis player on right
[408.67,,94.65,612.52,407.22]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Tennis player']
Detected Tennis player at location [96, 89, 294, 406] Detected tennis player on right at location [413, 92, 610, 404]
[ 413, 92, 610, 404 ]
Detected tennis player at location [413, 92, 610, 404]
Detected Tennis player at location [96, 89, 294, 406] Detected tennis player at location [413, 92, 610, 404]
[ 413, 92, 610, 404 ]
COCO_train2014_000000525232.jpg
refcoco_images/COCO_train2014_000000525232.jpg
Guy in foreground on phone
[233.76,126.8,,425.68,328.57]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Guy', 'Phone']
Detected Guy at location [290, 164, 546, 426] Detected Phone at location [388, 248, 445, 291] Detected Guy in foreground on phone at location [290, 165, 544, 426]
[ 290, 165, 544, 426 ]
Detected Guy at location [290, 165, 544, 426]
Detected Guy at location [290, 164, 546, 426] Detected Phone at location [388, 248, 445, 291] Detected Guy at location [290, 165, 544, 426]
[ 290, 165, 544, 426 ]
COCO_train2014_000000552234.jpg
refcoco_images/COCO_train2014_000000552234.jpg
player in dugout under bat
[273.87,295.27,398.37,470.79]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Player', 'Dugout', 'Bat']
Detected Player at location [34, 59, 367, 591] Detected Player at location [35, 107, 365, 589] Detected player in dugout under bat at location [274, 295, 397, 469]
[ 274, 295, 397, 469 ]
Detected player at location [274, 295, 397, 469]
Detected Player at location [34, 59, 367, 591] Detected Player at location [35, 107, 365, 589] Detected player at location [274, 295, 397, 469]
[ 274, 295, 397, 469 ]
COCO_train2014_000000554534.jpg
refcoco_images/COCO_train2014_000000554534.jpg
bear
[,29.09,,,0.86,479.14,640.,,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Bear']
Detected Bear at location [31, 0, 479, 639] Detected bear at location [30, 0, 479, 639]
[ 30, 0, 479, 639 ]
Detected bear at location [30, 0, 479, 639]
Detected Bear at location [31, 0, 479, 639] Detected bear at location [30, 0, 479, 639]
[ 30, 0, 479, 639 ]
COCO_train2014_000000519610.jpg
refcoco_images/COCO_train2014_000000519610.jpg
lesbian playing guitar
[,24.94,,41.56,294.23,640.,,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Lesbian', 'Guitar']
Detected Lesbian,Guitar at location [100, 290, 426, 639] Detected lesbian playing guitar at location [24, 36, 301, 639]
[ 24, 36, 301, 639 ]
Detected lesbian at location [24, 36, 301, 639]
Detected Lesbian,Guitar at location [100, 290, 426, 639] Detected lesbian at location [24, 36, 301, 639]
[ 24, 36, 301, 639 ]
COCO_train2014_000000556811.jpg
refcoco_images/COCO_train2014_000000556811.jpg
Giraffe on left
[,56.77,,24.18,416.32,286.41]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Giraffe']
Detected Giraffe at location [199, 36, 533, 384] Detected Giraffe on left at location [56, 25, 415, 286]
[ 56, 25, 415, 286 ]
Detected Giraffe at location [56, 25, 415, 286]
Detected Giraffe at location [199, 36, 533, 384] Detected Giraffe at location [56, 25, 415, 286]
[ 56, 25, 415, 286 ]
COCO_train2014_000000580625.jpg
refcoco_images/COCO_train2014_000000580625.jpg
red chair
[288.,,,155.87,441.81,283.87]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Red chair']
Detected Red chair at location [285, 154, 444, 286] Detected red chair at location [288, 157, 441, 282]
[ 288, 157, 441, 282 ]
Detected red chair at location [288, 157, 441, 282]
Detected Red chair at location [285, 154, 444, 286] Detected red chair at location [288, 157, 441, 282]
[ 288, 157, 441, 282 ]
COCO_train2014_000000554524.jpg
refcoco_images/COCO_train2014_000000554524.jpg
left whole pizza
[,99.56,,49.49,291.02,155.75]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Pizza']
Detected Pizza at location [100, 50, 481, 155] Detected left whole pizza at location [100, 50, 290, 156]
[ 100, 50, 290, 156 ]
Detected whole pizza at location [100, 50, 290, 156]
Detected Pizza at location [100, 50, 481, 155] Detected whole pizza at location [100, 50, 290, 156]
[ 100, 50, 290, 156 ]
COCO_train2014_000000543661.jpg
refcoco_images/COCO_train2014_000000543661.jpg
kid half seen
[441.55,112.88,609.7,,414.88]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Kid']
Detected Kid at location [226, 104, 489, 338] Detected kid half seen at location [443, 114, 609, 415]
[ 443, 114, 609, 415 ]
Detected kid at location [443, 114, 609, 415]
Detected Kid at location [226, 104, 489, 338] Detected kid at location [443, 114, 609, 415]
[ 443, 114, 609, 415 ]
COCO_train2014_000000552291.jpg
refcoco_images/COCO_train2014_000000552291.jpg
guy in right corner, partial
[508.27,173.93,638.72,423.24]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Guy']
Detected Guy at location [283, 176, 437, 429] Detected guy in right corner, partial at location [507, 174, 639, 429]
[ 507, 174, 639, 429 ]
Detected guy at location [507, 174, 639, 429]
Detected Guy at location [283, 176, 437, 429] Detected guy at location [507, 174, 639, 429]
[ 507, 174, 639, 429 ]
COCO_train2014_000000560909.jpg
refcoco_images/COCO_train2014_000000560909.jpg
dog in purse
[220.18,161.38,413.67,353.2,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Dog', 'Purse']
Detected Dog at location [218, 156, 639, 352] Detected Dog at location [218, 159, 415, 353] Detected Dog at location [403, 160, 639, 317] Detected Purse at location [156, 227, 336, 379] Detected dog in purse at location [219, 162, 414, 353]
[ 219, 162, 414, 353 ]
Detected dog at location [219, 162, 414, 353]
Detected Dog at location [218, 156, 639, 352] Detected Dog at location [218, 159, 415, 353] Detected Dog at location [403, 160, 639, 317] Detected Purse at location [156, 227, 336, 379] Detected dog at location [219, 162, 414, 353]
[ 219, 162, 414, 353 ]
COCO_train2014_000000539631.jpg
refcoco_images/COCO_train2014_000000539631.jpg
white hair in foreground
[367.62,312.34,620.99,405.39]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['White hair']
Detected White hair at location [373, 313, 619, 409] Detected white hair in foreground at location [371, 312, 619, 405]
[ 371, 312, 619, 405 ]
Detected white hair at location [371, 312, 619, 405]
Detected White hair at location [373, 313, 619, 409] Detected white hair at location [371, 312, 619, 405]
[ 371, 312, 619, 405 ]
COCO_train2014_000000579165.jpg
refcoco_images/COCO_train2014_000000579165.jpg
bowl/spoon on the left with the white stuff in it
[,,0.,,,176.24,302.64,480.,,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Bowl', 'Spoon', 'Stuff']
Detected Bowl at location [2, 171, 306, 479] Detected Bowl at location [219, 27, 537, 215] Detected Spoon at location [2, 131, 241, 337] Detected bowl/spoon on the left with the white stuff in it at location [0, 132, 240, 335]
[ 0, 132, 240, 335 ]
Detected bowl at location [0, 132, 240, 335]
Detected Bowl at location [2, 171, 306, 479] Detected Bowl at location [219, 27, 537, 215] Detected Spoon at location [2, 131, 241, 337] Detected bowl at location [0, 132, 240, 335]
[ 0, 132, 240, 335 ]
COCO_train2014_000000560388.jpg
refcoco_images/COCO_train2014_000000560388.jpg
pink
[355.96,143.13,528.54,473.2,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Pink']
Detected Pink at location [352, 139, 527, 479] Detected pink at location [356, 143, 527, 479]
[ 356, 143, 527, 479 ]
Detected pink at location [356, 143, 527, 479]
Detected Pink at location [352, 139, 527, 479] Detected pink at location [356, 143, 527, 479]
[ 356, 143, 527, 479 ]
COCO_train2014_000000578718.jpg
refcoco_images/COCO_train2014_000000578718.jpg
first arm closest to us
[264.63,316.5,,509.12,632.9,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Arm']
Detected arm at location [262, 312, 511, 639] Detected arm at location [263, 313, 511, 604] Detected first arm closest to us at location [264, 260, 511, 639]
[ 264, 260, 511, 639 ]
Detected arm at location [264, 260, 511, 639]
Detected arm at location [262, 312, 511, 639] Detected arm at location [263, 313, 511, 604] Detected arm at location [264, 260, 511, 639]
[ 264, 260, 511, 639 ]
COCO_train2014_000000561529.jpg
refcoco_images/COCO_train2014_000000561529.jpg
teddy bear that is closer to the screen and is to the right
[359.72,157.49,602.53,474.49]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Teddy bear']
Detected Teddy bear at location [359, 157, 600, 479] Detected teddy bear that is closer to the screen and is to the right at location [360, 157, 601, 479]
[ 360, 157, 601, 479 ]
Detected teddy bear at location [360, 157, 601, 479]
Detected Teddy bear at location [359, 157, 600, 479] Detected teddy bear at location [360, 157, 601, 479]
[ 360, 157, 601, 479 ]
COCO_train2014_000000522256.jpg
refcoco_images/COCO_train2014_000000522256.jpg
Girl bottom right
[376.53,122.77,503.55,481.28]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Girl']
Detected Girl at location [84, 61, 299, 425] Detected Girl bottom right at location [375, 123, 508, 481]
[ 375, 123, 508, 481 ]
Detected Girl at location [375, 123, 508, 481]
Detected Girl at location [84, 61, 299, 425] Detected Girl at location [375, 123, 508, 481]
[ 375, 123, 508, 481 ]
COCO_train2014_000000533599.jpg
refcoco_images/COCO_train2014_000000533599.jpg
far right horse
[331.27,102.92,639.42,353.3,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Horse']
Detected Horse at location [85, 100, 639, 359] Detected Horse at location [329, 102, 639, 359] Detected far right horse at location [332, 103, 639, 359]
[ 332, 103, 639, 359 ]
Detected horse at location [332, 103, 639, 359]
Detected Horse at location [85, 100, 639, 359] Detected Horse at location [329, 102, 639, 359] Detected horse at location [332, 103, 639, 359]
[ 332, 103, 639, 359 ]
COCO_train2014_000000559935.jpg
refcoco_images/COCO_train2014_000000559935.jpg
black bag on top
[146.46,242.58,275.64,375.8,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Bag']
Detected Bag at location [81, 379, 257, 579] Detected black bag on top at location [147, 243, 274, 375]
[ 147, 243, 274, 375 ]
Detected black bag at location [147, 243, 274, 375]
Detected Bag at location [81, 379, 257, 579] Detected black bag at location [147, 243, 274, 375]
[ 147, 243, 274, 375 ]
COCO_train2014_000000581227.jpg
refcoco_images/COCO_train2014_000000581227.jpg
Far left player
[,42.46,120.08,125.9,,333.,,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Player']
Detected Player at location [262, 10, 426, 333] Detected Far left player at location [42, 120, 125, 333]
[ 42, 120, 125, 333 ]
Detected player at location [42, 120, 125, 333]
Detected Player at location [262, 10, 426, 333] Detected player at location [42, 120, 125, 333]
[ 42, 120, 125, 333 ]
COCO_train2014_000000578950.jpg
refcoco_images/COCO_train2014_000000578950.jpg
Dumpling in front on right
[,52.78,177.52,540.23,427.,,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Dumpling']
Detected Dumpling at location [0, 0, 639, 426] Detected Dumpling in front on right at location [52, 178, 540, 426]
[ 52, 178, 540, 426 ]
Detected Dumpling at location [52, 178, 540, 426]
Detected Dumpling at location [0, 0, 639, 426] Detected Dumpling at location [52, 178, 540, 426]
[ 52, 178, 540, 426 ]
COCO_train2014_000000540735.jpg
refcoco_images/COCO_train2014_000000540735.jpg
dark brown dow in front between two black cows
[130.72,157.6,,306.99,348.32]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Dow', 'Cow']
Detected Cow at location [128, 157, 307, 349] Detected Cow at location [369, 155, 512, 357] Detected dark brown dow in front between two black cows at location [130, 158, 305, 347]
[ 130, 158, 305, 347 ]
Detected dark brown dow at location [130, 158, 305, 347]
Detected Cow at location [128, 157, 307, 349] Detected Cow at location [369, 155, 512, 357] Detected dark brown dow at location [130, 158, 305, 347]
[ 130, 158, 305, 347 ]
COCO_train2014_000000530293.jpg
refcoco_images/COCO_train2014_000000530293.jpg
girl left not with ball
[163.31,,49.27,283.74,358.56]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Girl', 'Ball']
Detected Girl at location [160, 46, 286, 364] Detected Girl at location [257, 71, 423, 300] Detected Ball at location [258, 267, 293, 299] Detected girl left not with ball at location [162, 48, 284, 361]
[ 162, 48, 284, 361 ]
Detected girl at location [162, 48, 284, 361]
Detected Girl at location [160, 46, 286, 364] Detected Girl at location [257, 71, 423, 300] Detected Ball at location [258, 267, 293, 299] Detected girl at location [162, 48, 284, 361]
[ 162, 48, 284, 361 ]
COCO_train2014_000000542643.jpg
refcoco_images/COCO_train2014_000000542643.jpg
ctacher
[383.41,111.86,472.17,363.56]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['ctacher']
Detected ctacher at location [299, 138, 386, 379] Detected ctacher at location [384, 113, 470, 364]
[ 384, 113, 470, 364 ]
Detected catcher at location [384, 113, 470, 364]
Detected ctacher at location [299, 138, 386, 379] Detected catcher at location [384, 113, 470, 364]
[ 384, 113, 470, 364 ]
COCO_train2014_000000532565.jpg
refcoco_images/COCO_train2014_000000532565.jpg
boy
[169.38,,38.17,427.,,,640.,,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Boy']
Detected Boy at location [166, 39, 426, 639] Detected boy at location [167, 39, 426, 639]
[ 167, 39, 426, 639 ]
Detected boy at location [167, 39, 426, 639]
Detected Boy at location [166, 39, 426, 639] Detected boy at location [167, 39, 426, 639]
[ 167, 39, 426, 639 ]
COCO_train2014_000000568840.jpg
refcoco_images/COCO_train2014_000000568840.jpg
cup in back on left
[,71.87,,,0.,,,193.8,,134.76]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Cup']
Detected Cup at location [406, 116, 639, 409] Detected cup in back on left at location [71, 0, 192, 134]
[ 71, 0, 192, 134 ]
Detected cup at location [71, 0, 192, 134]
Detected Cup at location [406, 116, 639, 409] Detected cup at location [71, 0, 192, 134]
[ 71, 0, 192, 134 ]
COCO_train2014_000000532504.jpg
refcoco_images/COCO_train2014_000000532504.jpg
long bench left
[,,0.,,,382.32,191.28,632.57]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Bench']
Detected Bench at location [0, 381, 189, 639] Detected Bench at location [244, 384, 426, 639] Detected long bench left at location [0, 383, 190, 639]
[ 0, 383, 190, 639 ]
Detected long bench at location [0, 383, 190, 639]
Detected Bench at location [0, 381, 189, 639] Detected Bench at location [244, 384, 426, 639] Detected long bench at location [0, 383, 190, 639]
[ 0, 383, 190, 639 ]
COCO_train2014_000000551125.jpg
refcoco_images/COCO_train2014_000000551125.jpg
right toilet
[463.82,174.74,609.44,385.08]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Toilet']
Detected Toilet at location [68, 166, 612, 390] Detected Toilet at location [464, 174, 609, 388] Detected Toilet at location [70, 174, 208, 390] Detected Toilet at location [275, 174, 398, 388] Detected right toilet at location [463, 174, 608, 384]
[ 463, 174, 608, 384 ]
Detected toilet at location [463, 174, 608, 384]
Detected Toilet at location [68, 166, 612, 390] Detected Toilet at location [464, 174, 609, 388] Detected Toilet at location [70, 174, 208, 390] Detected Toilet at location [275, 174, 398, 388] Detected toilet at location [463, 174, 608, 384]
[ 463, 174, 608, 384 ]
COCO_train2014_000000569851.jpg
refcoco_images/COCO_train2014_000000569851.jpg
fruit next to the bottom right apple
[342.,,343.1,518.,,422.1]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Fruit', 'Apple']
Detected Fruit at location [0, 42, 639, 422] Detected fruit next to the bottom right apple at location [343, 344, 517, 426]
[ 343, 344, 517, 426 ]
Detected apple at location [343, 344, 517, 426]
Detected Fruit at location [0, 42, 639, 422] Detected apple at location [343, 344, 517, 426]
[ 343, 344, 517, 426 ]
COCO_train2014_000000544449.jpg
refcoco_images/COCO_train2014_000000544449.jpg
middle column directly to right of orange
[470.23,,78.99,621.78,257.53]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Column', 'Orange']
Detected column at location [280, 71, 556, 333] Detected Orange at location [281, 72, 556, 333] Detected middle column directly to right of orange at location [473, 80, 607, 262]
[ 473, 80, 607, 262 ]
Detected middle column at location [473, 80, 607, 262]
Detected column at location [280, 71, 556, 333] Detected Orange at location [281, 72, 556, 333] Detected middle column at location [473, 80, 607, 262]
[ 473, 80, 607, 262 ]
COCO_train2014_000000566700.jpg
refcoco_images/COCO_train2014_000000566700.jpg
with red dot onn it
[436.3,,,97.03,594.54,300.91]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Dot']
Detected Dot at location [0, 56, 639, 335] Detected with red dot onn it at location [437, 96, 594, 299]
[ 437, 96, 594, 299 ]
Detected red dot at location [437, 96, 594, 299]
Detected Dot at location [0, 56, 639, 335] Detected red dot at location [437, 96, 594, 299]
[ 437, 96, 594, 299 ]
COCO_train2014_000000562147.jpg
refcoco_images/COCO_train2014_000000562147.jpg
blue clothing
[356.89,,12.44,584.46,473.78]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Blue clothing']
Detected Blue clothing at location [359, 134, 587, 479] Detected blue clothing at location [355, 11, 585, 479]
[ 355, 11, 585, 479 ]
Detected blue clothing at location [355, 11, 585, 479]
Detected Blue clothing at location [359, 134, 587, 479] Detected blue clothing at location [355, 11, 585, 479]
[ 355, 11, 585, 479 ]
COCO_train2014_000000537924.jpg
refcoco_images/COCO_train2014_000000537924.jpg
woman looking at camera while eating
[128.36,,,0.,,,456.27,473.53]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Woman', 'Camera', 'Food']
Detected Woman at location [121, 5, 457, 479] Detected woman looking at camera while eating at location [127, 4, 456, 479]
[ 127, 4, 456, 479 ]
Detected woman at location [127, 4, 456, 479]
Detected Woman at location [121, 5, 457, 479] Detected woman at location [127, 4, 456, 479]
[ 127, 4, 456, 479 ]
COCO_train2014_000000554617.jpg
refcoco_images/COCO_train2014_000000554617.jpg
head, by the wooden bar, on the left
[,25.79,236.04,228.28,413.69]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Head', 'Wooden bar']
Detected Head at location [155, 24, 616, 404] Detected bar at location [219, 302, 471, 405] Detected head, by the wooden bar, on the left at location [27, 237, 229, 416]
[ 27, 237, 229, 416 ]
Detected head at location [27, 237, 229, 416]
Detected Head at location [155, 24, 616, 404] Detected bar at location [219, 302, 471, 405] Detected head at location [27, 237, 229, 416]
[ 27, 237, 229, 416 ]
COCO_train2014_000000542718.jpg
refcoco_images/COCO_train2014_000000542718.jpg
left some person
[,,0.,,,,99.24,106.43,339.42]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Person']
Detected Person at location [6, 72, 319, 639] Detected left some person at location [0, 99, 105, 338]
[ 0, 99, 105, 338 ]
Detected person at location [0, 99, 105, 338]
Detected Person at location [6, 72, 319, 639] Detected person at location [0, 99, 105, 338]
[ 0, 99, 105, 338 ]
COCO_train2014_000000542129.jpg
refcoco_images/COCO_train2014_000000542129.jpg
kid right
[148.13,,68.59,480.,,,520.19]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Kid']
Detected Kid at location [145, 75, 479, 517] Detected kid right at location [146, 69, 479, 519]
[ 146, 69, 479, 519 ]
Detected kid at location [146, 69, 479, 519]
Detected Kid at location [145, 75, 479, 517] Detected kid at location [146, 69, 479, 519]
[ 146, 69, 479, 519 ]
COCO_train2014_000000561875.jpg
refcoco_images/COCO_train2014_000000561875.jpg
Yellow and black vase
[376.51,155.39,542.46,365.81]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Vase']
Detected Vase at location [206, 136, 421, 423] Detected Yellow and black vase at location [380, 0, 543, 363]
[ 380, 0, 543, 363 ]
Detected yellow and black vase at location [380, 0, 543, 363]
Detected Vase at location [206, 136, 421, 423] Detected yellow and black vase at location [380, 0, 543, 363]
[ 380, 0, 543, 363 ]
COCO_train2014_000000532188.jpg
refcoco_images/COCO_train2014_000000532188.jpg
boy with hand over his mouth
[261.75,360.99,371.06,634.25]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Boy', 'Hand']
Detected Boy at location [264, 362, 366, 639] Detected boy with hand over his mouth at location [260, 360, 372, 639]
[ 260, 360, 372, 639 ]
Detected boy at location [260, 360, 372, 639]
Detected Boy at location [264, 362, 366, 639] Detected boy at location [260, 360, 372, 639]
[ 260, 360, 372, 639 ]
COCO_train2014_000000569878.jpg
refcoco_images/COCO_train2014_000000569878.jpg
black shirt lady right
[486.49,,76.76,640.,,,422.2,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Black shirt', 'Lady']
Detected Black shirt at location [455, 77, 639, 425] Detected Lady at location [455, 77, 640, 425] Detected black shirt lady right at location [453, 78, 639, 426]
[ 453, 78, 639, 426 ]
Detected black shirt at location [453, 78, 639, 426]
Detected Black shirt at location [455, 77, 639, 425] Detected Lady at location [455, 77, 640, 425] Detected black shirt at location [453, 78, 639, 426]
[ 453, 78, 639, 426 ]
COCO_train2014_000000519723.jpg
refcoco_images/COCO_train2014_000000519723.jpg
black chair, bottom right of photo
[432.8,,330.65,629.85,480.,,]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Black chair']
Detected Black chair at location [219, 244, 390, 479] Detected black chair, bottom right of photo at location [434, 331, 629, 479]
[ 434, 331, 629, 479 ]
Detected black chair at location [434, 331, 629, 479]
Detected Black chair at location [219, 244, 390, 479] Detected black chair at location [434, 331, 629, 479]
[ 434, 331, 629, 479 ]
COCO_train2014_000000544570.jpg
refcoco_images/COCO_train2014_000000544570.jpg
donut with icing
[417.66,290.49,629.61,472.52]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Donut', 'Icing']
Detected Donut at location [6, 44, 639, 479] Detected donut with icing at location [416, 290, 629, 479]
[ 416, 290, 629, 479 ]
Detected donut at location [416, 290, 629, 479]
Detected Donut at location [6, 44, 639, 479] Detected donut at location [416, 290, 629, 479]
[ 416, 290, 629, 479 ]
COCO_train2014_000000579165.jpg
refcoco_images/COCO_train2014_000000579165.jpg
bowl with broccoli
[220.62,,29.05,536.98,214.65]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Bowl', 'Broccoli']
Detected Bowl at location [219, 27, 537, 215] Detected broccoli at location [336, 82, 481, 178] Detected bowl with broccoli at location [220, 29, 536, 216]
[ 220, 29, 536, 216 ]
Detected bowl at location [220, 29, 536, 216]
Detected Bowl at location [219, 27, 537, 215] Detected broccoli at location [336, 82, 481, 178] Detected bowl at location [220, 29, 536, 216]
[ 220, 29, 536, 216 ]
COCO_train2014_000000546723.jpg
refcoco_images/COCO_train2014_000000546723.jpg
3rd sheep from the left
[181.18,110.16,283.59,293.49]
<|im_start|>system You are a coordinate selection system. Your task is to: 1. Parse the given bounding box coordinates 2. For left/right selection: - Compare the starting x-coordinates (first number in each box) - Left means smaller x-coordinate - Right means larger x-coordinate 3. For top/bottom selection: ...
['Sheep']
Detected Sheep at location [25, 120, 639, 408] Detected 3rd sheep from the left at location [331, 149, 453, 383]
[ 331, 149, 453, 383 ]
Detected sheep at location [331, 149, 453, 383]
Detected Sheep at location [25, 120, 639, 408] Detected sheep at location [331, 149, 453, 383]
[ 331, 149, 453, 383 ]