Dataset Preview
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
Job manager crashed while running this job (missing heartbeats).
Error code: JobManagerCrashedError
Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
image
string | id
string | height
int64 | width
int64 | conversations
list |
|---|---|---|---|---|
coco/train2017/000000272026.jpg
|
00000001
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nA <ref>woman</ref><box>[[67, 311, 639, 874]]</box> is seated at the forefront of the image.\n\nLayer 2: \nBeside her to one side, another <ref>woman</ref><box>[[647, 313, 999, 874]]</box> is present, <pred>sitting at</pred><box>[[647, 313, 999, 874]]</box><box>[[351, 490, 821, 836]]</box> the <ref>dining table</ref><box>[[351, 490, 821, 836]]</box>. A <ref>wall</ref><box>[[0, 124, 819, 874]]</box> runs behind them, and a <ref>chair</ref><box>[[29, 745, 96, 874]]</box> is on the side.\n\nLayer 3: \nIn the background, another <ref>woman</ref><box>[[547, 291, 786, 646]]</box> <pred>is sitting at</pred><box>[[547, 291, 786, 646]]</box><box>[[351, 490, 821, 836]]</box> the <ref>dining table</ref><box>[[351, 490, 821, 836]]</box>. Various <ref>drinks</ref> like <ref>bottles</ref><box>[[552, 574, 580, 666], [399, 522, 449, 667]]</box> and <ref>cups</ref><box>[[519, 627, 561, 679]]</box> are arranged on the <ref>table</ref><box>[[351, 490, 821, 836]]</box>.\n\nLayer 4: \nFurther back, more <ref>cups</ref><box>[[583, 656, 666, 760]]</box> can be seen <pred>on</pred><box>[[583, 656, 666, 760]]</box><box>[[351, 490, 821, 836]]</box> the table.\n\nLayer 5: \nIn the distance, additional <ref>cups</ref><box>[[460, 595, 518, 664]]</box> are placed <pred>on</pred><box>[[460, 595, 518, 664]]</box><box>[[351, 490, 821, 836]]</box> the dining table."
}
] |
coco/train2017/000000422170.jpg
|
00000002
| 548
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nThere are <ref>people</ref><box>[[636, 72, 778, 345]]</box> and another <ref>person</ref><box>[[0, 77, 224, 633]]</box> who appear to be <pred>walking on</pred><box>[[636, 72, 778, 345], [0, 77, 224, 633]]</box> the <ref>stairs</ref><box>[[0, 71, 999, 927]]</box>. In this scene, the people add dynamism, moving across the expanse of the <ref>stairs</ref><box>[[0, 71, 999, 927]]</box>, hinting at transition or arrival. \n\nLayer 2: \nA <ref>bottle</ref><box>[[623, 783, 705, 808]]</box> rests on the <ref>stairs</ref><box>[[0, 71, 999, 927]]</box>. Another <ref>person</ref><box>[[75, 475, 443, 880]]</box> is visible, possibly <pred>sitting on</pred><box>[[75, 475, 443, 880]]</box><box>[[0, 71, 999, 927]]</box> the <ref>stairs</ref><box>[[0, 71, 999, 927]]</box>.\n\nLayer 3: \nA <ref>person</ref><box>[[694, 172, 999, 679]]</box> is present, with a <ref>book</ref><box>[[666, 422, 803, 483]]</box> held in their hands.\n\nLayer 4: \nTo the side, there's a <ref>handbag</ref><box>[[229, 806, 398, 927]]</box>, possibly belonging to one of the seated people. \n\nLayer 5: \nAnother <ref>person</ref><box>[[0, 299, 43, 459]]</box> is further down the <ref>stairs</ref><box>[[0, 71, 999, 927]]</box>, adding a sense of depth and continuity to the scene."
}
] |
coco/train2017/000000440284.jpg
|
00000003
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the forefront, there is a <ref>backpack</ref><box>[[714, 274, 828, 428]]</box> being <pred>carried by</pred> a <ref>man</ref><box>[[694, 196, 822, 661]]</box>. Nearby, a <ref>bicycle</ref><box>[[416, 409, 586, 736]]</box> is visible <pred>touching</pred> by the same <ref>man</ref><box>[[694, 196, 822, 661]]</box>. Another <ref>bicycle</ref><box>[[301, 416, 460, 754]]</box> is positioned <pred>beside</pred><box>[[125, 415, 349, 839]]</box>. Underneath, the <ref>pavement</ref><box>[[0, 406, 999, 874]]</box> stretches across the scene.\n\nLayer 2: A <ref>man</ref><box>[[694, 196, 822, 661]]</box> stands next to a <ref>bicycle</ref><box>[[607, 353, 797, 721]]</box>, while another <ref>man</ref><box>[[469, 208, 615, 669]]</box> is contact with a <ref>bicycle</ref><box>[[416, 409, 586, 736]]</box> and a <ref>pavement</ref><box>[[0, 406, 999, 874]]</box> is beneath them.\n\nLayer 3: In the background, a <ref>tree</ref><box>[[555, 124, 999, 571]]</box> is seen beyond the foreground activities.\n\nLayer 4: Higher above, the <ref>roof</ref><box>[[371, 124, 999, 177]]</box> of a nearby structure caps the distant backdrop.\n\nLayer 5: Further away, a <ref>man</ref><box>[[171, 168, 343, 772]]</box> is also <pred>carrying</pred> a <ref>backpack</ref><box>[[209, 256, 226, 386]]</box>. A <ref>bicycle</ref><box>[[125, 415, 349, 839]]</box> and <ref>pavement</ref><box>[[0, 406, 999, 874]]</box> complete the layered layout.\n\nLayer 6: Finally, in the far distance, a <ref>building</ref><box>[[0, 124, 999, 350]]</box> stands, part of the more extended background setting."
}
] |
coco/train2017/000000257958.jpg
|
00000004
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nThe closest to the viewer, we see a <ref>giraffe</ref><box>[[718, 364, 835, 793]]</box>, which is beside a large <ref>rock</ref><box>[[0, 294, 999, 836]]</box>. Additionally, there is a <ref>tree</ref><box>[[0, 124, 999, 552]]</box> that stands in front of this giraffe and beside the rock.\n\nLayer 2: \nIn this layer, another giraffe is eating from the <ref>grass</ref><box>[[479, 364, 789, 441]]</box>, which spans across the mid-section of the image, creating a natural interaction between the layers.\n\nLayer 3: \nAnother <ref>giraffe</ref><box>[[474, 327, 694, 808]]</box> appears in front of the rock, positioned further back in the scene, adding to the depth perception with its layered positioning.\n\nLayer 4: \nThe farthest from the camera, a <ref>giraffe</ref><box>[[134, 168, 538, 831]]</box> is interacting with the environment, peering toward the other elements of the scene, contributing to the sense of perspective and scale in the composition."
}
] |
coco/train2017/000000357386.jpg
|
00000005
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the image, a <ref>black bear</ref><box>[[223, 316, 615, 545]]</box> is present, <pred>running on</pred><box>[[223, 316, 615, 545]]</box><box>[[0, 344, 999, 874]]</box> the <ref>grass</ref><box>[[0, 344, 999, 874]]</box>. Nearby, there is a <ref>tree</ref><box>[[0, 124, 999, 617]]</box>.\n\nLayer 2: There are <ref>large rocks</ref><box>[[135, 283, 999, 541]]</box> <pred>beside</pred><box>[[135, 283, 999, 541]]</box><box>[[0, 124, 999, 617]]</box> the tree and <pred>on</pred><box>[[135, 283, 999, 541]]</box><box>[[0, 344, 999, 874]]</box> the grass."
}
] |
coco/train2017/000000246074.jpg
|
00000006
| 500
| 376
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \n- A <ref>person</ref><box>[[784, 877, 830, 999]]</box> is seen among the scene.\n\nLayer 2: \n- The <ref>pavement</ref><box>[[770, 781, 875, 999]]</box> runs alongside the road.\n\nLayer 3: \n- A <ref>bicycle</ref><box>[[740, 687, 766, 811]]</box> is <pred>parked on</pred><box>[[225, 81, 842, 999]]</box> the side of the road.\n- An <ref>Asian street</ref><box>[[123, 0, 875, 999]]</box> with various <ref>shops</ref><box>[[123, 0, 875, 999]]</box> and <ref>signs</ref><box>[[223, 0, 875, 941]]</box> lining the buildings.\n- A <ref>road</ref><box>[[225, 81, 842, 999]]</box> is visible with Chinese character markings.\n- A <ref>light</ref><box>[[137, 557, 303, 787]]</box> is <pred>over</pred><box>[[297, 739, 391, 868]]</box> one of the bicycles.\n- A <ref>window</ref><box>[[123, 0, 299, 563]]</box> is <pred>over</pred><box>[[311, 355, 395, 453]]</box> a <ref>roof</ref><box>[[311, 355, 395, 453]]</box> on the building facades.\n\nLayer 4: \n- A <ref>motorcycle</ref><box>[[648, 487, 724, 564]]</box> is <pred>visible among</pred><box>[[630, 473, 660, 561]]</box> the parked vehicles.\n\nLayer 5: \n- A <ref>bicycle</ref><box>[[630, 473, 660, 561]]</box> is <pred>parked on</pred><box>[[225, 81, 842, 999]]</box> the side of the road.\n\nLayer 6: \n- A <ref>tie</ref><box>[[590, 381, 596, 403]]</box> is <pred>worn by</pred><box>[[569, 363, 612, 447]]</box> one of the persons in the street.\n\nLayer 7: \n- A <ref>handbag</ref><box>[[569, 403, 578, 426]]</box> is <pred>held by</pred><box>[[569, 363, 612, 447]]</box> one of the persons.\n\nLayer 8: \n- A <ref>person</ref><box>[[569, 363, 612, 447]]</box> is <pred>standing on</pred><box>[[225, 81, 842, 999]]</box> the road.\n\nLayer 9: \n- A <ref>bicycle</ref><box>[[566, 317, 617, 373]]</box> is <pred>parked on</pred><box>[[225, 81, 842, 999]]</box> the side of the road.\n\nLayer 10: \n- Another <ref>person</ref><box>[[465, 173, 478, 213]]</box> is <pred>walking on</pred><box>[[225, 81, 842, 999]]</box> the road.\n\nLayer 11: \n- A <ref>bicycle</ref><box>[[353, 605, 367, 645]]</box> is <pred>parked on</pred><box>[[225, 81, 842, 999]]</box> the side of the road.\n\nLayer 12: \n- A <ref>bicycle</ref><box>[[332, 493, 399, 571]]</box> is <pred>parked on</pred><box>[[225, 81, 842, 999]]</box> the side of the road.\n\nLayer 13: \n- A <ref>bicycle</ref><box>[[297, 739, 391, 868]]</box> is <pred>parked on</pred><box>[[225, 81, 842, 999]]</box> the side of the road.\n\nLayer 14: \n- A <ref>wall</ref><box>[[123, 153, 189, 293]]</box> is seen on the left side of the image."
}
] |
coco/train2017/000000325956.jpg
|
00000007
| 426
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \n- A <ref>skateboard</ref><box>[[155, 466, 319, 557]]</box> is present with one of the <ref>backpacks</ref><box>[[899, 414, 999, 531]]</box> nearby.\n- A <ref>ceiling</ref><box>[[0, 167, 566, 485]]</box> spans the scene.\n\nLayer 2: \n- One of the <ref>men</ref><box>[[853, 249, 991, 695]]</box> watches the activities.\n\nLayer 3: \n- The <ref>floor</ref><box>[[0, 672, 999, 831]]</box> stretches across the image.\n\nLayer 4: \n- Additional <ref>men</ref><box>[[705, 291, 838, 719]]</box> are gathered, observing the skateboard action.\n\nLayer 5: \n- More of the <ref>men</ref><box>[[612, 313, 718, 712]]</box> are positioned in view.\n\nLayer 6: \n- Another <ref>backpacks</ref><box>[[472, 650, 524, 719]]</box> is set against the scene.\n- Several <ref>men</ref><box>[[521, 350, 632, 716]]</box> remain present nearby.\n\nLayer 7: \n- Additional presence of the <ref>men</ref><box>[[444, 425, 543, 664]]</box> further occupies the space.\n\nLayer 8:\n- A <ref>person</ref><box>[[4, 171, 452, 508]]</box> is engaged in athletic movements.\n\nLayer 9:\n- A <ref>backpacks</ref><box>[[391, 622, 490, 730]]</box> is noticeable within the frame.\n\nLayer 10:\n- The presence of more <ref>men</ref><box>[[388, 435, 451, 642]]</box> is evident.\n\nLayer 11:\n- Another spectator, one of the <ref>men</ref><box>[[18, 549, 67, 734]]</box>.\n\nLayer 12:\n- Yet another observer from the group of <ref>men</ref><box>[[0, 557, 29, 738]]</box>.\n\nLayer 13:\n- A <ref>light</ref><box>[[0, 314, 75, 372]]</box> fixture is mounted, illuminating the surroundings."
}
] |
coco/train2017/000000231655.jpg
|
00000008
| 426
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>boat</ref><box>[[833, 179, 977, 242]]</box> is depicted in close proximity.\n\nLayer 2: More <ref>boats</ref><box>[[767, 167, 905, 328]]</box> and <ref>boats</ref><box>[[722, 182, 938, 242]]</box> appear on the <ref>sea</ref><box>[[0, 192, 999, 831]]</box>.\n\nLayer 3: Additional <ref>boats</ref><box>[[736, 227, 938, 276]]</box> float on the <ref>sea</ref><box>[[0, 192, 999, 831]]</box>.\n\nLayer 4: Several <ref>boats</ref><box>[[683, 305, 772, 363]]</box> dot the vast expanse of the <ref>sea</ref><box>[[0, 192, 999, 831]]</box>, and a particular <ref>boat</ref><box>[[28, 167, 408, 500]]</box> stands out.\n\nLayer 5: A group of <ref>people</ref><box>[[62, 183, 113, 357]]</box> are seen against the backdrop of the <ref>sea</ref><box>[[0, 192, 999, 831]]</box>, alongside <ref>boats</ref><box>[[28, 167, 408, 500]]</box>.\n\nLayer 6: A <ref>dog</ref><box>[[582, 642, 616, 694]]</box> joins the scene, positioned on the <ref>sea</ref><box>[[0, 192, 999, 831]]</box>, and highlights the <ref>backpack</ref><box>[[482, 594, 552, 674]]</box> near it.\n\nLayer 7: More <ref>people</ref><box>[[424, 206, 524, 364]]</box> can be seen on the <ref>sea</ref><box>[[0, 192, 999, 831]]</box> amidst the presence of <ref>boats</ref><box>[[28, 167, 408, 500]]</box>.\n\nLayer 8: A central <ref>person</ref><box>[[402, 571, 568, 716]]</box> is <pred>sitting on</pred><box>[[402, 571, 568, 716]]</box><box>[[313, 638, 746, 760]]</box> and <pred>driving</pred><box>[[402, 571, 568, 716]]</box><box>[[313, 638, 746, 760]]</box> a <ref>boat</ref><box>[[313, 638, 746, 760]]</box> over the expansive <ref>sea</ref><box>[[0, 192, 999, 831]]</box>.\n\nLayer 9: A few more <ref>people</ref><box>[[188, 216, 207, 312]]</box> speckle the <ref>sea</ref><box>[[0, 192, 999, 831]]</box> beside <ref>boats</ref><box>[[28, 167, 408, 500]]</box>.\n\nLayer 10: Additional <ref>people</ref><box>[[121, 190, 173, 299]]</box> linger on the <ref>sea</ref><box>[[0, 192, 999, 831]]</box> near <ref>boats</ref><box>[[28, 167, 408, 500]]</box>."
}
] |
coco/train2017/000000557339.jpg
|
00000009
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>sink</ref><box>[[936, 619, 999, 672]]</box> is located at the far end, near a <ref>microwave</ref><box>[[215, 249, 477, 405]]</box>, and a lengthy <ref>counter</ref><box>[[0, 543, 999, 694]]</box> is visible. A <ref>cup</ref><box>[[117, 561, 155, 607]]</box> sits on the counter. The kitchen scene is dominated by a large, spanning <ref>cabinet</ref><box>[[0, 165, 999, 831]]</box> at the backdrop.\n\nLayer 2: Several <ref>bottles</ref><box>[[911, 474, 942, 575], [977, 493, 997, 581], [930, 477, 961, 583], [944, 495, 981, 614]]</box> are spread across the countertop. A <ref>potted plant</ref><box>[[844, 276, 991, 531]]</box> adds greenery to the space.\n\nLayer 3: A <ref>cup</ref><box>[[897, 575, 938, 621]]</box> can be seen, contributing to the kitchen items.\n\nLayer 4: Another <ref>bottle</ref><box>[[885, 507, 911, 589]]</box> sits further back with the <ref>potted plant</ref><box>[[844, 276, 991, 531]]</box> resting nearby.\n\nLayer 5: A <ref>vase</ref><box>[[880, 444, 899, 466]]</box> and a prominent <ref>wall</ref><box>[[800, 165, 999, 501]]</box> are present in this layer, along with more <ref>bottles</ref><box>[[118, 485, 138, 567], (892, 425, 913, 508), [910, 424, 938, 515]]</box>. A decorative <ref>wall-tile</ref><box>[[0, 396, 285, 575]]</box> enhances the kitchen ambiance.\n\nLayer 6: A <ref>bottle</ref><box>[[866, 516, 892, 575]]</box> is set against the background, alongside the enduring presence of the <ref>potted plant</ref><box>[[844, 276, 991, 531]]</box>.\n\nLayer 7: One <ref>person</ref><box>[[379, 242, 735, 722]]</box> appears near the cabinet. The long <ref>dining table</ref><box>[[220, 686, 825, 820]]</box> offers a central aspect of the kitchen layout.\n\nLayer 8: A <ref>cup</ref><box>[[366, 725, 435, 829]]</box> is positioned, along with the continued view of the <ref>microwave</ref><box>[[215, 249, 477, 405]]</box> and the backdrop cabinetry.\n\nLayer 9: The <ref>oven</ref><box>[[366, 614, 418, 727]]</box> is visible alongside the <ref>microwave</ref><box>[[215, 249, 477, 405]]</box>.\n\nLayer 10: A hanging <ref>towel</ref><box>[[26, 495, 99, 524]]</box> adds a touch of daily utility to this layer.\n\nLayer 11: A <ref>table</ref><box>[[218, 806, 830, 831]]</box> sets the scene near a <ref>person</ref><box>[[600, 241, 856, 803]]</box>.\n\nLayer 12: The kitchen includes a <ref>toaster</ref><box>[[21, 515, 110, 591]]</box>, and once again, the green scene from the <ref>potted plant</ref><box>[[844, 276, 991, 531]]</box>.\n\nLayer 13: Lastly, a third <ref>person</ref><box>[[120, 332, 441, 825]]</box> engages with others, with the expansive cabinet framing the background."
}
] |
coco/train2017/000000534314.jpg
|
00000010
| 425
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: <ref>group of people</ref><box>[[85, 697, 143, 722]]</box>, <ref>group of people</ref><box>[[92, 720, 170, 830]]</box>, <ref>group of people</ref><box>[[641, 666, 825, 820]]</box>, <ref>group of people</ref><box>[[416, 690, 775, 822]]</box>, <ref>group of people</ref><box>[[6, 483, 997, 830]]</box>, and a <ref>street</ref><box>[[0, 167, 999, 830]]</box> bustling with activity.\n\nLayer 2: <ref>group of people</ref><box>[[6, 637, 124, 681]]</box> among the crowd on the <ref>street</ref><box>[[0, 167, 999, 830]]</box>, beneath some <ref>black umbrellas</ref><box>[[635, 512, 669, 541]]</box>, <ref>black umbrellas</ref><box>[[649, 590, 680, 617]]</box>, and <ref>black umbrellas</ref><box>[[618, 623, 967, 746]]</box>.\n\nLayer 3: More <ref>group of people</ref><box>[[45, 690, 98, 830]]</box> are found in the bustling scene, while the <ref>buildings</ref><box>[[0, 167, 999, 733]]</box> provide the backdrop.\n\nLayer 4: A <ref>double decker bus</ref><box>[[390, 505, 646, 830]]</box> moves through the street, displaying a <ref>banner</ref><box>[[401, 471, 607, 532]]</box> for a celebration.\n\nLayer 5: Another part of the <ref>group of people</ref><box>[[373, 714, 391, 728]]</box> adds to the gathering around the bus.\n\nLayer 6: The scene features <ref>group of people</ref><box>[[334, 717, 362, 813]]</box> near one another.\n\nLayer 7: A <ref>group of people</ref><box>[[334, 708, 357, 717]]</box> standing with <ref>black umbrellas</ref><box>[[334, 708, 357, 717]]</box> overhead.\n\nLayer 8: Further along, there are <ref>group of people</ref><box>[[296, 710, 338, 815]]</box> closely assembled.\n\nLayer 9: The <ref>group of people</ref><box>[[291, 698, 335, 719]]</box> intermingle within the scene of excitement.\n\nLayer 10: Among them, a <ref>person</ref><box>[[216, 711, 291, 817]]</box> is distinct as they are <pred>wearing</pred><box>[[216, 711, 291, 817]]</box><box>[[249, 744, 263, 795]]</box> a <ref>tie</ref><box>[[249, 744, 263, 795]]</box>.\n\nLayer 11: Positioned next are more <ref>group of people</ref><box>[[238, 708, 310, 731]]</box> with several <ref>black umbrellas</ref><box>[[238, 708, 310, 731]]</box> above.\n\nLayer 12: Nearby, another <ref>group of people</ref><box>[[207, 708, 235, 779]]</box> appears.\n\nLayer 13: More individuals in the <ref>group of people</ref><box>[[199, 706, 216, 748]]</box> highlight the crammed environment.\n\nLayer 14: The composition includes more <ref>group of people</ref><box>[[157, 734, 174, 760]]</box> engaging in the event.\n\nLayer 15: A <ref>group of people</ref><box>[[142, 692, 176, 736]]</box> mingles amidst the gathering.\n\nLayer 16: Here, <ref>black umbrellas</ref><box>[[140, 692, 159, 711]]</box> are present among the <ref>group of people</ref><box>[[92, 720, 170, 830]]</box>.\n\nLayer 17: Finally, a prominent <ref>group of people</ref><box>[[138, 756, 202, 830]]</box> is part of the street filled with activity."
}
] |
coco/train2017/000000374966.jpg
|
00000011
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nIn the foreground, we see <ref>people</ref><box>[[903, 428, 955, 610]]</box> and <ref>people</ref><box>[[727, 370, 864, 769]]</box>, some of whom are <pred>playing</pred> with their <ref>skateboards</ref><box>[[878, 549, 935, 616]]</box> on the <ref>pavement</ref><box>[[0, 397, 999, 874]]</box>. Another <ref>person</ref><box>[[316, 471, 462, 650]]</box> joins the activity on the ground.\n\nLayer 2: \nMultiple <ref>people</ref><box>[[75, 400, 106, 516]]</box> and a <ref>person</ref><box>[[856, 422, 897, 447]]</box> are also present, enhancing the vibrant atmosphere.\n\nLayer 3: \nA <ref>skateboard</ref><box>[[703, 703, 777, 753]]</box> is visible, highlighting some more <pred>playing</pred> that’s happening in the area. \n\nLayer 4:\n<ref>People</ref><box>[[416, 427, 479, 545]]</box> contribute to the lively scene. The expansive <ref>pavement</ref><box>[[0, 397, 999, 874]]</box> accommodates the activities.\n\nLayer 5: \n<ref>People</ref><box>[[869, 364, 900, 481]]</box> are scattered around the area, with some near the <ref>trees</ref><box>[[0, 124, 999, 547]]</box> and lush <ref>grass</ref><box>[[132, 405, 999, 494]]</box>.\n\nLayer 6: \n<ref>people</ref><box>[[881, 477, 997, 601]]</box>, nearby <ref>skateboards</ref><box>[[493, 594, 565, 607]]</box> continue lending energy to the park. \n\nLayer 7: \nAdditional <ref>people</ref><box>[[49, 406, 72, 475]]</box> add to the crowd, among vibrant greenery. The <ref>building</ref><box>[[351, 124, 999, 436]]</box> stands further back, surrounded by <ref>trees</ref><box>[[0, 124, 999, 547]]</box>.\n\nLayer 8: \nA distant <ref>building</ref><box>[[351, 124, 999, 436]]</box> defines the background.\n\nLayer 9: \n<ref>skateboards</ref><box>[[327, 644, 438, 667]]</box> rest on the <ref>pavement</ref><box>[[0, 397, 999, 874]]</box>.\n\nLayer 10: \nMore distant <ref>people</ref><box>[[869, 364, 900, 481]]</box> are spaced apart, around <ref>grass</ref><box>[[132, 405, 999, 494]]</box>.\n\nLayer 11: \nThe <ref>sky</ref><box>[[251, 124, 881, 244]]</box> stretches above everything. \n\nLayer 12:\nIn the back, <ref>people</ref><box>[[150, 394, 174, 474]]</box> continue to dot the park landscape.\n\nLayer 13: \nFinally, some <ref>people</ref><box>[[101, 424, 123, 513]]</box> appear amongst the foliage, part of the day's activities."
}
] |
coco/train2017/000000163457.jpg
|
00000012
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>wooden wall</ref><box>[[856, 455, 981, 595]]</box> is visible, adding to the nearby scene. The <ref>truck</ref><box>[[84, 223, 789, 823]]</box>, carrying a load in the back, is seen close to the camera. It is moving along the <ref>road</ref><box>[[0, 631, 999, 874]]</box>. A <ref>building</ref><box>[[925, 400, 999, 608]]</box> stands to the right of the scene, alongside the <ref>road</ref><box>[[0, 631, 999, 874]]</box>. A <ref>tree</ref><box>[[77, 389, 157, 542]]</box> is on the left, contributing to the roadside scenery. The <ref>sky</ref><box>[[0, 124, 999, 445]]</box> encompasses the entire scene above.\n\nLayer 2: Patches of <ref>sand</ref><box>[[786, 513, 999, 697]]</box> are scattered across the area.\n\nLayer 3: Some areas display spots of <ref>dirt</ref><box>[[771, 603, 955, 686]]</box> scattered around.\n\nLayer 4: A person sits within the <ref>truck</ref><box>[[84, 223, 789, 823]]</box> towards the back, <ref>people</ref><box>[[618, 477, 696, 555]]</box>.\n\nLayer 5: In the <ref>truck</ref><box>[[84, 223, 789, 823]]</box>, another group of <ref>people</ref><box>[[494, 466, 560, 538]]</box> appears to be on board.\n\nLayer 6: A person occupies the driving seat of the <ref>truck</ref><box>[[84, 223, 789, 823]]</box>, <ref>people</ref><box>[[402, 478, 491, 568]]</box>.\n\nLayer 7: The magnificent <ref>mountain</ref><box>[[0, 357, 936, 656]]</box> serves as the background for the scene."
}
] |
coco/train2017/000000322738.jpg
|
00000013
| 421
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>person</ref><box>[[925, 331, 970, 427]]</box> is closest to the camera, and they're positioned next to the expansive <ref>sand</ref><box>[[0, 405, 999, 827]]</box> in the arena. The arena's <ref>wall</ref><box>[[0, 170, 999, 470]]</box> stands alongside the sand-filled ground. \n\nLayer 2: Additionally, there is a <ref>tree</ref><box>[[830, 354, 866, 371]]</box> in a similar layer. Another <ref>person</ref><box>[[828, 281, 899, 369]]</box> is slightly further back in this layer.\n\nLayer 3: The <ref>horse</ref><box>[[600, 255, 824, 739]]</box>, carrying its rider, appears deeper in the scene, depicted in motion.\n\nLayer 4: A <ref>person</ref><box>[[526, 206, 685, 581]]</box> is mounted on the horse, navigating the sandy terrain.\n\nLayer 5: A <ref>person</ref><box>[[285, 339, 348, 368]]</box> is situated at a more distant point, near the beginning of an architectural <ref>wall</ref><box>[[0, 170, 999, 470]]</box> that features a visible <ref>banner</ref><box>[[0, 170, 251, 444]]</box>.\n\nLayer 6: The farthest individuals are a <ref>person</ref><box>[[224, 336, 271, 368]]</box> located alongside the wall where the banner is placed."
}
] |
coco/train2017/000000536648.jpg
|
00000014
| 494
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: The scene shows two <ref>people</ref><box>[[621, 427, 885, 783], [354, 230, 682, 642]]</box> performing water skiing tricks in mid-air. They are equipped with <ref>skis</ref><box>[[490, 590, 831, 873], [224, 499, 578, 749]]</box>, demonstrating an impressive feat of athleticism over a stretch of <ref>water</ref><box>[[0, 766, 999, 885]]</box>. Nearby, a portion of the <ref>crowd</ref><box>[[847, 807, 903, 882], [763, 787, 853, 885]]</box> is positioned at the closest point, observing the action unfold.\n\nLayer 2: Another segment of the watching <ref>crowd</ref><box>[[699, 828, 764, 885]]</box> can be seen, amazed by the performers as they continue their aquatic escapade.\n\nLayer 3: A third group of spectators makes up the <ref>crowd</ref><box>[[623, 806, 697, 885]]</box>, indicating varying levels of engagement and perspectives on the scene. \n\nLayer 4: Adjacent, additional bystanders within the <ref>crowd</ref><box>[[551, 803, 615, 885]]</box> also enjoy the spectacle alongside the natural elements of the environment, which include a gracefully positioned <ref>house</ref><box>[[18, 630, 743, 781]]</box> and a backdrop of lush <ref>trees</ref><box>[[0, 475, 999, 793]]</box>.\n\nLayer 5: More participants in the <ref>crowd</ref><box>[[521, 859, 543, 885]]</box> remain visible, capturing the allure of lakeside entertainment.\n\nLayer 6: The final subset of the <ref>crowd</ref><box>[[494, 867, 519, 885]]</box> completes the scene from a slightly farther vantage point, enriching the overall dynamic with a multitude of observers scattered along the shore.\n\nLayer 7: Above all, a <ref>sky</ref><box>[[0, 113, 999, 566]]</box> arching over the panoramic scene complements the action beneath it, encompassing both the vibrant performance and the tranquility of nature."
}
] |
coco/train2017/000000480939.jpg
|
00000015
| 426
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nA <ref>paper</ref><box>[[874, 675, 955, 695]]</box> is on the <ref>pavement</ref><box>[[0, 619, 999, 831]]</box> near a <ref>bench</ref><box>[[16, 357, 997, 720]]</box>. A <ref>woman</ref><box>[[23, 409, 540, 543]]</box> is lying on the bench. The pavement extends around the scene.\n\nLayer 2:\nA <ref>man</ref><box>[[546, 227, 808, 676]]</box> is sitting next to the woman on the bench and holding a <ref>cell phone</ref><box>[[705, 277, 752, 344]]</box> to his ear. Beside him, there's a <ref>backpack</ref><box>[[513, 536, 800, 819]]</box> and two <ref>bottles</ref><box>[[555, 666, 600, 744]]</box> and <box>[[772, 686, 808, 742]]</box> on the pavement.\n\nLayer 3:\nIn the background, a <ref>fence</ref><box>[[0, 268, 999, 620]]</box> and <ref>traffic lights</ref><box>[[521, 249, 533, 274]]</box> can be observed as part of the urban setting.\n\nLayer 4:\nMore <ref>traffic lights</ref><box>[[232, 304, 245, 318]]</box> and <box>[[229, 299, 234, 313]]</box> are visible, contributing to the urban details.\n\nLayer 5:\nFurther in the background, additional <ref>traffic lights</ref><box>[[199, 268, 213, 291]]</box> are observed along with a sprawling <ref>building</ref><box>[[0, 167, 999, 332]]</box>."
}
] |
coco/train2017/000000157986.jpg
|
00000016
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the forefront, there are several <ref>men</ref><box>[[3, 497, 359, 821]]</box> standing next to a large, white <ref>truck</ref><box>[[42, 165, 999, 823]]</box>. Behind them, a <ref>tree</ref><box>[[0, 165, 999, 599]]</box> provides a backdrop. \n\nLayer 2: Among the <ref>men</ref>, one is <pred>wearing</pred><box>[[218, 495, 398, 831]]</box> a <ref>tie</ref><box>[[284, 575, 308, 671]]</box>, indicating formal attire. \n\nLayer 3: The truck is <pred>on</pred><box>[[42, 165, 999, 823]]</box> a <ref>road</ref><box>[[185, 711, 477, 831]]</box>.\n\nLayer 4: In the distance, above the scene, is the <ref>sky</ref><box>[[18, 165, 285, 486]]</box> hovering over the truck. To its side, a <ref>building</ref><box>[[0, 165, 29, 494]]</box> is visible, and a small portion of the <ref>pavement</ref><box>[[0, 517, 31, 545]]</box> can be seen."
}
] |
coco/train2017/000000221223.jpg
|
00000017
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: There is a <ref>clock</ref><box>[[916, 233, 961, 294]]</box> <pred>hanging from</pred><box>[[916, 233, 961, 294]]</box><box>[[0, 124, 999, 688]]</box> the wall.\n\nLayer 2: A <ref>potted plant</ref><box>[[908, 328, 999, 529]]</box> and another <ref>potted plant</ref><box>[[760, 124, 939, 227]]</box> are visible in the room.\n\nLayer 3: The <ref>teddy bear</ref><box>[[763, 321, 841, 383]]</box> is <pred>in</pred><box>[[763, 321, 841, 383]]</box><box>[[28, 208, 952, 751]]</box> the cabinet of the entertainment center.\n\nLayer 4: Another <ref>clock</ref><box>[[652, 180, 672, 200]]</box> is <pred>on</pred><box>[[652, 180, 672, 200]]</box><box>[[117, 124, 899, 499]]</box> the shelf, part of the larger <ref>entertainment center</ref><box>[[28, 208, 952, 751]]</box>, along with several different <ref>bowls</ref><box>[[728, 778, 869, 850]]</box>, <box>[[0, 800, 24, 842]]</box>, <box>[[0, 766, 71, 805]]</box>. A <ref>table</ref><box>[[0, 679, 999, 874]]</box> is visible in the image. The <ref>floor</ref><box>[[856, 717, 902, 751]]</box> can be partially seen also.\n\nLayer 5: There is a <ref>vase</ref><box>[[60, 221, 90, 252]]</box> and a <ref>TV</ref><box>[[266, 230, 694, 559]]</box>, which is displaying news. The <ref>TV</ref><box>[[266, 230, 694, 559]]</box> is part of the <ref>entertainment center</ref><box>[[28, 208, 952, 751]]</box>. The <ref>elephant</ref><box>[[652, 672, 796, 787]]</box> figurine is <pred>beside</pred><box>[[652, 672, 796, 787]]</box><box>[[728, 778, 869, 850]]</box> one of the bowls.\n\nLayer 6: A <ref>bowl</ref><box>[[56, 802, 150, 848]]</box> is placed on the table.\n\nLayer 7: Multiple <ref>potted plants</ref><box>[[53, 597, 182, 781]]</box> and <ref>potted plants</ref><box>[[652, 672, 796, 787]]</box> can be seen.\n\nLayer 8: Another <ref>bowl</ref><box>[[515, 778, 563, 809]]</box> is also present.\n\nLayer 9: The <ref>tie</ref><box>[[427, 408, 463, 499]]</box> is <pred>worn by</pred><box>[[427, 408, 463, 499]]</box><box>[[298, 247, 597, 502]]</box> the person.\n\nLayer 10: A different <ref>bowl</ref><box>[[423, 764, 490, 809]]</box> is visible.\n\nLayer 11: One more <ref>bowl</ref><box>[[360, 776, 408, 808]]</box> can be seen.\n\nLayer 12: A <ref>person</ref><box>[[298, 247, 597, 502]]</box> is depicted in the image.\n\nLayer 13: Finally, another <ref>bowl</ref><box>[[106, 766, 187, 814]]</box> is placed in the scene."
}
] |
coco/train2017/000000501381.jpg
|
00000018
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>skateboard</ref><box>[[426, 411, 575, 453]]</box> is positioned in the scene, in close association with a <ref>fence</ref><box>[[382, 439, 866, 621]]</box>. Overhead, <ref>lights</ref><box>[[193, 186, 697, 222]]</box> are <pred>on</pred><box>[[193, 186, 697, 222]]</box> and illuminate the area, providing visibility within the skatepark. A <ref>wall</ref><box>[[0, 314, 999, 637]]</box> forms part of the backdrop.\n\nLayer 2: A <ref>person</ref><box>[[341, 222, 535, 444]]</box> is <pred>playing</pred><box>[[341, 222, 535, 444]]</box> while using the skateboard, captured in an aerial maneuver above the skatepark.\n\nLayer 3: The scene includes a <ref>door</ref><box>[[127, 452, 188, 628]]</box>, serving as a part of the spatial framework visible in the background alongside the <ref>wall</ref><box>[[0, 314, 999, 637]]</box>.\n\nLayer 4: A <ref>ceiling</ref><box>[[0, 165, 999, 360]]</box> spans the top portion of the image, encasing the indoor skatepark environment."
}
] |
coco/train2017/000000022215.jpg
|
00000019
| 414
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: There are a <ref>tree</ref><box>[[939, 371, 999, 580]]</box> present, along with a <ref>railroad</ref><box>[[16, 503, 999, 822]]</box> track visible. A <ref>traffic light</ref><box>[[84, 287, 154, 428]]</box> is also observed, positioned near a <ref>building</ref><box>[[0, 241, 999, 631]]</box> that forms part of the backdrop. The <ref>sky</ref><box>[[0, 176, 999, 484]]</box> arches above the scene.\n\nLayer 2: We can see <ref>people</ref><box>[[649, 510, 697, 545]]</box> inside a <ref>train</ref><box>[[212, 449, 738, 666]]</box> traveling on the railroad. The <ref>platform</ref><box>[[0, 570, 999, 822]]</box> runs adjacent to the track.\n\nLayer 3: Additional <ref>people</ref><box>[[503, 528, 513, 541]]</box> are visible inside the train, enhancing the sense of passengers on board.\n\nLayer 4: Further <ref>people</ref><box>[[451, 516, 466, 537]]</box> appear within the train, adding to the picture of a populated transport scene.\n\nLayer 5: A small patch of <ref>sand</ref><box>[[24, 574, 75, 603]]</box> is situated near the platform, contributing to the details of the ground scenery."
}
] |
coco/train2017/000000259666.jpg
|
00000020
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1:\n- A <ref>piece of paper</ref><box>[[838, 569, 931, 685]]</box> is attached to the <ref>wall</ref><box>[[0, 124, 999, 874]]</box>.\n- A colorful <ref>blanket</ref><box>[[0, 672, 686, 874]]</box> is placed on the <ref>bed</ref><box>[[0, 397, 668, 859]]</box>.\n- There is a wooden <ref>cabinet</ref><box>[[460, 600, 956, 874]]</box> beside the bed.\n- A <ref>tv</ref><box>[[811, 161, 999, 367]]</box> is mounted on the wall.\n- A <ref>ceiling</ref><box>[[0, 124, 908, 221]]</box> overhead.\n\nLayer 2:\n- Some <ref>remotes</ref><box>[[638, 610, 664, 624]]</box> are beside the bed.\n\nLayer 3:\n- Additional <ref>remotes</ref><box>[[622, 611, 646, 625]]</box> are beside the bed.\n\nLayer 4:\n- More <ref>remotes</ref><box>[[605, 610, 630, 627]]</box> are beside the bed.\n\nLayer 5:\n- A <ref>rug</ref><box>[[568, 756, 682, 826]]</box> is placed on the floor.\n- The <ref>bed</ref><box>[[0, 397, 668, 859]]</box> is with a wooden frame.\n- The <ref>cabinet</ref><box>[[460, 600, 956, 874]]</box> is beside the bed.\n- The <ref>wall</ref><box>[[0, 124, 999, 874]]</box> is part of the room structure.\n- The <ref>ceiling</ref><box>[[0, 124, 908, 221]]</box>.\n\nLayer 6:\n- A <ref>light</ref><box>[[413, 124, 607, 636]]</box> is hanging from the ceiling."
}
] |
coco/train2017/000000337452.jpg
|
00000021
| 368
| 500
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nIn the foreground, several <ref>people</ref><box>[[547, 381, 809, 764], [663, 309, 793, 725]]</box> engage with their <ref>skateboards</ref><box>[[615, 706, 809, 792]]</box>. One individual appears to be <pred>holding</pred><box>[[901, 527, 978, 754]]</box> a <ref>skateboard</ref><box>[[901, 527, 978, 754]]</box>. The <ref>pavement</ref><box>[[0, 529, 999, 867]]</box> serves as the base upon which these activities take place.\n\nLayer 2:\nFurther back, another <ref>person</ref><box>[[865, 293, 999, 775]]</box> is <pred>standing on</pred><box>[[0, 529, 999, 867]]</box> the pavement, blending into the dynamic scene as people continue to maneuver skateboards.\n\nLayer 3:\nContinuing to the third layer, <ref>people</ref><box>[[645, 277, 701, 425]]</box> are visible, contributing to the vibrant activity occurring near the skate bowl. Their actions create a lively atmosphere against the <ref>pavement</ref><box>[[0, 529, 999, 867]]</box>.\n\nLayer 4:\nOn a slightly deeper plane, <ref>backpacks</ref><box>[[421, 530, 465, 600]]</box> can be seen resting <pred>on</pred><box>[[0, 529, 999, 867]]</box> the pavement, hinting at transient pauses amidst the bustling scene. \n\nLayer 5:\nIn this layer, there are additional <ref>people</ref><box>[[459, 367, 543, 678], [301, 293, 429, 669]]</box> surrounding the skate area, blending into the crowd while more <ref>skateboards</ref><box>[[413, 577, 491, 670]]</box> rest on the ground.\n\nLayer 6:\nA further depth reveals yet more <ref>people</ref><box>[[211, 375, 311, 642], [243, 563, 274, 594]]</box> and their accompanying <ref>skateboards</ref><box>[[215, 565, 305, 644]]</box>, amidst various belongings punctuating the scene with stationary contrast to the dynamic activity.\n\nLayer 7:\nA pair of <ref>people</ref><box>[[219, 323, 317, 439], [217, 351, 243, 385]]</box> is discernible, contributing to the layers of human presence in this vibrant setting.\n\nLayer 8:\nIn the distant view, a small number of <ref>people</ref><box>[[165, 327, 235, 600]]</box> linger, framing the skate park activity.\n\nLayer 9:\nFinally, a long <ref>fence</ref><box>[[0, 321, 205, 449]]</box> outlines the area, with a prominent <ref>building</ref><box>[[0, 131, 999, 728]]</box> as the backdrop, affirming the surrounding landscape and its connections to the skateboarding culture, as suggested by its signage."
}
] |
coco/train2017/000000130069.jpg
|
00000022
| 412
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nA single <ref>pair of skis</ref><box>[[99, 627, 120, 633]]</box> placed over the <ref>snow</ref><box>[[0, 542, 999, 821]]</box>, which blankets the path. <ref>Trees</ref><box>[[0, 202, 999, 624]]</box> line the sides of the path, and the sky is clear <ref>above</ref><box>[[0, 177, 999, 430]]</box>.\n\nLayer 2: \nAnother <ref>person</ref><box>[[99, 575, 118, 625]]</box> is on the path, amidst the <ref>snow</ref><box>[[0, 542, 999, 821]]</box>, surrounded by the <ref>trees</ref><box>[[0, 202, 999, 624]]</box>, beneath the open <ref>sky</ref><box>[[0, 177, 999, 430]]</box>.\n\nLayer 3: \n<ref>Skis</ref><box>[[906, 747, 999, 761]]</box> can be spotted at this level, along with a <ref>person</ref><box>[[885, 511, 988, 761]]</box>, who is perhaps <pred>carrying</pred><box>[[885, 511, 988, 761]]</box> a <ref>backpack</ref><box>[[952, 558, 999, 625]]</box>. Above them spreads the tranquil <ref>sky</ref><box>[[0, 177, 999, 430]]</box>.\n\nLayer 4: \nAnother <ref>person</ref><box>[[85, 568, 103, 619]]</box> stands on the snowy path enveloped by <ref>trees</ref><box>[[0, 202, 999, 624]]</box>, with the endless <ref>sky</ref><box>[[0, 177, 999, 430]]</box> extending above.\n\nLayer 5: \nThere is a group of active <ref>people</ref><box>[[755, 723, 836, 738], [746, 535, 810, 730], [216, 555, 311, 755], [376, 557, 469, 759], [199, 547, 245, 698], [885, 511, 988, 761]]</box> <pred>playing</pred><box>[[746, 535, 810, 730], [376, 557, 469, 759]]</box>, some of whom have <ref>skis</ref><box>[[357, 742, 543, 785], [755, 723, 836, 738], [173, 697, 284, 717]]</box> and are carrying <ref>backpacks</ref><box>[[952, 558, 999, 625], [760, 562, 805, 627]]</box>, amidst the snowy landscape.\n\nLayer 6: \nHere stands another <ref>person</ref><box>[[70, 569, 92, 627]]</box> set in the snowy terrain, lined by the wandering <ref>trees</ref><box>[[0, 202, 999, 624]]</box>, all under the wide-open <ref>sky</ref><box>[[0, 177, 999, 430]]</box>.\n\nLayer 7: \nA <ref>person</ref><box>[[48, 568, 72, 624]]</box> is in this layer, surrounded by nature's <ref>trees</ref><box>[[0, 202, 999, 624]]</box>, draped under the expansive <ref>sky</ref><box>[[0, 177, 999, 430]]</box>.\n\nLayer 8: \n<ref>Skis</ref><box>[[259, 747, 338, 783]]</box> are part of this scene, along with more <ref>people</ref><box>[[746, 535, 810, 730], [376, 557, 469, 759], [199, 547, 245, 698]]</box>, and they are part of the active mood set on the snowy <ref>path</ref><box>[[0, 542, 999, 821]]</box>, with some carrying <ref>backpacks</ref><box>[[952, 558, 999, 625], [760, 562, 805, 627], [391, 575, 429, 645]]</box>.\n\nLayer 9: \nThere’s a <ref>person</ref><box>[[145, 572, 170, 620]]</box> among the layer, located among <ref>trees</ref><box>[[0, 202, 999, 624]]</box> that stand before a scenic <ref>mountain</ref><box>[[568, 252, 827, 383]]</box>, against the widespread <ref>sky</ref><box>[[0, 177, 999, 430]]</box>.\n\nLayer 10: \nHere one finds yet another <ref>person</ref><box>[[125, 571, 148, 622]]</box>, situated with <ref>trees</ref><box>[[0, 202, 999, 624]]</box> framing a distant <ref>mountain</ref><box>[[568, 252, 827, 383]]</box>, below the open <ref>sky</ref><box>[[0, 177, 999, 430]]</box>."
}
] |
coco/train2017/000000557394.jpg
|
00000023
| 426
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the upper right corner, there is a <ref>traffic light</ref><box>[[871, 457, 897, 521]]</box>. \n\nLayer 2: To the side, a <ref>light</ref><box>[[867, 299, 897, 327]]</box> is positioned. \n\nLayer 3: Nearby, a <ref>person</ref><box>[[769, 466, 885, 786]]</box> is on a <ref>bicycle</ref><box>[[769, 600, 881, 806]]</box>, <pred>carrying</pred> a <ref>backpack</ref><box>[[800, 513, 866, 588]]</box>.\n\nLayer 4: On the road, there are <ref>trucks</ref><box>[[449, 358, 782, 684], [771, 441, 821, 555]]</box>.\n\nLayer 5: Further away, a <ref>bus</ref><box>[[1, 212, 460, 767]]</box> is traveling."
}
] |
coco/train2017/000000516116.jpg
|
00000024
| 425
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the foreground, several <ref>girls</ref><box>[[959, 375, 999, 431], [206, 302, 237, 380], [479, 301, 508, 377], [797, 313, 835, 431], [471, 297, 491, 352], [447, 289, 969, 430]]</box> are visible. They are either sitting or standing on or near the <ref>playingfield</ref><box>[[0, 313, 999, 830]]</box>. A <ref>fence</ref><box>[[0, 267, 999, 377]]</box> surrounds the field, with <ref>trees</ref><box>[[0, 167, 999, 332]]</box> just beyond it.\n\nLayer 2: Further back, more <ref>girls</ref><box>[[789, 289, 916, 621], [536, 293, 560, 356], [206, 302, 237, 380], [479, 301, 508, 377], [618, 286, 738, 591], [447, 289, 969, 430]]</box> can be seen, actively involved in the game, under the same enclosure of the <ref>fence</ref><box>[[0, 267, 999, 377]]</box>.\n\nLayer 3: Slightly further back, a mix of participants <ref>girls</ref><box>[[716, 370, 736, 424], [206, 302, 237, 380], [479, 301, 508, 377], [797, 313, 835, 431], [447, 289, 969, 430]]</box> engage with their surroundings; they inhabit the dimensions defined by the <ref>playingfield</ref><box>[[0, 313, 999, 830]]</box> and boundary of the <ref>fence</ref><box>[[0, 267, 999, 377]]</box> with <ref>trees</ref><box>[[0, 167, 999, 332]]</box> in the distance.\n\nLayer 4: Expanded views reveal more players like <ref>girls</ref><box>[[67, 280, 112, 408], [479, 301, 508, 377], [471, 297, 491, 352], [447, 289, 969, 430]]</box> on field margins. The scene remains set by the recognizable features of the <ref>fence</ref><box>[[0, 267, 999, 377]]</box>.\n\nLayer 5: There are focal points on <ref>chairs</ref><box>[[510, 360, 538, 386], [440, 358, 502, 414]]</box> alongside bustling <ref>girls</ref><box>[[536, 293, 560, 356], [344, 325, 699, 704], [165, 310, 369, 736], [305, 316, 557, 726]]</box>. A <ref>sports ball</ref><box>[[547, 671, 622, 746]]</box> is engaged in play, highlighting interaction.\n\nLayer 6: Views encompass distant sections featuring <ref>girls</ref><box>[[143, 272, 216, 489], [206, 302, 237, 380], [479, 301, 508, 377], [447, 289, 969, 430]]</box>, indicating the near-perimeter confines of activity dictated by the <ref>playingfield</ref><box>[[0, 313, 999, 830]]</box> and guarded by <ref>fence</ref><box>[[0, 267, 999, 377]]</box>."
}
] |
coco/train2017/000000542442.jpg
|
00000025
| 321
| 500
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A portion of the <ref>sky</ref><box>[[921, 231, 941, 261]]</box> is visible above the scene.\n\nLayer 2: A <ref>man</ref><box>[[633, 247, 981, 810]]</box> wearing a <ref>tie</ref><box>[[785, 385, 833, 616]]</box> is standing, accompanied by another <ref>man</ref><box>[[1, 291, 261, 810]]</box> wearing a <ref>tie</ref><box>[[129, 442, 155, 667]]</box>.\n\nLayer 3: Another set of <ref>potted plants</ref><box>[[583, 435, 649, 527]]</box> is visible, with a nearby <ref>tree</ref><box>[[575, 177, 961, 491]]</box>.\n\nLayer 4: More <ref>potted plants</ref><box>[[581, 363, 661, 423]]</box> can be seen, alongside the same <ref>tree</ref><box>[[575, 177, 961, 491]]</box>, adjacent to a section of the <ref>building</ref><box>[[830, 177, 999, 652]]</box>.\n\nLayer 5: One of the middle <ref>men</ref><box>[[311, 237, 595, 810]]</box> is <pred>wearing</pred><box>[[423, 372, 463, 605]]</box> a <ref>tie</ref><box>[[423, 372, 463, 605]]</box>, walking beside another <ref>man</ref><box>[[633, 247, 981, 810]]</box>.\n\nLayer 6: A <ref>man</ref><box>[[311, 237, 595, 810]]</box> is <pred>holding</pred><box>[[287, 695, 365, 819]]</box> a <ref>handbag</ref><box>[[287, 695, 365, 819]]</box>.\n\nLayer 7: One <ref>tie</ref><box>[[129, 442, 155, 667]]</box> appears worn by the <ref>man</ref><box>[[1, 291, 261, 810]]</box>.\n\nLayer 8: The scene unfolds on a <ref>pavement</ref><box>[[0, 555, 999, 819]]</box>, surrounded by <ref>potted plants</ref><box>[[201, 401, 281, 527]]</box> and set in front of a backdrop containing a <ref>tree</ref><box>[[575, 177, 961, 491]]</box>.\n\nLayer 9: In the background, there is a long <ref>wall</ref><box>[[0, 177, 653, 620]]</box>, along with the towering <ref>tree</ref><box>[[575, 177, 961, 491]]</box> and nearby <ref>building</ref><box>[[830, 177, 999, 652]]</box>."
}
] |
coco/train2017/000000036333.jpg
|
00000026
| 640
| 425
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nThe <ref>person</ref><box>[[337, 270, 726, 750]]</box> is <pred>playing</pred><box>[[337, 270, 726, 750]]</box> with a <ref>skateboard</ref><box>[[461, 675, 547, 810]]</box> on the <ref>pavement</ref><box>[[167, 802, 830, 999]]</box>. The <ref>brick wall</ref><box>[[678, 0, 830, 878]]</box> with colorful graffiti is in the backdrop, alongside <ref>mirror-stuff</ref><box>[[688, 0, 814, 145]]</box> that's attached to the wall. \nLayer 2: \nAbove, a clear <ref>sky</ref><box>[[167, 0, 712, 155]]</box> is visible, completing the scene."
}
] |
coco/train2017/000000574350.jpg
|
00000027
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>handbag</ref><box>[[963, 415, 991, 486]]</box> rests on the <ref>pavement</ref><box>[[0, 514, 999, 874]]</box>. The <ref>umbrellas</ref><box>[[40, 324, 260, 469]]</box> and another group of <ref>umbrellas</ref><box>[[849, 269, 953, 308]]</box> are closer to the viewer.\n\nLayer 2: Two <ref>people</ref><box>[[889, 347, 995, 649]]</box> and <ref>people</ref><box>[[866, 333, 980, 669]]</box> appear within this layer, positioned as if together, creating a sense of closeness on the <ref>pavement</ref><box>[[0, 514, 999, 874]]</box>. The <ref>umbrellas</ref><box>[[40, 324, 260, 469]]</box> and <ref>umbrellas</ref><box>[[849, 269, 953, 308]]</box> are still seen in the scene.\n\nLayer 3: A prominent view of another <ref>handbag</ref><box>[[761, 486, 821, 578]]</box> accompanies a group of <ref>people</ref><box>[[154, 380, 290, 760]]</box> and <ref>people</ref><box>[[792, 360, 886, 711]]</box>. These individuals are walking on the <ref>pavement</ref><box>[[0, 514, 999, 874]]</box> beneath their <ref>umbrellas</ref><box>[[40, 324, 260, 469]]</box> and <ref>umbrellas</ref><box>[[714, 271, 913, 392]]</box>. Additionally, the <ref>handbag</ref><box>[[187, 428, 254, 578]]</box> figures prominently on an individual’s arm, with another <ref>umbrellas</ref><box>[[849, 269, 953, 308]]</box> above.\n\nLayer 4: A <ref>white dog</ref><box>[[633, 639, 730, 705]]</box> appears as if in motion across the <ref>pavement</ref><box>[[0, 514, 999, 874]]</box> near some <ref>grass</ref><box>[[0, 460, 999, 547]]</box>. Additional <ref>people</ref><box>[[693, 382, 817, 698]]</box> and <ref>people</ref><box>[[866, 333, 980, 669]]</box> also feature in this depth, surrounded by commonly seen <ref>umbrellas</ref><box>[[40, 324, 260, 469]]</box> and <ref>umbrellas</ref><box>[[849, 269, 953, 308]]</box>. Handbags<box>[[187, 428, 254, 578]]</box> add to the scape.\n\nLayer 5: The <ref>grass</ref><box>[[0, 460, 999, 547]]</box> retreats slightly further back, blending into the image’s mid-ground along with the predictable <ref>umbrellas</ref><box>[[40, 324, 260, 469]]</box>.\n\nLayer 6: Farther still, a towering <ref>wall</ref><box>[[0, 124, 999, 519]]</box> becomes an imposing boundary to the entire scene, continuing the motif of the <ref>umbrellas</ref><box>[[40, 324, 260, 469]]</box> throughout."
}
] |
coco/train2017/000000180480.jpg
|
00000028
| 640
| 426
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nA group of <ref>people</ref><box>[[772, 443, 831, 885], [521, 507, 624, 895], [683, 540, 788, 878]]</box> is present, observing the scene.\n\nLayer 2: \nAnother <ref>person</ref><box>[[606, 560, 676, 792]]</box> appears, <pred>leaning on</pred><box>[[606, 560, 676, 792]]</box><box>[[167, 210, 831, 899]]</box> a <ref>wall</ref><box>[[167, 210, 831, 899]]</box>. The <ref>skateboard</ref><box>[[300, 536, 499, 679]]</box> is seen in motion. The <ref>floor</ref><box>[[167, 846, 831, 999]]</box> supports these activities, and a <ref>backpack</ref><box>[[620, 767, 755, 891]]</box> is positioned on it.\n\nLayer 3: \nA <ref>man</ref><box>[[197, 33, 605, 636]]</box> is captured in mid-air as he <pred>plays</pred><box>[[197, 33, 605, 636]]</box><box>[[300, 536, 499, 679]]</box> and <pred>looks at</pred><box>[[197, 33, 605, 636]]</box><box>[[300, 536, 499, 679]]</box> the <ref>skateboard</ref><box>[[300, 536, 499, 679]]</box>, skillfully performing a trick. He is observed by several individuals, including <ref>people</ref><box>[[197, 719, 252, 900]]</box>. Overhead, <ref>lights</ref><box>[[167, 0, 831, 610]]</box> <pred>hanging from</pred><box>[[167, 0, 831, 610]]</box><box>[[167, 0, 831, 657]]</box> the <ref>ceiling</ref><box>[[167, 0, 831, 657]]</box> illuminate the scene.\n\nLayer 4: \nMore <ref>people</ref><box>[[176, 724, 218, 900]]</box> are oriented towards the event, contributing to the audience.\n\nLayer 5: \nThe scene is completed with <ref>people</ref><box>[[167, 728, 192, 899]]</box> engaged in watching the exciting trick being executed."
}
] |
coco/train2017/000000298051.jpg
|
00000029
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>car</ref><box>[[952, 547, 999, 581]]</box> is present. There is a <ref>crowd of people</ref><box>[[755, 564, 828, 749], [185, 485, 215, 535]]</box> gathered on the <ref>pavement</ref><box>[[60, 530, 927, 753]]</box>, standing near the <ref>portable stage</ref><box>[[288, 346, 694, 573]]</box>. This scene is placed in front of a <ref>building</ref><box>[[0, 165, 999, 641]]</box>. There is a <ref>road</ref><box>[[691, 557, 922, 627]]</box> close by.\n\nLayer 2: Another <ref>car</ref><box>[[746, 530, 880, 575]]</box> is visible near the <ref>building</ref><box>[[0, 165, 999, 641]]</box>. A small <ref>crowd of people</ref><box>[[213, 505, 257, 545]]</box> is also present in this layer. \n\nLayer 3: Part of the <ref>crowd of people</ref><box>[[591, 627, 699, 829]]</box> is further back compared to the previous groups.\n\nLayer 4: A <ref>clock</ref><box>[[547, 263, 598, 313]]</box> is hanging, with the surrounding <ref>building</ref><box>[[0, 165, 999, 641]]</box> and the <ref>sky</ref><box>[[226, 165, 999, 363]]</box> in the background.\n\nLayer 5: More of the <ref>crowd of people</ref><box>[[415, 627, 605, 823]]</box> gathers here.\n\nLayer 6: In this layer, another group of the <ref>crowd of people</ref><box>[[398, 641, 510, 793]]</box> is visible.\n\nLayer 7: Additional members of the <ref>crowd of people</ref><box>[[363, 639, 457, 770], [755, 564, 828, 749], [786, 625, 913, 821], [621, 623, 827, 825]]</box> are also present.\n\nLayer 8: Another section of the <ref>crowd of people</ref><box>[[324, 645, 423, 825]]</box> is seen.\n\nLayer 9: This layer shows a smaller <ref>crowd of people</ref><box>[[24, 650, 140, 829], [621, 623, 827, 825]]</box> still present.\n\nLayer 10: A larger group of the <ref>crowd of people</ref><box>[[229, 649, 404, 831], [621, 623, 827, 825]]</box> can be seen here.\n\nLayer 11: Another smaller group of the <ref>crowd of people</ref><box>[[137, 625, 273, 831]]</box> is visible.\n\nLayer 12: Finally, additional parts of the <ref>crowd of people</ref><box>[[117, 617, 184, 784], [621, 623, 827, 825]]</box> make up the most distant layer."
}
] |
coco/train2017/000000257275.jpg
|
00000030
| 354
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nA <ref>person</ref><box>[[234, 393, 421, 597]]</box> is positioned closest to the camera, with <ref>people</ref><box>[[797, 225, 838, 369]]</box> and other <ref>people</ref><box>[[771, 223, 807, 362], [98, 223, 999, 639]]</box> standing on a <ref>snow</ref><box>[[0, 223, 999, 775]]</box>-covered slope. \n\nLayer 2: \nA different group of <ref>people</ref><box>[[618, 243, 658, 341]]</box> is positioned slightly further back on the snow slope, observing the event. They also stand on the same <ref>snow</ref><box>[[0, 223, 999, 775]]</box>.\n\nLayer 3: \nThe scene transitions into a layer where <ref>skis</ref><box>[[316, 527, 454, 624]]</box> lie <pred>on</pred><box>[[316, 527, 454, 624]]</box><box>[[154, 566, 999, 775]]</box> the surface of the <ref>water</ref><box>[[154, 566, 999, 775]]</box>. The <ref>snow</ref><box>[[0, 223, 999, 775]]</box> is also <pred>on</pred><box>[[0, 223, 999, 775]]</box><box>[[154, 566, 999, 775]]</box> with the same layer of water, illustrating the mix of snow and water.\n\nLayer 4: \nFinally, observing from a greater distance, the original <ref>person</ref><box>[[234, 393, 421, 597]]</box> is depicted <pred>playing</pred><box>[[234, 393, 421, 597]]</box> on the skis, with vast stretches of the <ref>snow</ref><box>[[0, 223, 999, 775]]</box> in the backdrop, signifying their path down the snowy slope to the pool of water."
}
] |
coco/train2017/000000575949.jpg
|
00000031
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nThere is a <ref>car</ref><box>[[959, 444, 999, 529]]</box> parked in the scene. A <ref>kite</ref><box>[[1, 124, 501, 742]]</box> is visible, adding a dynamic element to the sky. A <ref>banner</ref><box>[[780, 124, 947, 442]]</box>, likely a decoration or advertising signage, can also be seen. Additionally, the scene features a <ref>building</ref><box>[[188, 188, 638, 522]]</box> along with the <ref>sky</ref><box>[[310, 124, 999, 458]]</box>, serving as a backdrop to the festivities.\n\nLayer 2: \nSeveral <ref>people</ref> are partaking in the scene: one <ref>person</ref><box>[[730, 424, 816, 637]]</box> and another <ref>person</ref><box>[[871, 424, 989, 715]]</box> are walking together, potentially suggesting companionship. A third <ref>person</ref><box>[[683, 377, 919, 598]]</box> adds to the ensemble, all walking on the <ref>road</ref><box>[[0, 497, 999, 874]]</box>. Another <ref>kite</ref><box>[[799, 424, 853, 519]]</box> is flying, contributing to a potentially festive mood. Furthermore, a <ref>car</ref><box>[[394, 478, 463, 569]]</box> is seen on the road, blending into the bustling scene.\n\nLayer 3: \nA pair of <ref>people</ref> are observed, including a <ref>person</ref><box>[[174, 319, 426, 862]]</box> and another <ref>person</ref><box>[[396, 411, 669, 736]]</box>, both possibly indicating friendship as they walk in close proximity. A <ref>kite</ref><box>[[449, 555, 969, 874]]</box> decorates the atmosphere, hovering in the broader scene. Additionally, another instance of the aforementioned <ref>banner</ref><box>[[780, 124, 947, 442]]</box> is visible, suggesting the importance or celebration of the event.\n\nLayer 4:\nLastly, a <ref>car</ref><box>[[103, 536, 207, 673]]</box> is part of the visual setup in the further depths of the image. Also, an additional instance of the previously mentioned <ref>kite</ref><box>[[1, 124, 501, 742]]</box> and <ref>banner</ref><box>[[780, 124, 947, 442]]</box>, enhancing the theme or possibly signaling of the event, is featured in this layer."
}
] |
coco/train2017/000000411674.jpg
|
00000032
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nA group of two <ref>people</ref><box>[[657, 589, 721, 764], [716, 600, 778, 763]]</box> are walking together, occupying the forefront of the scene. To their side, a red <ref>double-decker bus</ref><box>[[188, 304, 686, 756]]</box> is visible. The <ref>window</ref> of a <ref>building</ref><box>[[53, 124, 675, 461]]</box> is close by, adding an architectural element. Additionally, there's an adjacent <ref>fence</ref><box>[[808, 594, 999, 815]]</box> and a <ref>traffic light</ref><box>[[910, 409, 959, 486]]</box> seen in the urban setting.\n\nLayer 2: \nA <ref>bicycle</ref><box>[[568, 669, 636, 778]]</box> being ridden by another <ref>person</ref><box>[[561, 589, 635, 745]]</box> is moving along the <ref>road</ref><box>[[0, 682, 999, 874]]</box>. This scene indicates an active street environment. The <ref>fence</ref><box>[[808, 594, 999, 815]]</box> extends into this layer as well.\n\nLayer 3: \nA third <ref>person</ref> is seen driving or present with the <ref>double-decker bus</ref><box>[[246, 542, 298, 595]]</box> in the scenery. The <ref>road</ref><box>[[0, 682, 999, 874]]</box>, on which all this activity is taking place, stretches out, maintaining its visible presence throughout the image.\n\nLayer 4: \nThe overarching background is dominated by the <ref>building</ref><box>[[0, 124, 999, 607]]</box>, providing an urban environment that contextualizes all the activities occurring."
}
] |
coco/train2017/000000355304.jpg
|
00000033
| 375
| 500
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the scene, a <ref>wall</ref><box>[[0, 123, 999, 399]]</box> forms the backdrop, next to a <ref>person</ref><box>[[929, 271, 999, 402]]</box> engaged in an activity. Nearby, there is a visible <ref>cardboard</ref><box>[[199, 199, 641, 245]]</box> placed within a <ref>cabinet</ref><box>[[109, 223, 789, 383]]</box>.\n\nLayer 2: Moving outward, the scene expands to include several <ref>people</ref><box>[[925, 339, 999, 510]]</box><box>[[507, 171, 623, 402]]</box><box>[[745, 255, 999, 738]]</box>, seated or standing around a <ref>dining table</ref><box>[[57, 449, 675, 551]]</box><box>[[17, 477, 183, 766]]</box> on the <ref>floor</ref><box>[[0, 540, 999, 801]]</box>.\n\nLayer 3: More activity is captured as additional <ref>people</ref><box>[[507, 171, 623, 402]]</box><box>[[745, 255, 999, 738]]</box><box>[[577, 219, 779, 639]]</box> and <ref>chairs</ref><box>[[17, 477, 183, 766]]</box> are situated around the <ref>dining table</ref><box>[[57, 449, 675, 551]]</box>.\n\nLayer 4: A distinct group features other <ref>people</ref><box>[[814, 233, 861, 265]]</box> around the expanding space, comfortable in their surroundings.\n\nLayer 5: Additional <ref>people</ref><box>[[699, 243, 787, 349]]</box> subtly blend into the scene, adding more depth.\n\nLayer 6: Overhead, a <ref>light</ref><box>[[449, 123, 511, 143]]</box> illuminates the setting below.\n\nLayer 7: In the boy's hand, a <ref>donut</ref><box>[[274, 429, 513, 533]]</box>, garnished with sprinkles, invites attention.\n\nLayer 8: Above, the <ref>ceiling</ref><box>[[271, 123, 365, 155]]</box> spans across the entire setting.\n\nLayer 9: Anchor of the scene, a <ref>young boy</ref><box>[[179, 161, 797, 862]]</box>, seated and enjoying his donut.\n\nLayer 10: Below, the <ref>floor</ref><box>[[0, 540, 999, 801]]</box> supports the weight of the communal gathering.\n\nLayer 11: Witnesses to the event are more <ref>people</ref><box>[[117, 313, 230, 423]]</box>, blending into the spatial distribution.\n\nLayer 12: A standalone <ref>door-stuff</ref><box>[[107, 373, 139, 408]]</box> presents an entranceway or partition.\n\nLayer 13: Integral parts of the seating arrangement, <ref>chairs</ref><box>[[0, 474, 67, 603]]</box> are scattered about.\n\nLayer 14: At a distance lies a <ref>window</ref><box>[[0, 189, 123, 373]]</box>, providing a glance into the outside world."
}
] |
coco/train2017/000000261805.jpg
|
00000034
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: At the forefront, there is a <ref>grass</ref><box>[[93, 124, 999, 554]]</box> area.\n\nLayer 2: The scene includes several <ref>people</ref><box>[[412, 532, 497, 874], [167, 506, 249, 874], [666, 529, 789, 761]]</box> on a <ref>pavement</ref><box>[[0, 597, 974, 874]]</box>. Some are <pred>carrying</pred><box>[[419, 773, 546, 861]]</box> <ref>suitcases</ref><box>[[419, 773, 546, 861]]</box>, and a couple others <pred>carrying</pred><box>[[37, 725, 181, 864]]</box> <ref>suitcases</ref><box>[[37, 725, 181, 864]]</box>. Additionally, some people have <pred>wearing</pred><box>[[412, 532, 497, 874]]</box><ref>handbags</ref><box>[[382, 584, 476, 733]]</box> and one is <pred>carrying</pred><box>[[127, 580, 184, 705]]</box><ref>backpacks</ref><box>[[127, 580, 184, 705]]</box>, while another is <pred>carrying</pred><box>[[657, 725, 769, 839]]</box><ref>backpack</ref><box>[[657, 725, 769, 839]]</box>. In the background of this layer is a <ref>fence</ref><box>[[744, 561, 999, 874]]</box>.\n\nLayer 3: There are more <ref>people</ref><box>[[334, 516, 415, 815]]</box> further back, standing on the <ref>pavement</ref><box>[[0, 597, 974, 874]]</box>, with the <ref>fence</ref><box>[[744, 561, 999, 874]]</box> still visible.\n\nLayer 4: A <ref>building</ref><box>[[0, 376, 999, 694]]</box> with visible windows is present.\n\nLayer 5: Above, the <ref>ceiling</ref><box>[[0, 124, 999, 438]]</box> is constructed from a geometric metal structure, allowing the partially visible sky to be seen."
}
] |
coco/train2017/000000499679.jpg
|
00000035
| 304
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nIn this layer, we observe a <ref>batter</ref><box>[[176, 333, 410, 686]]</box> <pred>swinging</pred><box>[[176, 333, 410, 686]]</box> a <ref>baseball bat</ref><box>[[84, 304, 212, 375]]</box> on the <ref>gravel</ref><box>[[0, 533, 999, 639]]</box>, which is part of the broader <ref>playingfield</ref><box>[[0, 558, 999, 736]]</box>. \n\nLayer 2: \nIn this layer, an <ref>umpire</ref><box>[[714, 443, 988, 728]]</box> can be seen close to a <ref>catcher</ref><box>[[483, 458, 758, 717]]</box> <pred>wearing</pred><box>[[485, 471, 541, 519]]</box> a <ref>baseball glove</ref><box>[[485, 471, 541, 519]]</box>.\n\nLayer 3: \nHere, we note various <ref>benches</ref><box>[[462, 549, 474, 566], [1, 500, 162, 528]]</box> that are positioned along the sidelines.\n\nLayer 4: \nThis layer features several <ref>chairs</ref><box>[[7, 302, 43, 329], [303, 296, 335, 319], [816, 313, 850, 324], [120, 296, 155, 314], [210, 290, 243, 313], [716, 305, 747, 318]]</box>, which are arranged around the perimeter of the <ref>playingfield</ref><box>[[0, 558, 999, 736]]</box>.\n\nLayer 5: \nFinally, in the background, we observe a <ref>fence</ref><box>[[0, 391, 999, 620]]</box> attached to which several <ref>people</ref><box>[[553, 369, 630, 588], [355, 358, 421, 558]]</box> can be seen standing. Additionally, a <ref>wall</ref><box>[[0, 366, 999, 579]]</box> encloses the playing area."
}
] |
coco/train2017/000000167110.jpg
|
00000036
| 425
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \n- A <ref>person</ref><box>[[398, 248, 636, 472]]</box> is seen <pred>playing</pred><box>[[398, 248, 636, 472]]</box> with a <ref>skateboard</ref><box>[[623, 405, 736, 541]]</box>.\n- A <ref>skateboard</ref><box>[[623, 405, 736, 541]]</box> is <pred>held</pred><box>[[623, 405, 736, 541]]</box> by another <ref>person</ref><box>[[707, 249, 819, 541]]</box>.\n- The <ref>fence</ref><box>[[0, 354, 999, 621]]</box> encloses the <ref>playingfield</ref><box>[[0, 514, 999, 830]]</box>.\n\nLayer 2:\n- A distant <ref>roof</ref><box>[[821, 323, 930, 388]]</box> can be seen with the <ref>sky</ref><box>[[0, 167, 607, 299]]</box> above it.\n- A <ref>tree</ref><box>[[0, 167, 999, 405]]</box> is observable within the scene.\n\nLayer 3:\n- A person <pred>stands on</pred><box>[[398, 248, 636, 472]]</box> the <ref>playingfield</ref><box>[[0, 514, 999, 830]]</box> with a performing action.\n\nLayer 4:\n- A <ref>person</ref><box>[[393, 394, 471, 474]]</box> appears to be <pred>looking at</pred><box>[[393, 394, 471, 474]]</box> another <ref>person</ref><box>[[398, 248, 636, 472]]</box>.\n- Another <ref>person</ref><box>[[707, 249, 819, 541]]</box> is also visible.\n\nLayer 5:\n- A couple of <ref>cars</ref><box>[[184, 548, 223, 570], [23, 550, 171, 580]]</box> are <pred>parked</pred><box>[[184, 548, 223, 570], [23, 550, 171, 580]]</box> in the distance beyond the playingfield.\n\nLayer 6:\n- The remaining area is the overall <ref>playingfield</ref><box>[[0, 514, 999, 830]]</box>, giving an open space for the activities seen."
}
] |
coco/train2017/000000199640.jpg
|
00000037
| 331
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: There is a <ref>person</ref><box>[[856, 294, 892, 377]]</box> prominently in the scene followed by another <ref>person</ref><box>[[305, 322, 352, 444]]</box> in close proximity. Nearby is a different <ref>person</ref><box>[[548, 344, 617, 472]]</box> and yet another <ref>person</ref><box>[[62, 329, 195, 485]]</box>. In the background of these individuals, a <ref>car</ref><box>[[418, 319, 838, 504]]</box> makes an appearance while a <ref>building</ref><box>[[0, 240, 999, 400]]</box> lines the street, establishing an urban atmosphere.\n\nLayer 2: A noteworthy <ref>person</ref><box>[[827, 351, 995, 727]]</box> is significant in this layer, joined by a <ref>motorcycle</ref><box>[[154, 486, 541, 748]]</box>. Another <ref>person</ref><box>[[487, 327, 553, 472]]</box> and adjoining <ref>motorcycle</ref><box>[[535, 499, 568, 552]]</box> further decorate the scene.\n\nLayer 3: Another <ref>motorcycle</ref><box>[[49, 458, 196, 652]]</box> gains attention. Among the others, a <ref>person</ref><box>[[305, 322, 352, 444]]</box> is visible, augmented by another arm of engaging <ref>people</ref><box>[[548, 344, 617, 472], [487, 327, 553, 472]]</box> alongside <ref>motorcycles</ref><box>[[689, 464, 999, 692]]</box>.\n\nLayer 4: The presence of a sole <ref>person</ref><box>[[449, 416, 688, 750]]</box> stands out in this layer. \n\nLayer 5: An individual with a <ref>handbag</ref><box>[[343, 438, 380, 593]]</box> adds to the complex integration; <ref>people</ref><box>[[548, 344, 617, 472], [487, 327, 553, 472]]</box> are interspersed among a densely packed row.\n\nLayer 6: Here emerges another scene of <ref>people</ref><box>[[257, 371, 337, 497]]</box> surrounded by others <ref>people</ref><box>[[548, 344, 617, 472], [487, 327, 553, 472]]</box> with a <ref>motorcycle</ref><box>[[689, 464, 999, 692]]</box> in visual reach.\n\nLayer 7: Evident is one <ref>person</ref><box>[[255, 372, 444, 656]]</box> interacting with another <ref>person</ref><box>[[487, 327, 553, 472]]</box>, close to <ref>motorcycles</ref><box>[[154, 486, 541, 748], [535, 499, 568, 552]]</box> linking the scenes from prior layers together cohesively.\n\nLayer 8: Beginning with an opening <ref>motorcycle</ref><box>[[1, 440, 56, 489]]</box>, supporting a <ref>person</ref><box>[[305, 322, 352, 444]]</box> is another visual highlight here, coupled with familiar outlines from <ref>people</ref><box>[[548, 344, 617, 472], [487, 327, 553, 472]]</box>. Similarly, assembled are <ref>people</ref><box>[[775, 352, 878, 489]]</box> standing near the substantial <ref>car</ref><box>[[418, 319, 838, 504]]</box> and necessary <ref>building</ref><box>[[0, 240, 999, 400]]</box> backdrop.\n\nLayer 9: Bringing up the rear, discernible is a <ref>motorcycle</ref><box>[[0, 415, 109, 597]]</box> teamed with <ref>people</ref><box>[[305, 322, 352, 444], [548, 344, 617, 472], [487, 327, 553, 472]]</box> combined with <ref>people</ref><box>[[775, 352, 878, 489]]</box>, closing the depth narrative with the urban jungle framed by the <ref>building</ref><box>[[0, 240, 999, 400]]</box> setting."
}
] |
coco/train2017/000000288164.jpg
|
00000038
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1:\n- A <ref>chair</ref><box>[[967, 766, 999, 853]]</box> is positioned nearby.\n- <ref>Another chair</ref><box>[[766, 666, 972, 864]]</box> is closely situated.\n- A <ref>dining table</ref><box>[[641, 586, 997, 862]]</box> stretches out, occupying ample space.\n- The <ref>table</ref><box>[[28, 571, 999, 874]]</box> spans across the room.\n\nLayer 2:\n- A <ref>chair</ref><box>[[858, 639, 913, 756]]</box> is part of the room's furnishings.\n- A <ref>person</ref><box>[[279, 421, 468, 865]]</box> is seen in the classroom.\n- Another <ref>person</ref><box>[[59, 406, 251, 864]]</box> stands nearby.\n- There's a <ref>dining table</ref><box>[[199, 631, 524, 672]]</box>, central to the scene.\n- A <ref>rug</ref><box>[[0, 659, 792, 874]]</box> lies underfoot.\n- A <ref>chair</ref><box>[[251, 601, 338, 769]]</box> adds to the seating arrangement.\n- Another <ref>chair</ref><box>[[202, 574, 251, 634]]</box> accompanies it.\n- Adjacent is a <ref>chair</ref><box>[[407, 598, 502, 776]]</box>.\n- The <ref>walls</ref><box>[[0, 124, 999, 822]]</box> encompass the setting.\n\nLayer 3:\n- A <ref>chair</ref><box>[[824, 611, 869, 712]]</box> forms part of the relaxing area.\n- Matching in proximity is another <ref>chair</ref><box>[[279, 421, 468, 865]]</box>.\n- Standing within is another <ref>person</ref><box>[[59, 406, 251, 864]]</box>.\n- The scene includes a <ref>dining table</ref><box>[[199, 631, 524, 672]]</box>.\n\nLayer 4:\n- Situated further back is a <ref>chair</ref><box>[[797, 592, 833, 678]]</box>.\n- Near it stands a <ref>person</ref><box>[[279, 421, 468, 865]]</box>.\n\nLayer 5:\n- At eye-level, a <ref>shelf</ref><box>[[705, 541, 774, 614]]</box> holds various items.\n\nLayer 6:\n- A <ref>tv</ref><box>[[682, 491, 771, 564]]</box> is mounted on the wall.\n\nLayer 7:\n- Toward the back, <ref>window-blinds</ref><box>[[503, 389, 679, 578]]</box> shield sunlight.\n\nLayer 8:\n- A <ref>toothbrush</ref><box>[[113, 605, 148, 613]]</box> is held in hand.\n- Positioned alongside is a <ref>cup</ref><box>[[913, 797, 952, 853]]</box>, resting on a surface.\n\nLayer 9:\n- A diminutive <ref>dining table</ref><box>[[0, 630, 70, 656]]</box> is subtly incorporated."
}
] |
coco/train2017/000000486717.jpg
|
00000039
| 397
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nThere is a <ref>baseball glove</ref><box>[[522, 600, 572, 656]]</box> located up close, <pred>worn by</pred><box>[[522, 600, 572, 656]]</box><box>[[519, 472, 758, 758]]</box> a <ref>catcher</ref>. Behind, a <ref>dirt</ref><box>[[0, 188, 999, 808]]</box> area is partially covered by the <ref>playingfield</ref><box>[[0, 188, 999, 644]]</box>. A <ref>batter</ref><box>[[137, 330, 482, 721]]</box> is positioned ready to <pred>swing</pred><box>[[137, 330, 482, 721]]</box><box>[[341, 258, 427, 385]]</box>, and an <ref>umpire</ref><box>[[705, 368, 920, 633]]</box> stands <pred>touching</pred><box>[[705, 368, 920, 633]]</box><box>[[519, 472, 758, 758]]</box> the catcher.\n\nLayer 2: \nA <ref>baseball bat</ref><box>[[341, 258, 427, 385]]</box> is in the hands of the <ref>batter</ref><box>[[137, 330, 482, 721]]</box>, appearing further away than other closer objects. The <ref>catcher</ref><box>[[519, 472, 758, 758]]</box>, being <pred>next to</pred><box>[[519, 472, 758, 758]]</box><box>[[137, 330, 482, 721]]</box> the batter, creates a scene of action in the game."
}
] |
coco/train2017/000000301554.jpg
|
00000040
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: There is a <ref>man</ref><box>[[578, 193, 728, 367]]</box>.\n\nLayer 2: Nearby, a <ref>person</ref><box>[[173, 491, 266, 686]]</box> is visible, and the expansive <ref>dirt</ref><box>[[150, 616, 999, 874]]</box> stretches across the ground.\n\nLayer 3: A short distance away, a <ref>person</ref><box>[[99, 496, 249, 708]]</box> can be seen, interacting with the surroundings.\n\nLayer 4: Close to the ground, a <ref>bench</ref><box>[[243, 625, 362, 739]]</box> lies positioned within the scene.\n\nLayer 5: In the middle of the scene, a large <ref>elephant</ref><box>[[176, 340, 895, 806]]</box> commands attention. Additionally, a group of <ref>people</ref><box>[[0, 438, 249, 865]]</box> stands nearby.\n\nLayer 6: One of the <ref>people</ref><box>[[0, 438, 249, 865]]</box> is seen with a <ref>handbag</ref><box>[[1, 581, 95, 714]]</box>.\n\nLayer 7: In the further distance, additional <ref>people</ref><box>[[0, 742, 120, 874]]</box> appear within the view.\n\nLayer 8: The background is adorned with lush <ref>trees</ref><box>[[0, 124, 999, 592]]</box>, completing the serene and natural setting."
}
] |
coco/train2017/000000279083.jpg
|
00000041
| 481
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the closest layer, a <ref>baseball player</ref><box>[[632, 475, 813, 742]]</box> is prominently visible, <pred>holding</pred><box>[[632, 475, 813, 742]]</box><box>[[617, 586, 655, 610]]</box> a <ref>baseball glove</ref><box>[[617, 586, 655, 610]]</box>. This <ref>player</ref><box>[[632, 475, 813, 742]]</box> is <pred>standing on</pred><box>[[0, 408, 999, 874]]</box> the <ref>playingfield</ref><box>[[0, 408, 999, 874]]</box>, which extends broadly in the scene. Part of the <ref>sand</ref><box>[[0, 458, 999, 806]]</box> can also be perceived, showing the boundary of the field.\n\nLayer 2: A small <ref>sports ball</ref><box>[[341, 351, 352, 364]]</box> is visible in the path of another <ref>player</ref><box>[[346, 303, 438, 525]]</box> who is <pred>swinging</pred><box>[[346, 303, 438, 525]]</box><box>[[326, 375, 379, 422]]</box> a <ref>baseball bat</ref><box>[[326, 375, 379, 422]]</box>, creating a dynamic action in the scene. The <ref>playingfield</ref><box>[[0, 408, 999, 874]]</box> continues to provide the surface for all activities.\n\nLayer 3: Another <ref>baseball glove</ref><box>[[268, 407, 298, 432]]</box> can be seen, subtly in the layers of this sports occasion.\n\nLayer 4: A visible <ref>player</ref><box>[[226, 378, 351, 505]]</box> is <pred>holding</pred><box>[[226, 378, 351, 505]]</box><box>[[268, 407, 298, 432]]</box> a <ref>baseball glove</ref><box>[[268, 407, 298, 432]]</box> and is also <pred>standing on</pred><box>[[226, 378, 351, 505]]</box><box>[[0, 458, 999, 806]]</box> the <ref>sand</ref><box>[[0, 458, 999, 806]]</box>.\n\nLayer 5: The <ref>dirt</ref><box>[[0, 400, 999, 510]]</box> area is visible behind the batter, adding texture to the ground layer of the scene.\n\nLayer 6: In the furthest backdrop, a <ref>banner</ref><box>[[0, 260, 999, 432]]</box> extends across the top portion of the image."
}
] |
coco/train2017/000000434618.jpg
|
00000042
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: Two <ref>people</ref><box>[[279, 372, 636, 874]]</box><box>[[532, 355, 999, 864]]</box> in ski gear are closest to the camera. These individuals are smiling and posing for the camera while seated in a ski lift.\n\nLayer 2: Another <ref>person</ref><box>[[134, 299, 433, 862]]</box>, also in ski gear, is situated slightly farther from the camera.\n\nLayer 3: There is <ref>snow</ref><box>[[112, 288, 999, 731]]</box> covering the ground beneath the lift.\n\nLayer 4: In this layer, there are additional <ref>skis</ref><box>[[0, 126, 140, 864]]</box> visible, and portions of the <ref>people</ref><box>[[532, 355, 999, 864]]</box> and <ref>mountains</ref><box>[[112, 202, 999, 633]]</box> in the distance can also be seen.\n\nLayer 5: The <ref>sky</ref><box>[[0, 124, 999, 382]]</box> is present, forming the backdrop of the image, stretching across the horizon and <pred>over</pred><box>[[0, 124, 999, 382]]</box><box>[[112, 288, 999, 731], [112, 202, 999, 633]]</box> both the snowy landscape and the distant mountains."
}
] |
coco/train2017/000000462750.jpg
|
00000043
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>curtain</ref><box>[[939, 320, 999, 591]]</box> is highlighted in the closest view, alongside a <ref>rug</ref><box>[[74, 717, 936, 831]]</box> sprawled across the floor.\n\nLayer 2: A <ref>laptop</ref><box>[[914, 630, 999, 686]]</box> rests on a <ref>couch</ref><box>[[847, 586, 999, 823]]</box>, portraying a workspace area. The prominent <ref>woman</ref><box>[[388, 557, 652, 697]]</box> is <pred>lying on</pred><box>[[388, 557, 652, 697]]</box><box>[[149, 525, 885, 800]]</box> a <ref>bed</ref><box>[[149, 525, 885, 800]]</box>, adding to the softness of the space.\n\nLayer 3: A <ref>light</ref><box>[[210, 452, 975, 659]]</box> brightens the room, setting the atmosphere. \n\nLayer 4: Overhead, the <ref>ceiling</ref><box>[[168, 165, 999, 228]]</box> provides cover for the inviting interior. \n\nLayer 5: A <ref>door</ref><box>[[0, 194, 191, 831]]</box> leads into the room, part of the domestic landscape."
}
] |
coco/train2017/000000045366.jpg
|
00000044
| 376
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: The scene begins with a <ref>train</ref><box>[[113, 392, 497, 786]]</box> that is <pred>beside</pred><box>[[113, 392, 497, 786]]</box><box>[[443, 392, 526, 687]]</box> another <ref>train</ref><box>[[443, 392, 526, 687]]</box>, which is <pred>parked on</pred><box>[[443, 392, 526, 687]]</box><box>[[0, 372, 999, 792]]</box> the <ref>railroad</ref><box>[[0, 372, 999, 792]]</box>. Along the area, there's a <ref>pavement</ref><box>[[508, 424, 588, 792]]</box> <pred>beside</pred><box>[[508, 424, 588, 792]]</box><box>[[0, 372, 999, 792]]</box> the railroad. The <ref>road</ref><box>[[674, 404, 999, 539]]</box> features a <ref>truck</ref><box>[[975, 432, 999, 466]]</box> lying on it.\n\nLayer 2: A <ref>bus</ref><box>[[895, 469, 950, 499]]</box> lies on the <ref>road</ref><box>[[674, 404, 999, 539]]</box>. Meanwhile, more <ref>trains</ref><box>[[113, 392, 497, 786], [535, 386, 999, 789]]</box> are visible with <ref>gravel</ref><box>[[563, 427, 999, 792]]</box> <pred>beside</pred><box>[[563, 427, 999, 792]]</box><box>[[0, 372, 999, 792]]</box> the <ref>railroad</ref><box>[[0, 372, 999, 792]]</box>.\n\nLayer 3: Elevated above the streetscape are <ref>traffic lights</ref><box>[[878, 422, 900, 460]]</box>, adding to the urban detail. More <ref>trains</ref><box>[[113, 392, 497, 786], [535, 386, 999, 789]]</box> stand highlighted by surrounding <ref>gravel</ref><box>[[563, 427, 999, 792]]</box>.\n\nLayer 4: Setting the urban backdrop is a <ref>wall</ref><box>[[719, 386, 743, 419]]</box>, part of the surrounding <ref>buildings</ref><box>[[0, 280, 999, 482]]</box>. Above it all is the expansive <ref>sky</ref><box>[[0, 206, 999, 363]]</box> that <pred>over</pred><box>[[0, 206, 999, 363]]</box><box>[[0, 322, 699, 402], [643, 296, 999, 432]]</box> a <ref>mountain</ref><box>[[0, 322, 699, 402]]</box> and a solitary <ref>tree</ref><box>[[643, 296, 999, 432]]</box>.\n\nLayer 5: Additional <ref>traffic light</ref><box>[[664, 413, 694, 469]]</box>, closer to the buildings, contribute to the transport network. Trains and <ref>gravel</ref><box>[[563, 427, 999, 792]]</box> enhance the layered complexity of the scene.\n\nLayer 6: Another <ref>train</ref><box>[[0, 433, 295, 600]]</box> moves through the area alongside others. Together these layers compound to create a multifaceted view of this <ref>train station</ref><box>[[0, 372, 999, 792]]</box>, animated and bustling in nature."
}
] |
coco/train2017/000000305224.jpg
|
00000045
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>table</ref><box>[[810, 124, 999, 334]]</box> is present.\n\nLayer 2: Moving further back, there is a <ref>backpack</ref><box>[[733, 127, 838, 346]]</box> <pred>on</pred><box>[[733, 127, 838, 346]]</box><box>[[0, 142, 999, 874]]</box> the <ref>floor</ref><box>[[0, 142, 999, 874]]</box>. A <ref>person</ref><box>[[362, 126, 752, 353]]</box> is <pred>sitting on</pred><box>[[362, 126, 752, 353]]</box><box>[[268, 124, 622, 314]]</box> a different couch in the background. A <ref>chair</ref><box>[[0, 127, 274, 344]]</box> is visible as well.\n\nLayer 3: Further into the scene, a <ref>handbag</ref><box>[[617, 391, 977, 855]]</box> is <pred>sitting on</pred><box>[[617, 391, 977, 855]]</box><box>[[0, 142, 999, 874]]</box> the floor. A large pink <ref>couch</ref><box>[[177, 294, 816, 861]]</box> is present.\n\nLayer 4: At this depth, a <ref>cell phone</ref><box>[[518, 364, 600, 416]]</box> is <pred>on</pred><box>[[518, 364, 600, 416]]</box><box>[[177, 294, 816, 861]]</box> the couch beside a <ref>woman</ref><box>[[67, 175, 735, 789]]</box> who is <pred>lying on</pred><box>[[67, 175, 735, 789]]</box><box>[[177, 294, 816, 861]]</box> the large pink couch. She is <pred>looking at</pred><box>[[67, 175, 735, 789]]</box><box>[[273, 188, 508, 471]]</box> and <pred>touching</pred><box>[[67, 175, 735, 789]]</box><box>[[273, 188, 508, 471]]</box> a <ref>laptop</ref><box>[[273, 188, 508, 471]]</box> that is <pred>in front of</pred><box>[[67, 175, 735, 789]]</box><box>[[273, 188, 508, 471]]</box> her.\n\nLayer 5: Another pink <ref>couch</ref><box>[[3, 124, 81, 246]]</box> is visible.\n\nLayer 6: Lastly, the <ref>laptop</ref><box>[[273, 188, 508, 471]]</box> remains at the furthest visible depth point in this scene."
}
] |
coco/train2017/000000345142.jpg
|
00000046
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>chair</ref><box>[[858, 246, 917, 338]]</box> is present on the <ref>playingfield</ref><box>[[0, 298, 999, 874]]</box>, positioned <pred>beside</pred><box>[[782, 258, 853, 337]]</box><box>[[858, 246, 917, 338]]</box> another <ref>chair</ref><box>[[782, 258, 853, 337]]</box>. A <ref>banner</ref><box>[[0, 238, 674, 382]]</box> is also visible on the playing field.\n\nLayer 2: A <ref>wall</ref><box>[[780, 124, 872, 208]]</box> can be seen, forming part of the background structure.\n\nLayer 3: The main subject, a <ref>tennis player</ref><box>[[636, 341, 838, 691]]</box>, is seen over the <ref>playingfield</ref><box>[[0, 298, 999, 874]]</box>.\n\nLayer 4: A <ref>tennis racket</ref><box>[[600, 311, 660, 433]]</box> accompanies the tennis player on the playing field.\n\nLayer 5: A small <ref>sports ball</ref><box>[[46, 346, 74, 361]]</box> is airborne above the playing field, while multiple <ref>chairs</ref><box>[[782, 258, 853, 337]]</box> are arranged on it. The <ref>banner</ref><box>[[0, 238, 674, 382]]</box> adds to the scene's decor.\n\nLayer 6: In the background, there are <ref>stairs</ref><box>[[124, 124, 229, 269]]</box>, indicating part of the indoor setting of the venue."
}
] |
coco/train2017/000000302424.jpg
|
00000047
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>child</ref><box>[[721, 403, 885, 745]]</box> is present in the scene. Close by, there is a <ref>cow</ref><box>[[72, 191, 503, 790]]</box>.\n\nLayer 2: A part of a <ref>wall</ref><box>[[647, 124, 821, 214]]</box> is present.\n\nLayer 3: There is a <ref>flower</ref><box>[[610, 200, 700, 272]]</box> attached to the wall.\n\nLayer 4: The <ref>calf</ref><box>[[541, 376, 788, 675]]</box> statue is positioned on the <ref>pavement</ref><box>[[0, 263, 999, 874]]</box>.\n\nLayer 5: Nearby, a <ref>bench</ref><box>[[491, 271, 763, 386]]</box> is found standing, with the pavement stretching below.\n\nLayer 6: The <ref>door</ref><box>[[251, 124, 438, 226]]</box> appears in view alongside parts of the cow.\n\nLayer 7: A supporting structure, a <ref>brick wall</ref><box>[[24, 124, 872, 321]]</box>, extends into the scene.\n\nLayer 8: Above the scene, a <ref>window</ref><box>[[0, 124, 999, 274]]</box> adds to the view."
}
] |
coco/train2017/000000095210.jpg
|
00000048
| 408
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the foreground, we see a <ref>person</ref><box>[[931, 519, 944, 550]]</box> near a <ref>bus</ref><box>[[638, 413, 995, 636]]</box>. The <ref>pavement</ref><box>[[296, 545, 449, 634]]</box> is adjoining a grand <ref>building</ref><box>[[0, 181, 732, 613]]</box> embellished with <ref>windows</ref><box>[[0, 181, 713, 386]]</box>. There are overarching <ref>trees</ref><box>[[0, 181, 999, 817]]</box> casting a leafy canopy. \n\nLayer 2: Slightly further, another <ref>person</ref><box>[[902, 519, 924, 534]]</box> is visible. The surrounding setting features the <ref>pavement</ref><box>[[296, 545, 449, 634]]</box> and the encompassing urban landscape with a patch of open <ref>sky</ref><box>[[855, 404, 874, 432]]</box>. \n\nLayer 3: A different <ref>person</ref><box>[[585, 488, 627, 524]]</box> is seen interacting with one of the <ref>buses</ref><box>[[1, 405, 298, 705]]</box>. The <ref>road</ref><box>[[0, 597, 999, 817]]</box> forms a path under the view of more of the metropolitan structures. \n\nLayer 4: A distinct <ref>person</ref><box>[[552, 471, 590, 530]]</box> is in this mid-ground, juxtaposed against the line of buses and shared urban infrastructure. \n\nLayer 5: A <ref>person</ref><box>[[472, 481, 519, 525]]</box> stands near another parked <ref>bus</ref><box>[[1, 405, 298, 705]]</box>, with the road facilitating an organized transit of elements. \n\nLayer 6: Further back, yet another <ref>person</ref><box>[[343, 502, 396, 619]]</box> is viewable within the scene, encapsulated by the vibrant urban dwellings and trees. \n\nLayer 7: Farthest in the imagery, a notable <ref>backpack</ref><box>[[341, 530, 362, 558]]</box> is apparent on a member in the image, rounding off the cadence of the visual with an air of hustle-and-bustle."
}
] |
coco/train2017/000000266200.jpg
|
00000049
| 384
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>group of people</ref><box>[[93, 528, 118, 572]]</box>, another <ref>group of people</ref><box>[[298, 528, 359, 575]]</box>, and yet another <ref>group of people</ref><box>[[394, 477, 424, 563]]</box> are visible. The <ref>beach</ref><box>[[0, 496, 999, 799]]</box> scene features several <ref>umbrellas</ref><box>[[769, 457, 844, 544], [775, 418, 853, 457], [841, 457, 935, 516], [538, 503, 563, 517], [0, 446, 74, 493], [57, 496, 140, 519]]</box>. A <ref>chair</ref><box>[[799, 497, 817, 544]]</box> is also present, adding to the beach setting.\n\nLayer 2: A <ref>group of people</ref><box>[[810, 519, 875, 577]]</box> is seen, along with another <ref>group of people</ref><box>[[553, 563, 842, 639]]</box>. More individuals include a <ref>group of people</ref><box>[[616, 372, 727, 605]]</box>, as well as a <ref>group of people</ref><box>[[655, 535, 772, 595]]</box> and one more <ref>group of people</ref><box>[[67, 540, 106, 574]]</box>. A <ref>group of people</ref><box>[[488, 369, 547, 610]]</box> can be found in the scene, interacting with the sand.\n\nLayer 3: There are some <ref>chairs</ref><box>[[683, 634, 999, 790]]</box> on the beach.\n\nLayer 4: An additional <ref>group of people</ref><box>[[419, 503, 441, 514]]</box> is visible along with the <ref>umbrellas</ref><box>[[769, 457, 844, 544], [775, 418, 853, 457], [841, 457, 935, 516], [538, 503, 563, 517], [0, 446, 74, 493], [57, 496, 140, 519]]</box>, framing the lively beach.\n\nLayer 5: A <ref>chair</ref><box>[[391, 524, 399, 538]]</box> is settled along with the scene, while a <ref>group of people</ref><box>[[394, 477, 424, 563]]</box> can be observed.\n\nLayer 6: Another <ref>group of people</ref><box>[[374, 541, 462, 578]]</box> enjoying the beach, situated among the <ref>group</ref><box>[[488, 369, 547, 610]]</box> already interacting with the sand.\n\nLayer 7: See a <ref>group of people</ref><box>[[274, 499, 288, 543]]</box> and another <ref>group of people</ref><box>[[560, 489, 586, 546]]</box> engaged in beach activities.\n\nLayer 8: There's also a <ref>backpack</ref><box>[[265, 556, 298, 578]]</box> that rests on the sand.\n\nLayer 9: Finally, a <ref>group of people</ref><box>[[0, 588, 106, 623]]</box> occupies a similar spot alongside another <ref>group of people</ref><box>[[553, 563, 842, 639]]</box> while a different <ref>group of people</ref><box>[[488, 369, 547, 610]]</box> brings movement.\n\nLayer 10: The skyline is filled with the expansive <ref>sky</ref><box>[[0, 199, 999, 460]]</box>, creating a picturesque backdrop above the scene."
}
] |
coco/train2017/000000322507.jpg
|
00000050
| 333
| 500
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A group of <ref>people</ref><box>[[147, 317, 351, 723], [381, 403, 508, 679]]</box> is <pred>standing on</pred><box>[[801, 677, 903, 693], [69, 701, 171, 737]]</box> their <ref>skis</ref><box>[[801, 677, 903, 693], [69, 701, 171, 737]]</box>, and playing on the <ref>skis</ref><box>[[801, 677, 903, 693], [69, 701, 171, 737]]</box>. These <ref>people</ref><box>[[147, 317, 351, 723], [381, 403, 508, 679]]</box> are preparing for a race on the <ref>snow</ref><box>[[0, 525, 999, 831]]</box>-covered area. A <ref>fence</ref><box>[[147, 291, 511, 593]]</box> and a large <ref>banner</ref><box>[[181, 419, 771, 591]]</box> with event branding is also noticeable against the backdrop of a <ref>tree</ref><box>[[98, 165, 811, 399]]</box> just below the snow. Surrounding the area is the view of a <ref>building</ref><box>[[0, 165, 725, 423]]</box> along a side <ref>wall</ref><box>[[0, 275, 523, 471]]</box>.\n\nLayer 2: Another <ref>person</ref><box>[[771, 369, 919, 683]]</box> is also a part of the group preparing on their <ref>skis</ref><box>[[69, 701, 171, 737]]</box>, amidst interaction with the <ref>snow</ref><box>[[0, 525, 999, 831]]</box>. Continuing above processes are visual reflections of a racing area disclosed by another section of the same <ref>banner</ref><box>[[181, 419, 771, 591]]</box>, and an architectural blur of a partial <ref>building</ref><box>[[0, 165, 725, 423]]</box> extends into vision. A similar presence of <ref>tree</ref><box>[[98, 165, 811, 399]]</box> interlaces the scene accentuated by a <ref>fence</ref><box>[[147, 291, 511, 593]]</box> and the <ref>wall</ref><box>[[0, 275, 523, 471]]</box>.\n\nLayer 3: There is another <ref>person</ref><box>[[593, 415, 693, 663]]</box> seen from this perspective, they appear amidst the snow preparations, setting into the pace of the snowy <ref>area</ref><box>[[0, 525, 999, 831]]</box>. Continuing same manifestations of the circulations of the <ref>banner</ref><box>[[181, 419, 771, 591]]</box>, and technical materiality indulged through an affixed <ref>building</ref><box>[[0, 165, 725, 423]]</box>. Encroached forest remains with the featured <ref>tree</ref><box>[[98, 165, 811, 399]]</box>.\n\nLayer 4: A single, small <ref>window</ref><box>[[51, 165, 91, 377]]</box> is embedded within the structure, offering a fragmentary glimpse into the core of the entire scenic setting.\n\nLayer 5: Last part of the people’s arrangement includes another <ref>person</ref><box>[[27, 375, 185, 703]]</box> situated in depth, positioned interactively in snowy conditions, <pred>standing on</pred><box>[[193, 717, 274, 731]]</box> the skis, nested in the panorama. The <ref>fence</ref><box>[[147, 291, 511, 593]]</box>, and compartment of the <ref>wall</ref><box>[[0, 275, 523, 471]]</box> as a background presence accentuates the demeanor within this backdrop.\n\nLayer 6: A peripheral structure identified as a <ref>tent</ref><box>[[133, 165, 547, 311]]</box> becomes noticeable in the farther regions, held within the background, along with the communication of the <ref>building</ref><box>[[0, 165, 725, 423]]</box>, accompanied by the <ref>tree</ref><box>[[98, 165, 811, 399]]</box> which completes this distant composite."
}
] |
coco/train2017/000000203450.jpg
|
00000051
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>group of three people</ref><box>[[610, 363, 753, 695]]</box> is actively <pred>standing on</pred><box>[[610, 363, 753, 695]]</box><box>[[591, 658, 786, 725]]</box> <ref>skis</ref><box>[[591, 658, 786, 725]]</box>. They are on a <ref>snow</ref><box>[[0, 479, 999, 831]]</box> field. One individual is <pred>carrying a</pred><box>[[610, 363, 753, 695]]</box><box>[[603, 469, 630, 510]]</box> <ref>backpack</ref><box>[[603, 469, 630, 510]]</box>.\n\nLayer 2: Another <ref>member of the group</ref><box>[[469, 383, 551, 616]]</box> is similarly <pred>standing on</pred><box>[[469, 383, 551, 616]]</box><box>[[482, 608, 582, 641]]</box> <ref>skis</ref><box>[[482, 608, 582, 641]]</box>. This person also has a <pred>carrying a</pred><box>[[469, 383, 551, 616]]</box><box>[[603, 469, 630, 510]]</box> <ref>backpack</ref><box>[[603, 469, 630, 510]]</box>.\n\nLayer 3: A <ref>third person</ref><box>[[390, 419, 468, 572]]</box> is seen <pred>standing on</pred><box>[[390, 419, 468, 572]]</box><box>[[410, 572, 507, 581]]</box> <ref>skis</ref><box>[[410, 572, 507, 581]]</box>. This individual's hands appear free.\n\nLayer 4: In the backdrop, a view of <ref>trees</ref><box>[[0, 165, 999, 525]]</box> adds a natural border to the scene. Above, the <ref>sky</ref><box>[[0, 165, 999, 457]]</box> looms over the landscape, suggesting a cloudy day, enhancing the wintry setting."
}
] |
coco/train2017/000000346997.jpg
|
00000052
| 478
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \n\nA <ref>person</ref><box>[[578, 298, 972, 861]]</box> is sitting on a <ref>chair</ref><box>[[930, 614, 999, 770]]</box> and <pred>holding</pred><box>[[578, 298, 972, 861]]</box><box>[[694, 516, 728, 531]]</box> a <ref>cell phone</ref><box>[[694, 516, 728, 531]]</box>, appearing to be <pred>looking at</pred><box>[[578, 298, 972, 861]]</box><box>[[694, 516, 728, 531]]</box> or <pred>playing</pred> with it. There's also a <ref>pillow</ref><box>[[546, 579, 644, 791]]</box> on the couch nearby. \n\nLayer 2: \n\n<ref>Books</ref><box>[[79, 400, 101, 447]]</box> rest on a <ref>shelf</ref><box>[[182, 396, 210, 422]]</box>, and the <ref>wall</ref><box>[[221, 166, 999, 529]]</box> in the background holds a <ref>door-stuff</ref><box>[[635, 338, 783, 520]]</box>. The <ref>ceiling</ref><box>[[0, 126, 999, 288]]</box> is visible at the top.\n\nLayer 3 to Layer 18: \n\nVarious other <ref>books</ref> are scattered around in the room, filling the <ref>cabinet</ref><box>[[348, 357, 714, 656]]</box>. A <ref>group of people</ref><box>[[0, 443, 196, 664]]</box> and another <ref>group of people</ref><box>[[125, 440, 338, 676]]</box> are sitting on a <ref>couch</ref><box>[[0, 656, 373, 797]]</box>. One person is <pred>holding</pred><box>[[125, 440, 338, 676]]</box><box>[[212, 597, 245, 639]]</box> a <ref>cup</ref><box>[[212, 597, 245, 639]]</box>. The <ref>floor</ref><box>[[104, 775, 612, 872]]</box> is made of wood and reflects a warm ambiance with a <ref>light</ref><box>[[449, 254, 515, 288]]</box> fixture overhead. Another <ref>group of people</ref><box>[[298, 391, 573, 872]]</box> completes the gathering in this cozy room setting."
}
] |
coco/train2017/000000399594.jpg
|
00000053
| 490
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>bed</ref><box>[[9, 143, 999, 870]]</box> spans across the scene, serving as the main surface in the image.\n\nLayer 2: A <ref>door-stuff</ref><box>[[899, 117, 961, 176]]</box> is visible, possibly part of a door or associated fixture near the bed.\n\nLayer 3: Placed near the wall, a <ref>backpack</ref><box>[[530, 117, 721, 215]]</box> rests on the bed.\n\nLayer 4: Two <ref>dogs</ref><box>[[513, 180, 883, 813]]</box> and <box>[[0, 157, 710, 796]]</box> lie comfortably, seemingly interacting with each other on the bed.\n\nLayer 5: A <ref>blanket</ref><box>[[0, 538, 999, 881]]</box> covers a portion of the bed, adding an element of comfort to the scene.\n\nLayer 6: The upper side of the image features a <ref>wall</ref><box>[[0, 117, 999, 182]]</box>, providing context to the room's environment."
}
] |
coco/train2017/000000240449.jpg
|
00000054
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>person</ref><box>[[443, 360, 647, 689]]</box> is actively skiing on the <ref>snow</ref><box>[[0, 483, 999, 874]]</box>, wearing <ref>skis</ref><box>[[408, 649, 675, 703]]</box>. Beside the person, there is a <ref>fence</ref><box>[[0, 427, 999, 571]]</box> also placed on the snow.\n\nLayer 2: In the mid-ground, there are <ref>trees</ref><box>[[18, 124, 999, 511]]</box> standing tall, and the overcast <ref>sky</ref><box>[[0, 124, 999, 302]]</box> is shown above the trees.\n\nLayer 3: In the background, a <ref>house</ref><box>[[0, 210, 883, 494]]</box> can be seen along with a stretch of the <ref>sky</ref><box>[[0, 124, 999, 302]]</box>."
}
] |
coco/train2017/000000321860.jpg
|
00000055
| 413
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the room, there is a <ref>person</ref><box>[[593, 505, 686, 637]]</box> <pred>lying on</pred><box>[[593, 505, 686, 637]]</box><box>[[571, 502, 995, 812]]</box> a <ref>bed</ref><box>[[571, 502, 995, 812]]</box>. A <ref>handbag</ref><box>[[502, 651, 612, 719]]</box> is placed <pred>on</pred><box>[[502, 651, 612, 719]]</box><box>[[246, 633, 775, 821]]</box> the <ref>rug</ref><box>[[246, 633, 775, 821]]</box>. \n\nLayer 2: Nearby, two <ref>suitcases</ref><box>[[373, 566, 493, 639]]</box> are stacked on each other.\n\nLayer 3: Layers further back in the room show a second set of <ref>suitcases</ref><box>[[373, 514, 487, 581]]</box>.\n\nLayer 4: Above, there's a <ref>shelf</ref><box>[[286, 326, 476, 408]]</box> that extends across the wall.\n\nLayer 5: Covering part of the floor, the <ref>rug</ref><box>[[246, 633, 775, 821]]</box> also holds a <ref>table</ref><box>[[0, 658, 231, 821]]</box> and various <ref>cardboard</ref><box>[[120, 622, 527, 821]]</box> items scattered about.\n\nLayer 6: Overhead is the <ref>ceiling</ref><box>[[173, 176, 802, 229]]</box>, completing the room's structural features along with the <ref>door-stuff</ref><box>[[202, 269, 548, 651]]</box> on the wall.\n\nLayer 7: To the left, a <ref>laptop</ref><box>[[1, 511, 224, 821]]</box> rests on the table.\n\nLayer 8: Finally, the left side of the room has a <ref>window-blind</ref><box>[[0, 176, 79, 602]]</box> adorning one of the walls."
}
] |
coco/train2017/000000335216.jpg
|
00000056
| 375
| 500
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nIn the foreground, there’s a pair of <ref>skis</ref><box>[[61, 838, 159, 867]]</box>. Closeby, a <ref>person</ref><box>[[98, 628, 171, 844]]</box> is wearing them.\n\nLayer 2: \nA group of <ref>people</ref><box>[[363, 645, 407, 717], [98, 628, 171, 844], [945, 609, 997, 786], [285, 617, 369, 807], [73, 627, 119, 753]]</box> are visible, engaged in varied activities on the <ref>snowy mountain</ref><box>[[0, 646, 999, 873]]</box>.\n\nLayer 3: \nThere’s another <ref>backpack</ref><box>[[348, 790, 399, 850]]</box> lying unclaimed on the <ref>snow</ref><box>[[0, 646, 999, 873]]</box>.\n\nLayer 4: \nA <ref>person</ref><box>[[311, 606, 333, 624]]</box> is present, among several others closely standing, such as person with another <pred>carrying</pred><box>[[945, 609, 997, 786]]</box><box>[[957, 623, 999, 670]]</box> <ref>backpack</ref><box>[[957, 623, 999, 670]]</box>.\n\nLayer 5: \nTwo more <ref>people</ref><box>[[287, 618, 305, 658], [447, 618, 515, 730]]</box> add to the bustling snow scene.\n\nLayer 6: \nTwo other <ref>people</ref><box>[[271, 617, 283, 664], [447, 618, 515, 730]]</box> are seen interacting with their surroundings.\n\nLayer 7: \nFurther away, an additional <ref>person</ref><box>[[249, 609, 271, 669]]</box> comes into view.\n\nLayer 8: \nA couple more <ref>people</ref><box>[[209, 675, 219, 688], [661, 617, 693, 669]]</box> scatter the scene.\n\nLayer 9: \nMore individuals can be seen, such as the <ref>person</ref><box>[[19, 612, 65, 714]]</box>. In the backdrop, the <ref>roof of the ski lift</ref><box>[[215, 313, 673, 564]]</box> with its \"Mountain Top\" label is faintly visible, nestled among <ref>trees</ref><box>[[0, 293, 999, 723]]</box>, with the clear <ref>sky</ref><box>[[0, 123, 999, 654]]</box> overhead acting as a fitting scenic backdrop, and a prominent <ref>mountain</ref><box>[[559, 597, 701, 669]]</box> rising majestically in the <ref>background</ref><box>[[0, 521, 645, 711]]</box>."
}
] |
coco/train2017/000000383223.jpg
|
00000057
| 333
| 500
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the foreground, there is a <ref>person</ref><box>[[827, 207, 915, 471]]</box> standing on the <ref>playingfield</ref><box>[[0, 469, 999, 831]]</box>. A <ref>person</ref><box>[[143, 467, 527, 739]]</box> is actively <pred>holding</pred><box>[[143, 467, 527, 739]]</box><box>[[140, 449, 167, 525]]</box> and <pred>swinging</pred><box>[[143, 467, 527, 739]]</box><box>[[140, 449, 167, 525]]</box> a <ref>baseball bat</ref><box>[[140, 449, 167, 525]]</box>. In addition, they are <pred>wearing</pred><box>[[143, 467, 527, 739]]</box><box>[[131, 521, 167, 547]]</box> a <ref>baseball glove</ref><box>[[131, 521, 167, 547]]</box>. Nearby, another <ref>person</ref><box>[[709, 189, 833, 417]]</box> is engaged in the activity. A <ref>bench</ref><box>[[129, 325, 426, 469]]</box> is present, as well as a <ref>fence</ref><box>[[0, 279, 999, 500]]</box>, and a <ref>banner</ref><box>[[171, 165, 999, 331]]</box> can be seen against the backdrop of a <ref>wall</ref><box>[[0, 165, 999, 477]]</box> enclosing the area.\n\nLayer 2: In the middle ground, <ref>people</ref><box>[[7, 177, 155, 473]]</box> and <box>[[205, 181, 307, 471]]</box> can be observed. The <ref>fence</ref><box>[[0, 279, 999, 500]]</box>, the <ref>banner</ref><box>[[171, 165, 999, 331]]</box>, and the <ref>wall</ref><box>[[0, 165, 999, 477]]</box> remain visible here as well.\n\nLayer 3: Further back, a <ref>baseball glove</ref><box>[[583, 613, 617, 661]]</box> is visible on a <ref>person</ref><box>[[504, 457, 741, 707]]</box>. Another <ref>person</ref><box>[[143, 467, 527, 739]]</box> is present, still related to the interaction with the <ref>baseball glove</ref><box>[[131, 521, 167, 547]]</box>. A <ref>group of people</ref><box>[[750, 405, 927, 715]]</box> are positioned in this layer.\n\nLayer 4: The layer contains a <ref>sports ball</ref><box>[[571, 595, 588, 611]]</box> in context to the scene involving a <ref>person</ref><box>[[143, 467, 527, 739]]</box> and another <ref>person</ref><box>[[504, 457, 741, 707]]</box>, both of whom are associated with their respective <ref>baseball gloves</ref><box>[[131, 521, 167, 547]]</box>. Additionally, additional <ref>people</ref><box>[[750, 405, 927, 715]]</box> enrich the scene.\n\nLayer 5: The background features <ref>people</ref><box>[[557, 185, 691, 469]]</box> on the <ref>playingfield</ref><box>[[0, 469, 999, 831]]</box>, accompanied by a <ref>person</ref><box>[[143, 467, 527, 739]]</box> and another <ref>person</ref><box>[[709, 189, 833, 417]]</box>. Here, the <ref>bench</ref><box>[[129, 325, 426, 469]]</box>, the <ref>fence</ref><box>[[0, 279, 999, 500]]</box>, and the <ref>banner</ref><box>[[171, 165, 999, 331]]</box> continue to be present within the enclosure of the <ref>wall</ref><box>[[0, 165, 999, 477]]</box>.\n\nLayer 6: Farther back, <ref>people</ref><box>[[421, 205, 523, 471]]</box> can be found, along with an visible <ref>bench</ref><box>[[129, 325, 426, 469]]</box>, a <ref>fence</ref><box>[[0, 279, 999, 500]]</box>, a <ref>banner</ref><box>[[171, 165, 999, 331]]</box>, and enclosing <ref>wall</ref><box>[[0, 165, 999, 477]]</box>.\n\nLayer 7: In the very distant layer, the <ref>baseball bat</ref><box>[[140, 449, 167, 525]]</box> remains associated with the <ref>person</ref><box>[[143, 467, 527, 739]]</box> on the <ref>playingfield</ref><box>[[0, 469, 999, 831]]</box>."
}
] |
coco/train2017/000000381433.jpg
|
00000058
| 514
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>person</ref><box>[[952, 628, 999, 895]]</box> is most prominent in the scene, appearing closest to the viewer.\n\nLayer 2: There are multiple <ref>cars</ref>:\n- One car <box>[[930, 573, 938, 593]]</box>.\n- Another car <box>[[905, 573, 931, 616]]</box>.\n- Yet another car <box>[[878, 572, 910, 628]]</box>. \nThe streets are lined with <ref>buildings</ref><box>[[0, 98, 892, 626]]</box> and <ref>trees</ref><box>[[0, 98, 999, 558]]</box>, with the <ref>sky</ref><box>[[682, 98, 999, 510]]</box> visible above.\n\nLayer 3: Additional <ref>cars</ref> include:\n- A car <box>[[927, 577, 935, 596]]</box> overlapping with layer 2.\n \nLayer 4: The bustling scene includes a <ref>bus</ref><box>[[290, 260, 835, 781]]</box> traveling down the <ref>road</ref><box>[[0, 583, 984, 900]]</box>.\n\nLayer 5: There is more presence of people, such as <ref>people</ref><box>[[499, 340, 568, 403]]</box>.\n\nLayer 6: Further <ref>people</ref> appear:\n- One <ref>person</ref><box>[[249, 561, 266, 623]]</box>.\n- Additional individuals are denoted by overlapping boxes indicating the crowded nature.\n\nLayer 7 to Layer 10: The pedestrian and traffic activity continues with:\n- More <ref>people</ref> <box>[[159, 566, 181, 594]]</box>, <box>[[142, 556, 162, 591]]</box> visible.\n- Further <ref>cars</ref><box>[[1, 583, 238, 744]]</box> representing distant vehicles. \n\nThe scene depicts vibrant city life bustling with vehicular and pedestrian activity."
}
] |
coco/train2017/000000433998.jpg
|
00000059
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>tv</ref><box>[[758, 394, 999, 515]]</box> is <pred>attached to</pred><box>[[758, 394, 999, 515]]</box><box>[[0, 352, 771, 529]]</box> a <ref>wall</ref><box>[[0, 352, 771, 529]]</box>. A <ref>baseball player</ref><box>[[493, 460, 605, 701]]</box> is also present in this layer.\n\nLayer 2: The <ref>baseball player</ref><box>[[493, 460, 605, 701]]</box> is <pred>holding</pred><box>[[493, 460, 605, 701]]</box><box>[[515, 558, 524, 569]]</box> a <ref>sports ball</ref><box>[[515, 558, 524, 569]]</box> and <pred>wearing</pred><box>[[493, 460, 605, 701]]</box><box>[[490, 525, 522, 561]]</box> a <ref>baseball glove</ref><box>[[490, 525, 522, 561]]</box>. The player is <pred>looking at</pred><box>[[493, 460, 605, 701]]</box><box>[[423, 692, 612, 831]]</box> another <ref>person</ref><box>[[423, 692, 612, 831]]</box> in the distance.\n\nLayer 3: The <ref>baseball player</ref><box>[[493, 460, 605, 701]]</box> is <pred>standing on</pred><box>[[493, 460, 605, 701]]</box><box>[[0, 609, 999, 775]]</box> a <ref>sand</ref><box>[[0, 609, 999, 775]]</box> mound. The player is also <pred>interacting with</pred><box>[[493, 460, 605, 701]]</box> the surroundings described.\n\nLayer 4: The <ref>playingfield</ref><box>[[0, 508, 999, 831]]</box> encompassing the whole scene, including the <ref>baseball player</ref><box>[[493, 460, 605, 701]]</box> and the <ref>sand</ref><box>[[0, 609, 999, 775]]</box>.\n\nLayer 5: A <ref>fence</ref><box>[[0, 291, 959, 398]]</box> separates the field from the spectator area, indicating the boundary of the playing field."
}
] |
coco/train2017/000000395920.jpg
|
00000060
| 360
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \n - In this near layer, several <ref>traffic lights</ref><box>[[970, 478, 978, 502]]</box><box>[[900, 291, 961, 435]]</box> can be seen. Beyond, there appears to be some background elements like <ref>trees</ref><box>[[415, 218, 999, 525]]</box> and the <ref>sky</ref><box>[[0, 218, 999, 438]]</box>.\n \nLayer 2: \n - Further on, additional <ref>traffic lights</ref><box>[[900, 291, 961, 435]]</box> are visible. Additionally, the background remains the same with <ref>trees</ref><box>[[415, 218, 999, 525]]</box> and the <ref>sky</ref><box>[[0, 218, 999, 438]]</box> present.\n\nLayer 3: \n - This layer shows more <ref>traffic lights</ref><box>[[900, 291, 961, 435]]</box>. The backdrop of <ref>trees</ref><box>[[415, 218, 999, 525]]</box> and <ref>sky</ref><box>[[0, 218, 999, 438]]</box> continues to be visible in this layer as well.\n\nLayer 4: \n - At this depth, a <ref>person</ref><box>[[546, 440, 597, 529]]</box> is noticeable, likely <pred>sitting on</pred><box>[[546, 440, 597, 529]]</box> the <ref>green and white bus</ref><box>[[293, 313, 830, 702]]</box>. The urban setting includes <ref>buildings</ref><box>[[0, 218, 999, 599]]</box> lining the street.\n\nLayer 5: \n - Another <ref>person</ref><box>[[387, 408, 488, 502]]</box> is identified in this layer, likely <pred>driving</pred><box>[[387, 408, 488, 502]]</box><box>[[293, 313, 830, 702]]</box> the <ref>green and white bus</ref><box>[[293, 313, 830, 702]]</box>. The presence of <ref>buildings</ref><box>[[0, 218, 999, 599]]</box> continues in the backdrop.\n\nLayer 6: \n - This layer includes a <ref>handbag</ref><box>[[291, 514, 298, 543]]</box> on the pavement and the <ref>road</ref><box>[[0, 495, 999, 780]]</box> where the <ref>green and white bus</ref><box>[[293, 313, 830, 702]]</box> is <pred>driving on</pred><box>[[293, 313, 830, 702]]</box><box>[[0, 495, 999, 780]]</box>.\n\nLayer 7: \n - More urban environmental details are present, including <ref>traffic lights</ref><box>[[227, 412, 255, 474]]</box> near the <ref>green and white bus</ref><box>[[293, 313, 830, 702]]</box>, and a <ref>person</ref><box>[[273, 503, 298, 572]]</box> <pred>walking on</pred><box>[[273, 503, 298, 572]]</box> the <ref>pavement</ref><box>[[0, 514, 999, 780]]</box>."
}
] |
coco/train2017/000000372413.jpg
|
00000061
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the foreground, a <ref>motorcyclist</ref><box>[[766, 341, 913, 552]]</box> is navigating along a <ref>road</ref><box>[[0, 272, 999, 831]]</box>, which is bordered by <ref>grass</ref><box>[[0, 216, 999, 699]]</box>.\n\nLayer 2: Slightly further back, a <ref>person</ref><box>[[532, 502, 784, 594]]</box> is positioned on a <ref>motorcycle</ref><box>[[775, 410, 919, 613]]</box>, with a visible <ref>floor</ref><box>[[0, 564, 585, 723]]</box> area alongside the road.\n\nLayer 3: There is a configuration of <ref>cars</ref><box>[[340, 207, 376, 219]]</box> parked near the edge of the road.\n\nLayer 4: A short distance away, a small cluster of <ref>cars</ref><box>[[323, 212, 334, 218]]</box> can be observed.\n\nLayer 5: More vehicles, a few <ref>cars</ref><box>[[305, 208, 321, 219]]</box>, remain visible extending further.\n\nLayer 6: In this layer, several <ref>people</ref><box>[[291, 340, 329, 502], [31, 418, 135, 513], [149, 444, 202, 519]]</box> stand and sit along the <ref>grass</ref><box>[[0, 216, 999, 699]]</box>.\n\nLayer 7: The scene is set against a backdrop of more <ref>cars</ref><box>[[280, 205, 298, 219]]</box> alongside a prominent <ref>tree</ref><box>[[0, 165, 999, 361]]</box> near the road.\n\nLayer 8: There is an additional set of <ref>cars</ref><box>[[234, 205, 273, 219]]</box> next to the natural surroundings of a <ref>tree</ref><box>[[0, 165, 999, 361]]</box>.\n\nLayer 9: The furthest visible area includes <ref>cars</ref><box>[[12, 191, 54, 218]]</box>, with the expansive view of the <ref>tree</ref><box>[[0, 165, 999, 361]]</box> enhancing the scenic outdoor environment."
}
] |
coco/train2017/000000297266.jpg
|
00000062
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nA <ref>motorcycle</ref><box>[[591, 411, 959, 831]]</box> is closest to the camera. \n\nLayer 2: \nA <ref>person</ref><box>[[571, 177, 724, 341]]</box> is positioned nearby. \nAnother <ref>person</ref><box>[[700, 264, 841, 412]]</box> is standing close as well. \nYet another <ref>person</ref><box>[[900, 335, 980, 683]]</box> is part of the scene. \nAdditionally, there is a <ref>person</ref><box>[[271, 221, 373, 438]]</box>. \n\nLayer 3: \nA <ref>person</ref><box>[[472, 274, 633, 475]]</box> is in this layer. \nAnother <ref>person</ref><box>[[346, 255, 490, 475]]</box> is captured here. \nOne more <ref>person</ref><box>[[821, 328, 977, 817]]</box> appears in this layer. \nAccompanied by a further <ref>person</ref><box>[[900, 335, 980, 683]]</box>. \nAdditionally, a <ref>person</ref><box>[[572, 249, 711, 516]]</box> is visible. \nThere's another <ref>person</ref><box>[[137, 218, 332, 531]]</box>. \n\nLayer 4: \nHere, a <ref>person</ref><box>[[440, 235, 522, 360]]</box> is visible. \nAlso, a <ref>person</ref><box>[[700, 264, 841, 412]]</box> appears again. \nAnother <ref>person</ref><box>[[821, 328, 977, 817]]</box> is present. \nA further <ref>person</ref><box>[[900, 335, 980, 683]]</box> can be seen. \nA <ref>person</ref><box>[[271, 221, 373, 438]]</box> is also here. \nAdditionally, a <ref>person</ref><box>[[365, 191, 465, 328]]</box> appears in this layer. \nAnd another <ref>person</ref><box>[[137, 218, 332, 531]]</box> is included. \nFinally, a <ref>door</ref><box>[[234, 207, 700, 327]]</box> is visible. \n\nLayer 5: \nThere is a <ref>shelf</ref><box>[[4, 188, 74, 227]]</box> seen here. \n<ref>Water</ref><box>[[0, 165, 999, 472]]</box> dominates this layer. \nA <ref>window</ref><box>[[209, 165, 691, 222]]</box> appears overhead. \n\nLayer 6: \nA different <ref>motorcycle</ref><box>[[3, 430, 499, 815]]</box> is in the scene. \n\nLayer 7: \nAnother <ref>person</ref><box>[[0, 314, 168, 697]]</box> is positioned here. \nAn additional <ref>person</ref><box>[[821, 328, 977, 817]]</box> can be seen. \nAnother <ref>person</ref><box>[[572, 249, 711, 516]]</box> is present. "
}
] |
coco/train2017/000000507937.jpg
|
00000063
| 425
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the foreground, there is a <ref>person</ref><box>[[505, 577, 560, 621]]</box> in proximity to a <ref>bus</ref><box>[[286, 514, 603, 755]]</box>, suggesting that this person is the driver. The <ref>bus</ref><box>[[286, 514, 603, 755]]</box> is painted green and white and is positioned <pred>on</pred><box>[[286, 514, 603, 755]]</box><box>[[131, 536, 999, 830]]</box> the <ref>road</ref><box>[[131, 536, 999, 830]]</box>. Adjacent to the road is the <ref>fence</ref><box>[[0, 637, 416, 830]]</box> along the <ref>pavement</ref><box>[[0, 636, 316, 830]]</box>.\n\nLayer 2: There is a <ref>truck</ref><box>[[235, 590, 274, 631]]</box> positioned in front of the bus, slightly further back is the structure of a <ref>building</ref><box>[[149, 390, 999, 703]]</box>. Additionally, a sprawling <ref>tree</ref><box>[[0, 167, 999, 669]]</box> adds greenery beside the scene.\n\nLayer 3: Another vehicle, a <ref>car</ref><box>[[212, 592, 229, 609]]</box>, is noticeable. The expansive <ref>sky</ref><box>[[0, 167, 930, 472]]</box> can be seen overhead, offering a backdrop to the urban landscape.\n\nLayer 4: Positioned slightly further, another <ref>car</ref><box>[[195, 574, 202, 586]]</box> is discernible.\n\nLayer 5: Another <ref>truck</ref><box>[[176, 584, 201, 623]]</box> is visible on the road, further extending the vehicular presence.\n\nLayer 6: In the rear segment, a small <ref>car</ref><box>[[143, 614, 196, 644]]</box> is among the distant objects.\n\nLayer 7: Lastly, a <ref>person</ref><box>[[101, 606, 150, 728]]</box> is <pred>walking on</pred><box>[[101, 606, 150, 728]]</box><box>[[0, 636, 316, 830]]</box> the <ref>pavement</ref><box>[[0, 636, 316, 830]]</box>, providing a human element to the urban activity."
}
] |
coco/train2017/000000105647.jpg
|
00000064
| 429
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: The scene is bustling with <ref>pedestrians</ref><box>[[861, 228, 970, 321], [438, 413, 452, 475], [42, 408, 118, 574], [162, 397, 223, 566], [124, 391, 174, 525], [212, 327, 449, 373]]</box> moving through the area. They are surrounded by <ref>handbags</ref><box>[[107, 497, 132, 536], [224, 419, 280, 483], [285, 405, 330, 469]]</box> as part of their attire. The busy <ref>road</ref><box>[[529, 417, 999, 833]]</box> features a large <ref>bus</ref><box>[[452, 165, 994, 516]]</box> in motion, contributing to the lively environment. The scene is organized under a <ref>banner</ref><box>[[0, 163, 999, 324]]</box> and framed by a <ref>building</ref><box>[[0, 163, 999, 464]]</box>.\n\nLayer 2: Additional <ref>pedestrians</ref><box>[[784, 232, 841, 321]]</box> add to the crowd, accompanied by a <ref>traffic light</ref><box>[[38, 212, 127, 341]]</box> regulating street activity. A colorful display of <ref>fruit</ref><box>[[251, 435, 360, 530]]</box> is on offering amidst the busy foot traffic. The area is bordered by a <ref>pavement</ref><box>[[0, 338, 341, 833]]</box> where people walk.\n\nLayer 3: The bustling scene becomes more apparent with a <ref>car</ref><box>[[315, 363, 714, 650]]</box> navigating the road, filled with additional <ref>pedestrians</ref><box>[[693, 388, 777, 642]]</box> walking along. The passage is accompanied by a <ref>pavement</ref><box>[[0, 338, 341, 833]]</box> where people can stroll safely.\n\nLayer 4: Within this energetic area, more <ref>handbags</ref><box>[[507, 478, 530, 500]]</box> are visible, complementing the style of some of the <ref>pedestrians</ref><box>[[471, 430, 530, 539]]</box> traversing this lively street.\n\nLayer 5: The scene is lined with <ref>bicycles</ref><box>[[340, 580, 377, 642], [334, 531, 419, 706], [412, 572, 476, 716], [408, 586, 465, 732], [176, 544, 319, 805], [444, 708, 590, 822], [393, 541, 487, 710], [465, 589, 774, 833]]</box> parked against a sturdy <ref>fence</ref><box>[[262, 509, 479, 833]]</box>, hinting at the environmentally friendly transport options available.\n\nLayer 6: A single <ref>pedestrian</ref><box>[[338, 355, 363, 388]]</box> appears amidst the commotion, further confirming the vibrant street life.\n\nLayer 7: In the backdrop, a portion of the <ref>sky</ref><box>[[215, 163, 404, 247]]</box> is visible, adding a serene canvas above the bustling activities below."
}
] |
coco/train2017/000000358719.jpg
|
00000065
| 426
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: <ref>Window blinds</ref><box>[[924, 422, 953, 455]]</box> cover part of a <ref>window</ref><box>[[839, 304, 999, 517]]</box> that shapes the outer section of the <ref>walls</ref><box>[[0, 167, 999, 591]]</box> and reaches up to the <ref>ceiling</ref><box>[[308, 167, 999, 227]]</box>.\n\nLayer 2: A <ref>chair</ref><box>[[749, 505, 941, 808]]</box> resides comfortably below a <ref>light</ref><box>[[668, 194, 819, 511]]</box> that brightens the room, which these <ref>walls</ref><box>[[0, 167, 999, 591]]</box> enclose.\n\nLayer 3: Below the same <ref>light</ref><box>[[668, 194, 819, 511]]</box>, a <ref>table</ref><box>[[703, 563, 782, 633]]</box> tops with <ref>food</ref><box>[[711, 628, 794, 692]]</box> is visible in the dimly lit area.\n\nLayer 4: A settled <ref>person</ref><box>[[45, 634, 410, 722]]</box> reclines comfortably, potentially viewing a scene or gathering items within reach.\n\nLayer 5: The scene unfolds as the <ref>wooden floor</ref><box>[[607, 703, 963, 831]]</box> meets the floor's edge, giving balance to the room’s <ref>couch</ref><box>[[1, 535, 728, 825]]</box> along with its furniture arrangement and again supporting the <ref>walls</ref><box>[[0, 167, 999, 591]]</box>.\n\nLayer 6: Within the vicinity stands a <ref>woman</ref><box>[[519, 342, 760, 829]]</box> agile and engaged, potentially in a game.\n\nLayer 7: Nearby, a <ref>man</ref><box>[[129, 213, 493, 825]]</box> is also an active participant, drawing attention to the lively engagement across the scene."
}
] |
coco/train2017/000000572233.jpg
|
00000066
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the image, <ref>people</ref><box>[[725, 530, 749, 577]]</box>, as well as more <ref>people</ref><box>[[313, 467, 369, 527]], [580, 463, 658, 547], [410, 466, 432, 501], [196, 466, 259, 536], [833, 430, 994, 747], [689, 443, 774, 541], [188, 411, 368, 659]]</box>, are gathered at <ref>dining tables</ref><box>[[668, 552, 970, 820]]</box>. A <ref>chair</ref><box>[[922, 541, 944, 594]]</box> is positioned nearby. In the surroundings, <ref>potted plants</ref><box>[[109, 179, 369, 411], [132, 491, 227, 572], [7, 240, 165, 418], [749, 480, 866, 567]]</box> adorn the area.\n\nLayer 2: There is an <ref>individual</ref><box>[[677, 524, 696, 558]]</box> who appears engaged, and more <ref>people</ref><box>[[313, 467, 369, 527]], [580, 463, 658, 547], [410, 466, 432, 501]]</box> join them. A <ref>chair</ref><box>[[922, 541, 944, 594]]</box> remains present in the background, decorated with <ref>potted plants</ref><box>[[109, 179, 369, 411], [132, 491, 227, 572], [7, 240, 165, 418]]</box>.\n\nLayer 3: Seated with <ref>people</ref><box>[[647, 453, 713, 533]], [580, 463, 658, 547], [410, 466, 432, 501]]</box>, a group enjoys their time near the <ref>chair</ref><box>[[922, 541, 944, 594]]</box>. The setting continues to include charming <ref>potted plants</ref><box>[[109, 179, 369, 411], [132, 491, 227, 572]]</box>.\n\nLayer 4: A <ref>man</ref><box>[[346, 325, 714, 820]]</box> is seated on a <ref>chair</ref><box>[[551, 685, 658, 831]]</box> beside other <ref>people</ref><box>[[313, 467, 369, 527]]</box> at <ref>dining tables</ref><box>[[0, 571, 380, 823]]</box>.\n\nLayer 5: Adorning his attire, the <ref>man</ref><box>[[346, 325, 714, 820]]</box> is wearing a <ref>tie</ref><box>[[454, 485, 529, 549]]</box> while the festivities carry on.\n\nLayer 6: Allowing daylight, the room’s <ref>windows</ref><box>[[423, 165, 999, 294]]</box> filter light over <ref>people</ref><box>[[616, 467, 655, 508]]</box> as they sit near the <ref>chair</ref><box>[[922, 541, 944, 594]]</box> under the <ref>ceiling</ref><box>[[0, 165, 999, 406]]</box>.\n\nLayer 7: Tidily positioned, wine glasses sit near <ref>people</ref><box>[[154, 583, 191, 609]], [313, 467, 369, 527]]</box> together in their cheerful moment.\n\nLayer 8: From elegant <ref>wine glasses</ref><box>[[0, 577, 14, 622], [143, 607, 201, 715], [87, 591, 132, 671], [98, 578, 143, 661]]</box> at the <ref>dining tables</ref><box>[[0, 571, 380, 823]]</box>, the décor is completed with a <ref>knife</ref><box>[[125, 650, 255, 672]]</box> that complements the mise-en-scène.\n\nLayer 9: Completing the backdrop, the <ref>wooden wall</ref><box>[[0, 165, 999, 571]]</box> envelops the joyous gathering in elegant refinement."
}
] |
coco/train2017/000000278461.jpg
|
00000067
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the foreground, one of the <ref>men</ref><box>[[747, 357, 999, 874]]</box> is visible. This <ref>man</ref><box>[[747, 357, 999, 874]]</box> is <pred>wearing</pred><box>[[892, 520, 930, 670]]</box> a yellow <ref>tie</ref><box>[[892, 520, 930, 670]]</box>. Another <ref>man</ref><box>[[521, 295, 881, 874]]</box> is standing close by.\n\nLayer 2: Slightly further back, another individual in the group is wearing a yellow <ref>tie</ref><box>[[40, 477, 727, 789]]</box>.\n\nLayer 3: Moving deeper into the scene, there is a <ref>man</ref><box>[[173, 394, 340, 871]]</box>, forming part of the group.\n\nLayer 4: Positioned further, another <ref>man</ref><box>[[171, 344, 552, 864]]</box> becomes part of the assembled group.\n\nLayer 5: At the back, another <ref>man</ref><box>[[1, 355, 209, 874]]</box> stands, completing the lineup of individuals involved in the performance."
}
] |
coco/train2017/000000213008.jpg
|
00000068
| 425
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: The image features a <ref>light</ref><box>[[276, 358, 348, 421]]</box> that is mounted on the <ref>wall</ref><box>[[318, 167, 999, 577]]</box>, providing illumination to the room. A <ref>blanket</ref><box>[[71, 592, 961, 830]]</box> is spread across the <ref>bed</ref><box>[[0, 444, 999, 830]]</box>. Additionally, a <ref>chair</ref><box>[[212, 483, 332, 597]]</box> is present, positioned towards a <ref>curtain</ref><box>[[0, 167, 357, 627]]</box> that drapes toward the floor.\n\nLayer 2: Another <ref>chair</ref><box>[[20, 505, 145, 653]]</box> appears within the scene, slightly recessed compared to other objects.\n\nLayer 3: In this layer, there is a further receding <ref>chair</ref><box>[[127, 520, 142, 541]]</box>. Further into the scene, a <ref>dining table</ref><box>[[140, 520, 271, 561]]</box> complements the seating area. The <ref>curtain</ref><box>[[0, 167, 357, 627]]</box> seems to cover a more extended area, with the <ref>ceiling</ref><box>[[103, 167, 511, 230]]</box> enhancing the height perspective. Additionally, a <ref>flower</ref><box>[[401, 495, 437, 534]]</box> resides within a <ref>vase</ref><box>[[410, 517, 427, 544]]</box>, placed on a <ref>table</ref><box>[[393, 517, 430, 555]]</box>.\n\nLayer 4: There's a <ref>rug</ref><box>[[0, 606, 113, 830]]</box>, adding to the texture and color variety at the bottom end of the room.\n\nLayer 5: The <ref>bed</ref><box>[[0, 444, 999, 830]]</box> forms the backdrop, anchoring all furniture and decor within the scene's layout."
}
] |
coco/train2017/000000457754.jpg
|
00000069
| 448
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A single <ref>chair</ref><box>[[910, 472, 999, 695]]</box> is positioned prominently in the foreground.\n\nLayer 2: This layer includes several <ref>men</ref><box>[[129, 212, 338, 666], [625, 166, 861, 539]]</box>, along with a <ref>table</ref><box>[[0, 623, 999, 849]]</box> that the men appear to be using. <ref>Books</ref><box>[[167, 653, 462, 751]]</box> are noted <pred>on</pred><box>[[167, 653, 462, 751]]</box><box>[[0, 623, 999, 849]]</box> the table. A <ref>clock</ref><box>[[365, 330, 447, 399]]</box> is situated <pred>on</pred><box>[[365, 330, 447, 399]]</box><box>[[491, 461, 597, 675]]</box> a <ref>cabinet</ref><box>[[491, 461, 597, 675]]</box>.\n\nLayer 3: One <ref>man</ref><box>[[540, 392, 867, 688]]</box> is partially obscured, wearing a <ref>tie</ref><box>[[679, 532, 714, 650]]</box>.\n\nLayer 4: Another <ref>man</ref><box>[[251, 366, 530, 703]]</box> is present, along with the <ref>papers</ref><box>[[824, 639, 849, 653]]</box> on the <ref>table</ref><box>[[0, 623, 999, 849]]</box>. A <ref>clock</ref><box>[[457, 652, 466, 664]]</box> is noted <pred>worn</pred><box>[[251, 366, 530, 703]]</box><box>[[457, 652, 466, 664]]</box> by one of the men. Other <ref>ties</ref><box>[[348, 497, 399, 656]]</box> are seen.\n\nLayer 5: The <ref>door-stuff</ref><box>[[0, 149, 999, 744]]</box> along with a <ref>wall</ref><box>[[229, 149, 611, 496]]</box> are visible, forming the backdrop of the scene.\n\nLayer 6: Another set of <ref>men</ref><box>[[129, 212, 338, 666], [625, 166, 861, 539]]</box> are dressed with <ref>ties</ref><box>[[218, 316, 248, 429]]</box>, creating a layered effect in the composition."
}
] |
coco/train2017/000000080654.jpg
|
00000070
| 612
| 612
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \n- A <ref>wine glass</ref><box>[[406, 695, 452, 785]]</box> and some <ref>paper</ref><box>[[972, 587, 999, 617]]</box> are visible on the <ref>table</ref><box>[[17, 675, 999, 999]]</box>, adding to the ceremonial atmosphere.\n\nLayer 2: \n- Several individuals from the <ref>group of people</ref><box>[[777, 228, 982, 718]]</box><box>[[558, 169, 827, 728]]</box> are gathered around.\n\nLayer 3:\n- Other people in the <ref>group</ref> can be spotted, such as the one <pred>holding</pred><box>[[364, 434, 429, 515]]</box> the <ref>scissors</ref><box>[[364, 434, 429, 515]]</box> and those <pred>looking at</pred><box>[[0, 179, 120, 688]]</box> these scissors.\n\nLayer 4:\n- A <ref>wall</ref><box>[[749, 0, 999, 298]]</box> is part of the background, suggesting an indoor setting.\n\nLayer 5:\n- A person is <pred>wearing</pred><box>[[450, 372, 483, 502]]</box> a <ref>tie</ref><box>[[450, 372, 483, 502]]</box> within the <ref>group of people</ref><box>[[349, 226, 593, 708], [856, 166, 996, 450], [558, 169, 827, 728], [84, 117, 326, 692]]</box>.\n\nLayer 6:\n- Some people are part of the <ref>group</ref><box>[[760, 120, 909, 331], [220, 133, 300, 359]]</box>, contributing to the gathered assembly.\n\nLayer 7:\n- Additional members of the <ref>group</ref><box>[[298, 174, 470, 693], [760, 120, 909, 331], [220, 133, 300, 359]]</box> complete the scene.\n\nLayer 8:\n- A few people from the <ref>group</ref> can be observed as distant figures <box>[[279, 179, 388, 385]]</box>.\n\nLayer 9:\n- The <ref>table</ref><box>[[17, 675, 999, 999]]</box> plays a central role, supporting the ribbon-cutting ceremony.\n\nLayer 10:\n- Finally, a <ref>wall-brick</ref><box>[[0, 0, 752, 305]]</box> completes the indoor backdrop."
}
] |
coco/train2017/000000126301.jpg
|
00000071
| 425
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nA <ref>bicycle</ref><box>[[407, 416, 917, 725]]</box> is <pred>leaning on</pred> a <ref>bench</ref><box>[[730, 526, 999, 555]]</box>. \n\nLayer 2: \nNearby, another <ref>handbag</ref><box>[[295, 622, 418, 722]]</box> lies on the <ref>grass</ref><box>[[0, 690, 999, 830]]</box>.\n\nLayer 3: \nA <ref>woman</ref><box>[[182, 301, 351, 712]]</box> is <pred>playing</pred> with a <ref>laptop</ref><box>[[221, 436, 327, 494]]</box>. \nIn the background, a <ref>house</ref><box>[[0, 520, 999, 590]]</box> sits beside a <ref>building</ref><box>[[0, 167, 999, 572]]</box>. At a slightly distant point, a <ref>wall</ref><box>[[0, 550, 999, 719]]</box> stretches across the view.\n\nLayer 4: \nAnother <ref>handbag</ref><box>[[195, 483, 329, 522]]</box> is placed beside the <ref>woman</ref><box>[[182, 301, 351, 712]]</box> and on the <ref>grass</ref><box>[[0, 690, 999, 830]]</box>."
}
] |
coco/train2017/000000405137.jpg
|
00000072
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the foreground, a <ref>zebra</ref><box>[[733, 432, 880, 552]]</box> is present, <pred>standing on</pred> and <pred>eating</pred> the <ref>grass</ref><box>[[0, 268, 999, 591]]</box> spread across the field.\n\nLayer 2: To the side, a stretch of <ref>dirt</ref><box>[[0, 552, 999, 831]]</box> is visible, bordering the grassy area where the zebra grazes.\n\nLayer 3: In the background, <ref>trees</ref><box>[[0, 165, 999, 481]]</box> are dispersed across the landscape, adding depth to the peaceful scene."
}
] |
coco/train2017/000000136240.jpg
|
00000073
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>sports ball</ref><box>[[621, 511, 632, 530]]</box> is prominently displayed. The <ref>playingfield</ref><box>[[0, 165, 999, 831]]</box>, a baseball field, spreads across the scene. A <ref>baseball bat</ref><box>[[553, 493, 660, 557]]</box> is in proximity to the sports ball.\n\nLayer 2: A <ref>catcher</ref><box>[[398, 566, 575, 765]]</box> is <pred>wearing</pred><box>[[398, 566, 575, 765]]</box><box>[[494, 580, 535, 625]]</box> a <ref>baseball glove</ref><box>[[494, 580, 535, 625]]</box>. The <ref>batter</ref><box>[[399, 397, 553, 669]]</box> stands ready at the plate.\n\nLayer 3: Another <ref>person</ref><box>[[398, 566, 575, 765]]</box>, likely the catcher, is close to the batter in the composition.\n\nLayer 4: An <ref>umpire</ref><box>[[263, 545, 388, 801]]</box> stands further back in the field, overseeing the play."
}
] |
coco/train2017/000000038840.jpg
|
00000074
| 478
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the forefront, there is a <ref>snowboard</ref><box>[[583, 697, 727, 803]]</box> prominently visible.\n\nLayer 2: Behind it, a <ref>fence</ref><box>[[583, 437, 783, 493]]</box> can be seen, positioned <pred>over</pred><box>[[583, 437, 783, 493]]</box> the extensive <ref>snow</ref><box>[[0, 288, 999, 872]]</box> field.\n\nLayer 3: Above the scene is the expansive <ref>sky</ref><box>[[346, 126, 999, 249]]</box>, arching <pred>over</pred><box>[[346, 126, 999, 249]]</box> the entire setting.\n\nLayer 4: A <ref>person</ref><box>[[293, 340, 647, 817]]</box> is present, seemingly positioned <pred>on</pred><box>[[583, 697, 727, 803]]</box> the snowboard, and they appear <pred>attached to</pred><box>[[583, 697, 727, 803]]</box> it as well.\n\nLayer 5: Nearby, another <ref>person</ref><box>[[229, 383, 274, 496]]</box> is depicted.\n\nLayer 6: Alongside them, there are <ref>skis</ref><box>[[216, 487, 305, 502]]</box> evident in the scene, enhancing the wintry atmosphere.\n\nLayer 7: In the far background, a majestic <ref>mountain</ref><box>[[0, 126, 999, 477]]</box> adds context to the snowy environment."
}
] |
coco/train2017/000000342204.jpg
|
00000075
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: The closest item to the camera is a <ref>cup</ref><box>[[895, 627, 967, 711]]</box>, which belongs to one of the two <ref>women</ref><box>[[546, 330, 999, 862]]</box> present in the image. In the background, there are <ref>trees</ref><box>[[0, 124, 999, 608]]</box> visible, adding depth to the scene.\n\nLayer 2: The <ref>sky</ref><box>[[74, 124, 999, 463]]</box> is visible beyond the two women and provides a backdrop over them, giving the perspective of being at an outdoor location, possibly a park or an event.\n\nLayer 3: In this layer, we have a <ref>woman</ref><box>[[546, 330, 999, 862]]</box> sitting beside another <ref>woman</ref><box>[[42, 194, 700, 864]]</box>, together indicating the social setting of the image.\n\nLayer 4: The two <ref>women</ref><box>[[42, 194, 700, 864]]</box> are seated on the ground. One is enjoying a <ref>donut</ref><box>[[494, 463, 568, 542]]</box>, emphasizing a relaxed, casual gathering. The presence of <ref>trees</ref><box>[[0, 124, 999, 608]]</box> provides the context of an outdoor environment."
}
] |
coco/train2017/000000211107.jpg
|
00000076
| 428
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the <ref>living room</ref>, there is a <ref>couch</ref><box>[[493, 430, 933, 658]]</box>. Adjacent to it is a <ref>dining table</ref><box>[[188, 508, 468, 733]]</box> with a <ref>rug</ref><box>[[131, 531, 615, 826]]</box> beneath. The <ref>floor</ref><box>[[53, 505, 999, 833]]</box> is partially visible around these.\n\nLayer 2: In close proximity sits a <ref>chair</ref><box>[[463, 438, 535, 544]]</box> near the couch.\n\nLayer 3: There is a piece of <ref>cardboard</ref><box>[[430, 530, 471, 561]]</box> spotted on the floor.\n\nLayer 4: The <ref>wall</ref><box>[[0, 165, 999, 644]]</box> is primarily in view, with a <ref>remote</ref><box>[[316, 502, 344, 528]]</box> resting on the dining table.\n\nLayer 5: Over the <ref>wall</ref>, a <ref>curtain</ref><box>[[123, 165, 501, 566]]</box> is hanging.\n\nLayer 6: Lastly, on another <ref>table</ref><box>[[0, 681, 207, 833]]</box>, another <ref>remote</ref><box>[[12, 703, 160, 797]]</box> can be found, along with some <ref>paper</ref><box>[[109, 711, 181, 751]]</box>."
}
] |
coco/train2017/000000068751.jpg
|
00000077
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nA <ref>handbag</ref><box>[[961, 469, 994, 508]]</box> is visible closest to the camera. A <ref>man</ref><box>[[48, 333, 168, 699]]</box> is also present, standing on a <ref>pavement</ref><box>[[0, 483, 999, 831]]</box>.\n\nLayer 2: \nThere's another <ref>handbag</ref><box>[[911, 432, 961, 536]]</box> further back. The <ref>man</ref><box>[[48, 333, 168, 699]]</box> and <ref>pavement</ref><box>[[0, 483, 999, 831]]</box> continue to be visible in this layer.\n\nLayer 3: \nA <ref>backpack</ref><box>[[849, 437, 888, 489]]</box> is carried by a <ref>man</ref><box>[[571, 386, 641, 588]]</box>, who stands next to the previously mentioned man. Another <ref>man</ref><box>[[48, 333, 168, 699]]</box> and a <ref>pavement</ref><box>[[0, 483, 999, 831]]</box> are also in this layer. Additionally, a group of <ref>people</ref><box>[[849, 414, 914, 561]]</box> can be seen scattered on the pavement. A small section of <ref>grass</ref><box>[[0, 451, 767, 644]]</box> is visible, alongside a distant <ref>tent</ref><box>[[810, 360, 999, 444]]</box>.\n\nLayer 4: \nA <ref>person</ref><box>[[831, 420, 850, 513]]</box> is part of a larger crowd. The brick <ref>walkway</ref><box>[[20, 296, 958, 483]]</box> and some <ref>trees</ref><box>[[0, 165, 999, 561]]</box> encompass much of the scene. Parts of the <ref>grass</ref><box>[[0, 451, 767, 644]]</box> and <ref>tent</ref><box>[[810, 360, 999, 444]]</box> can still be observed from this depth.\n\nLayer 5: \nA <ref>skateboard</ref><box>[[572, 586, 618, 599]]</box> is introduced into the scene where the <ref>man</ref><box>[[48, 333, 168, 699]]</box> continues to be present on the <ref>pavement</ref><box>[[0, 483, 999, 831]]</box>. More distant, a <ref>group of people</ref><box>[[956, 400, 997, 575], [899, 424, 914, 511], [849, 414, 914, 561]]</box> appears to be passing by. The <ref>grass</ref><box>[[0, 451, 767, 644]]</box> area and <ref>tent</ref><box>[[810, 360, 999, 444]]</box> in the backdrop add layers to the setting.\n\nLayer 6: \nIn the farthest layers, a <ref>skateboard</ref><box>[[42, 685, 176, 733]]</box> is positioned on the <ref>pavement</ref><box>[[0, 483, 999, 831]]</box>."
}
] |
coco/train2017/000000128957.jpg
|
00000078
| 424
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nIn the image, there is a <ref>tie</ref><box>[[813, 441, 846, 526]]</box> which is being <pred>worn by</pred><box>[[813, 441, 846, 526]]</box><box>[[733, 386, 869, 716]]</box> a <ref>person</ref><box>[[733, 386, 869, 716]]</box>. This person is <pred>walking on</pred><box>[[733, 386, 869, 716]]</box><box>[[0, 168, 999, 830]]</box> the <ref>pavement</ref><box>[[0, 168, 999, 830]]</box>, and is <pred>beside</pred><box>[[733, 386, 869, 716]]</box><box>[[3, 169, 999, 639]]</box> a large <ref>airplane</ref><box>[[3, 169, 999, 639]]</box> on the same pavement.\n\nLayer 2: \nTwo other <ref>people</ref><box>[[53, 260, 134, 377], [168, 265, 254, 360]]</box> are present, with one <pred>beside</pred><box>[[53, 260, 134, 377]]</box><box>[[168, 265, 254, 360]]</box> the other. Both are <pred>walking on</pred><box>[[53, 260, 134, 377], [168, 265, 254, 360]]</box><box>[[0, 168, 999, 830], [0, 168, 999, 830]]</box> the <ref>pavement</ref><box>[[0, 168, 999, 830]]</box>, and appear to be focused on potentially inspecting or working on the <ref>airplane</ref><box>[[3, 169, 999, 639]]</box> nearby."
}
] |
coco/train2017/000000499584.jpg
|
00000079
| 441
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A small <ref>mountain</ref><box>[[95, 304, 160, 361]]</box> stands prominently in the forefront. An expanse of the <ref>sea</ref><box>[[0, 328, 703, 522]]</box> stretches behind it, with a <ref>boat</ref><box>[[502, 211, 828, 360]]</box> driving on it. A <ref>fence</ref><box>[[65, 286, 999, 541]]</box> forms part of the scene, providing structure to the view.\n\nLayer 2: A substantial <ref>wall</ref><box>[[796, 260, 999, 344]]</box> creates texture, and a broad <ref>cobblestone walkway</ref><box>[[182, 421, 877, 566]]</box> acts as a platform near the water.\n\nLayer 3: A <ref>bicycle</ref><box>[[760, 381, 956, 574]]</box> is seen leaning by the side, adding an element of leisure to the image.\n\nLayer 4: Central to the activity, a <ref>young boy</ref><box>[[596, 390, 808, 787]]</box> is actively holding and interacting with a <ref>kite</ref><box>[[532, 313, 578, 375]]</box>, part of which is in Layer 5 as well.\n\nLayer 5: The remainder of the <ref>kite</ref><box>[[571, 375, 603, 494]]</box> seems to stand out slightly, suggesting a dynamic movement in the scene.\n\nLayer 6: A distant tapestry of <ref>buildings</ref><box>[[148, 180, 999, 358]]</box> rests against the horizon. Lush <ref>trees</ref><box>[[0, 204, 999, 349]]</box> appear alongside the sea, forming a picturesque landscape.\n\nLayer 7: The vast <ref>sky</ref><box>[[0, 154, 999, 349]]</box> stretches above, completing the serene outdoor scene."
}
] |
coco/train2017/000000003860.jpg
|
00000080
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \n- A couple of <ref>people</ref><box>[[983, 491, 999, 545]]</box> are in the scene. \n- The <ref>trees</ref><box>[[0, 124, 999, 539]]</box> form a pleasant backdrop. \n- The <ref>sky</ref><box>[[341, 124, 999, 334]]</box> spreads beautifully above. \n- In the distance, a <ref>building</ref><box>[[24, 124, 999, 360]]</box> is visible. \n\nLayer 2: \n- Another <ref>person</ref><box>[[713, 481, 730, 545]]</box> appears nearby. \n- One <ref>horse</ref><box>[[733, 486, 783, 574]]</box> stands close by. \n- There are more <ref>trees</ref><box>[[0, 124, 999, 539]]</box> towering above. \n- The <ref>building</ref><box>[[24, 124, 999, 360]]</box> extends further along the horizon. \n\nLayer 3: \n- A <ref>person</ref><box>[[694, 490, 722, 562]]</box> interacts with the environment. \n- Another <ref>horse</ref><box>[[733, 486, 783, 574]]</box> accompanies individuals in the scene. \n- <ref>Trees</ref><box>[[0, 124, 999, 539]]</box> continue to provide shade and beauty. \n- The <ref>building</ref><box>[[24, 124, 999, 360]]</box> can still be seen prominently. \n\nLayer 4: \n- A solitary <ref>person</ref><box>[[51, 519, 62, 538]]</box> is positioned further away. \n- <ref>Trees</ref><box>[[0, 124, 999, 539]]</box> create a lush background. \n- The expansive <ref>sky</ref><box>[[341, 124, 999, 334]]</box> adds to the ambiance. \n- A <ref>building</ref><box>[[24, 124, 999, 360]]</box> stands far back, completing the scene. \n\nLayer 5: \n- Surrounded by activities are some <ref>people</ref><box>[[508, 519, 571, 544]], [[494, 514, 508, 544]], [[833, 503, 853, 562]], [[856, 513, 878, 577]], [[889, 486, 933, 577]]</box>. \n- Another <ref>horse</ref><box>[[733, 486, 783, 574]]</box> remains a part of the gathering. \n- Nearby, various <ref>benches</ref><box>[[212, 523, 276, 549]]</box> are scattered around. \n- <ref>Trees</ref><box>[[0, 124, 999, 539]]</box> enhance the scenery. \n- The <ref>sky</ref><box>[[341, 124, 999, 334]]</box> continues to provide a spacious feel. \n- A <ref>building</ref><box>[[24, 124, 999, 360]]</box> anchors the backdrop. \n\nLayer 6: \n- A <ref>person</ref><box>[[463, 480, 488, 572]]</box> adds life to the image's depth. \n- Nearby, a <ref>horse</ref><box>[[404, 480, 496, 575]]</box> appears interacting with the person. \n- The <ref>grassy area</ref><box>[[0, 497, 911, 874]]</box> enhances the ground texture. \n- <ref>Trees</ref><box>[[0, 124, 999, 539]]</box> are a constant feature lining the horizon. \n- A <ref>building</ref><box>[[24, 124, 999, 360]]</box> completes the look. \n\nLayer 7: \n- A few <ref>benches</ref><box>[[0, 532, 20, 552], [494, 514, 508, 544], [856, 513, 878, 577], [889, 486, 933, 577], [212, 523, 276, 549]]</box> dot the verdant <ref>grass</ref><box>[[0, 497, 911, 874]]</box>. \n- <ref>Trees</ref><box>[[0, 124, 999, 539]]</box> continue to outline the natural arena. \n- The encompassing <ref>sky</ref><box>[[341, 124, 999, 334]]</box> maintains an ethereal charm. \n- A distant <ref>building</ref><box>[[24, 124, 999, 360]]</box> stands on the horizon. "
}
] |
coco/train2017/000000352357.jpg
|
00000081
| 640
| 480
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: The image features a <ref>light</ref><box>[[724, 99, 776, 159]]</box> fixture, with three <ref>women</ref><box>[[577, 42, 851, 674], [227, 72, 806, 379], [236, 53, 422, 488]]</box> posed in the scene. Next to them, a section of a <ref>wooden floor</ref><box>[[753, 553, 874, 725]]</box> is visible.\n\nLayer 2: A portion of the <ref>ceiling</ref><box>[[669, 0, 874, 109]]</box> is seen in the top right corner.\n\nLayer 3: The composition includes a <ref>wooden wall</ref><box>[[647, 0, 874, 227]]</box> situated near the women.\n\nLayer 4: A <ref>vase</ref><box>[[442, 590, 639, 905]]</box> containing <ref>flowers</ref><box>[[288, 190, 776, 598]]</box> rests on a <ref>dining table</ref><box>[[124, 363, 874, 992]]</box>.\n\nLayer 5: One of the <ref>handbags</ref><box>[[240, 483, 442, 654]]</box> rests on the table among other items.\n\nLayer 6: Another <ref>handbag</ref><box>[[126, 472, 187, 593]]</box> is present on the surface of the table.\n\nLayer 7: The final <ref>handbag</ref><box>[[124, 476, 299, 669]]</box> can also be found on the table, adding to the scene."
}
] |
coco/train2017/000000271011.jpg
|
00000082
| 474
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: The scene features a <ref>fence</ref><box>[[0, 129, 243, 218]]</box> alongside a majestic <ref>mountain</ref><box>[[363, 204, 999, 316]]</box> with the clear <ref>sky</ref><box>[[148, 129, 999, 249]]</box> stretching over it. In the same layer, there is an array of <ref>snow</ref><box>[[0, 177, 999, 869]]</box> covering the ground, creating a perfect slope.\n\nLayer 2: A group of <ref>people</ref><box>[[622, 271, 916, 806]]</box><box>[[0, 400, 508, 869]]</box><box>[[753, 177, 905, 363]]</box> are engaging with their <ref>snowboards</ref><box>[[408, 638, 508, 859]]</box> near the ground. At this depth, some individuals are carrying their <ref>backpacks</ref><box>[[749, 285, 922, 438]]</box><box>[[0, 490, 162, 686]]</box><box>[[237, 465, 380, 569]]</box>.\n\nLayer 3: Additional <ref>people</ref><box>[[753, 177, 905, 363]]</box> can be found here using <ref>snowboards</ref><box>[[707, 507, 959, 592]]</box> as they interact on the slope. Concurrently, <ref>backpacks</ref><box>[[749, 285, 922, 438]]</box> accompany them.\n\nLayer 4: More <ref>people</ref><box>[[753, 177, 905, 363]]</box> are stationed along the slope with their <ref>snowboards</ref><box>[[707, 507, 959, 592]]</box>.\n\nLayer 5: Further away from the viewer, other <ref>people</ref><box>[[430, 279, 627, 461]]</box> can be observed with their gear.\n\nLayer 6: This includes the remaining <ref>people</ref><box>[[226, 391, 636, 692]]</box><box>[[622, 271, 916, 806]]</box><box>[[847, 191, 885, 251]]</box> sustaining their activities, accompanied by <ref>snowboards</ref><box>[[568, 531, 671, 756]]</box> and carrying more <ref>backpacks</ref><box>[[749, 285, 922, 438]]</box>."
}
] |
coco/train2017/000000581466.jpg
|
00000083
| 478
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>window</ref><box>[[99, 171, 999, 265]]</box> is present in the scene.\n\nLayer 2: There is a <ref>cow</ref><box>[[878, 252, 999, 310]]</box> in the image.\n\nLayer 3: \n- A <ref>cow</ref><box>[[784, 360, 997, 605]]</box> is seen.\n- Another <ref>cow</ref><box>[[568, 377, 988, 761]]</box> is nearby.\n- A <ref>cow</ref><box>[[385, 255, 485, 324]]</box> is part of the scene.\n- The <ref>fence</ref><box>[[262, 222, 999, 716]]</box> surrounds the area.\n- <ref>Grass</ref><box>[[67, 308, 999, 717]]</box> is in the vicinity.\n\nLayer 4: A <ref>ceiling</ref><box>[[553, 126, 999, 180]]</box> is overhead.\n\nLayer 5:\n- A <ref>cow</ref><box>[[273, 322, 496, 716]]</box> shares the space.\n- <ref>Cows</ref><box>[[3, 387, 149, 801]]</box> stand near the <ref>fence</ref><box>[[262, 222, 999, 716]]</box>.\n- The <ref>dirt</ref><box>[[0, 546, 649, 872]]</box> is underfoot.\n- <ref>Grass</ref><box>[[67, 308, 999, 717]]</box> lies around.\n\nLayer 6:\n- A <ref>cow</ref><box>[[0, 266, 168, 481]]</box> is spotted.\n- Another <ref>cow</ref><box>[[385, 255, 485, 324]]</box> is visible.\n- There is a distinct <ref>cow</ref><box>[[215, 263, 355, 391]]</box>.\n- A <ref>cow</ref><box>[[814, 274, 994, 364]]</box> stands out.\n- Yet another <ref>cow</ref><box>[[62, 258, 193, 396]]</box> populates the area.\n- Along with them, <ref>grass</ref><box>[[67, 308, 999, 717]]</box> is distributed.\n\nLayer 7: A <ref>wall</ref><box>[[0, 126, 999, 369]]</box> constructs part of the enclosure."
}
] |
coco/train2017/000000176588.jpg
|
00000084
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the foreground, a <ref>roof</ref><box>[[866, 332, 984, 398]]</box> of a tent structure is visible. Alongside, a <ref>person</ref><box>[[65, 463, 95, 539]]</box> appears actively engaged, possibly in play. Nearby are other <ref>people</ref><box>[[374, 471, 418, 625], [649, 469, 732, 655], [301, 477, 359, 687]]</box> participating in outdoor activities. The scene is part of a larger area with multiple <ref>people</ref><box>[[48, 425, 872, 719]]</box> enjoying together. This lively activity occurs <pred>on</pred><box>[[0, 458, 999, 831]]</box> the expansive <ref>dirt</ref><box>[[0, 458, 999, 831]]</box>. Background structures like a <ref>building</ref><box>[[286, 364, 999, 623]]</box> and the shade of a <ref>tree</ref><box>[[0, 165, 999, 505]]</box> complete the setting.\n\nLayer 2: At this level, another <ref>person</ref><box>[[863, 497, 895, 550]]</box> joins the group, amidst others <ref>people</ref><box>[[65, 463, 95, 539], [374, 471, 418, 625], [649, 469, 732, 655], [301, 477, 359, 687]]</box> engaging in their activities. The actions continue across the wide <ref>dirt</ref><box>[[0, 458, 999, 831]]</box> that encompasses the area, including the busy <ref>building</ref><box>[[286, 364, 999, 623]]</box> in the background.\n\nLayer 3: In this layer, we see a <ref>sports ball</ref><box>[[72, 613, 103, 642]]</box> involved with one <ref>person</ref><box>[[65, 463, 95, 539]]</box>. A group of <ref>people</ref><box>[[388, 635, 519, 820], [4, 457, 85, 711], [142, 420, 255, 786]]</box> are part of the activities taking place. This is all set against the earthy background of the <ref>dirt</ref><box>[[0, 458, 999, 831]]</box>.\n\nLayer 4: More <ref>people</ref><box>[[658, 529, 718, 669], [65, 463, 95, 539], [885, 485, 969, 700], [4, 457, 85, 711], [301, 477, 359, 687]]</box> can be seen scattered throughout the scene. The refreshingly green <ref>grass</ref><box>[[503, 569, 999, 725]]</box> enriches the view, outlined by the widespread <ref>dirt</ref><box>[[0, 458, 999, 831]]</box>. Outlying structures such as the prominent <ref>building</ref><box>[[286, 364, 999, 623]]</box> add depth to the setting.\n\nLayer 5: Yet more <ref>people</ref><box>[[561, 502, 596, 669], [65, 463, 95, 539], [649, 469, 732, 655], [301, 477, 359, 687]]</box> populate the scene, dynamically contributing to the suburb-like atmosphere of the <ref>dirt</ref><box>[[0, 458, 999, 831]]</box>. This bustling activity takes place against the taller backdrop of a <ref>building</ref><box>[[286, 364, 999, 623]]</box>.\n\nLayer 6: Within this setting, a handful of <ref>chairs</ref><box>[[530, 489, 558, 547]]</box> are positioned neatly, flanked by active <ref>people</ref><box>[[65, 463, 95, 539], [649, 469, 732, 655], [293, 453, 306, 495], [301, 477, 359, 687]]</box>. The sprawling <ref>dirt</ref><box>[[0, 458, 999, 831]]</box> continues to define the area, with sturdy structures like the <ref>building</ref><box>[[286, 364, 999, 623]]</box> looming in the expanse, under the calming influence of the <ref>tree</ref><box>[[0, 165, 999, 505]]</box>.\n\nLayer 7: People are aligned featuring prominently, including <ref>people</ref><box>[[49, 483, 101, 683], [388, 635, 519, 820], [4, 457, 85, 711], [142, 420, 255, 786]]</box> actively participating in the lively surroundings. Again, this bustling scene unfolds on the <ref>dirt</ref><box>[[0, 458, 999, 831]]</box>.\n\nLayer 8: Placed thoughtfully, comforting <ref>chairs</ref><box>[[440, 489, 460, 543]]</box> offer spots for relaxation amidst the active <ref>people</ref><box>[[65, 463, 95, 539], [649, 469, 732, 655], [293, 453, 306, 495], [301, 477, 359, 687]]</box>, encapsulated by the vast <ref>dirt</ref><box>[[0, 458, 999, 831]]</box> kingdom and accompanied by the robust <ref>building</ref><box>[[286, 364, 999, 623]]</box> and sheltering <ref>tree</ref><box>[[0, 165, 999, 505]]</box>.\n\nLayer 9: Overseeing the scene, a <ref>house</ref><box>[[0, 316, 380, 494]]</box> is nestled alongside the <ref>building</ref><box>[[286, 364, 999, 623]]</box> against the earthy backdrop of <ref>dirt</ref><box>[[0, 458, 999, 831]]</box>, completing the vibrant picture of village life."
}
] |
coco/train2017/000000193654.jpg
|
00000085
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nIn the scene, a <ref>room</ref><box>[[0, 212, 999, 831]]</box> encompasses the entire scene. In the room, a couple of the closest <ref>people</ref><box>[[623, 559, 718, 658]]</box> and <ref>people</ref><box>[[474, 573, 615, 821]]</box> are seen with another group, placing them near the foreground. One more <ref>person</ref><box>[[220, 573, 380, 800]]</box> can be observed as well.\n\nLayer 2: \nSlightly further back stands a <ref>person</ref><box>[[505, 378, 705, 513]]</box>, notably wearing a <ref>tie</ref><box>[[593, 444, 610, 503]]</box>. Also, seated in this layer are some attendees, including a seated <ref>person</ref><box>[[187, 491, 276, 649]]</box> and another <ref>person</ref><box>[[291, 486, 387, 635]]</box>, all depicting different postures as they likely focus on the speaker.\n\nLayer 3: \nTowards the middle distance, various other <ref>people</ref><box>[[586, 539, 647, 609]]</box> and<ref>people</ref><box>[[132, 538, 220, 621]]</box> can be identified, revealing a section of the audience engaging with the scene.\n\nLayer 4: \nFurther into the background of this image, a group of <ref>people</ref><box>[[279, 580, 568, 823]]</box> and <ref>person</ref><box>[[31, 613, 273, 821]]</box> contributes to the size and the engaged attendance in the room.\n\nLayer 5: \nLingering closer to the back, but still visibly expressive in the environment, a single <ref>person</ref><box>[[0, 583, 118, 829]]</box> amplifies the crowd's presence.\n\nLayer 6: \nFinally, providing context and setting for the entire event, a <ref>wall</ref><box>[[0, 165, 999, 578]]</box> frames and finalizes the background of this significant gathering space, completing the atmosphere of the room."
}
] |
coco/train2017/000000034120.jpg
|
00000086
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>friend</ref><box>[[326, 302, 548, 672]]</box> is <pred>sitting on</pred><box>[[326, 302, 548, 672]]</box> the <ref>couch</ref><box>[[1, 385, 344, 698]]</box>. There is a <ref>Nintendo Wii controller</ref><box>[[802, 349, 817, 367]]</box> and a <ref>rug</ref><box>[[49, 588, 999, 874]]</box> nearby. A large <ref>window</ref><box>[[0, 124, 462, 454]]</box> and a <ref>wall</ref><box>[[449, 124, 999, 647]]</box> are also part of this layer.\n\nLayer 2: Another <ref>friend</ref><box>[[668, 271, 842, 692]]</box> is also <pred>sitting on</pred><box>[[668, 271, 842, 692]]</box> the <ref>rug</ref><box>[[49, 588, 999, 874]]</box>, enjoying the game with a <ref>Nintendo Wii controller</ref><box>[[719, 418, 738, 433]]</box> in hand.\n\nLayer 3: There's a <ref>potted plant</ref><box>[[535, 370, 591, 466]]</box> nearby.\n\nLayer 4: A <ref>bottle</ref><box>[[511, 444, 524, 481]]</box> sits on a piece of furniture. \n\nLayer 5: Overhead, a <ref>light</ref><box>[[505, 124, 573, 180]]</box> illuminates the room. \n\nLayer 6: Another <ref>bottle</ref><box>[[491, 441, 507, 474]]</box> stands nearby.\n\nLayer 7: A <ref>potted plant</ref><box>[[449, 184, 529, 243]]</box> is seen against the wall.\n\nLayer 8: A different <ref>friend</ref><box>[[155, 421, 432, 838]]</box> is <pred>sitting on</pred><box>[[155, 421, 432, 838]]</box> the <ref>rug</ref><box>[[49, 588, 999, 874]]</box>. Another <ref>Nintendo Wii controller</ref><box>[[429, 402, 443, 427]]</box> is being used in this layer.\n\nLayer 9: A <ref>friend</ref><box>[[415, 230, 472, 385]]</box> stands nearby, contributing to the lively atmosphere.\n\nLayer 10: A <ref>friend</ref><box>[[3, 692, 151, 874]]</box> <pred>holding</pred><box>[[3, 692, 151, 874]]</box> a <ref>cup</ref><box>[[49, 770, 92, 817]]</box> can be seen in the scene. \n\nLayer 11: In the background, a <ref>friend</ref><box>[[118, 205, 218, 416]]</box> stands while furniture like <ref>chairs</ref><box>[[527, 308, 597, 418]]</box> frames the setting."
}
] |
coco/train2017/000000075183.jpg
|
00000087
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: The living room has a <ref>window</ref><box>[[839, 124, 999, 444]]</box> allowing natural light, and a <ref>wall</ref><box>[[0, 124, 895, 549]]</box> where various items are placed.\n\nLayer 2: A <ref>person</ref><box>[[697, 372, 860, 613]]</box> is seated on a <ref>couch</ref><box>[[101, 409, 630, 595]]</box>, with a <ref>laptop</ref><box>[[565, 535, 693, 622]]</box> nearby.\n\nLayer 3: Another <ref>person</ref><box>[[88, 628, 457, 865]]</box> is <pred>sitting on</pred><box>[[88, 628, 457, 865]]</box><box>[[1, 614, 496, 874]]</box> a <ref>couch</ref><box>[[1, 614, 496, 874]]</box>, accompanied by a <ref>laptop</ref><box>[[629, 722, 774, 808]]</box> <pred>on</pred><box>[[629, 722, 774, 808]]</box><box>[[507, 595, 902, 874]]</box> a <ref>table</ref><box>[[507, 595, 902, 874]]</box>.\n\nLayer 4: A <ref>laptop</ref><box>[[538, 625, 672, 686]]</box> is place on a lower <pred>surface</pred><box>[[538, 625, 672, 686]]</box>.\n\nLayer 5: There is a <ref>person</ref><box>[[29, 466, 344, 714]]</box> <pred>beside</pred><box>[[29, 466, 344, 714]]</box><box>[[195, 471, 588, 630]]</box> another, engaged with a <ref>laptop</ref><box>[[142, 538, 279, 639]]</box>.\n\nLayer 6: A <ref>mirror</ref><box>[[151, 172, 323, 350]]</box> <pred>hanging from</pred><box>[[151, 172, 323, 350]]</box><box>[[0, 124, 895, 549]]</box> and a <ref>couch</ref><box>[[741, 403, 997, 753]]</box> is also present in the room. \n\nLayer 7: A <ref>laptop</ref><box>[[142, 538, 279, 639]]</box> is being <pred>used by</pred><box>[[29, 466, 344, 714]]</box><box>[[142, 538, 279, 639]]</box> one of the <ref>people</ref><box>[[29, 466, 344, 714]]</box>."
}
] |
coco/train2017/000000285990.jpg
|
00000088
| 426
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: A <ref>person</ref><box>[[916, 300, 938, 346]]</box> is present on the <ref>snow</ref><box>[[0, 167, 999, 831]]</box>-covered slope.\n\nLayer 2: Nearby, a <ref>snowboard</ref><box>[[910, 336, 942, 346]]</box> can be seen lying on the snow next to <ref>skis</ref><box>[[902, 333, 947, 350]]</box>.\n\nLayer 3: A distant <ref>person</ref><box>[[789, 371, 792, 376]]</box> is present along with another <ref>person</ref><box>[[451, 481, 491, 550]]</box> who is <pred>playing</pred><box>[[451, 481, 491, 550]]</box> with a <ref>snowboard</ref><box>[[469, 533, 503, 555]]</box>. Another <ref>person</ref><box>[[227, 564, 257, 648]]</box> is present playing with <ref>skis</ref><box>[[213, 639, 251, 660]]</box>.\n\nLayer 4: Another <ref>person</ref><box>[[761, 388, 767, 403]]</box> stands in the snowy environment, partially obscured from the distance.\n\nLayer 5: A <ref>person</ref><box>[[718, 411, 724, 417]]</box> is standing still, the snow-capped backdrop extending into the distance.\n\nLayer 6: With a closer view, yet another <ref>person</ref><box>[[602, 449, 607, 461]]</box> is seen within the snow's embrace, along with the previously mentioned <ref>person</ref><box>[[451, 481, 491, 550]]</box>.\n\nLayer 7: Further <ref>people</ref><box>[[582, 466, 590, 479]]</box> are spread out across the snowy terrain.\n\nLayer 8: As we move slightly further back, another <ref>person</ref><box>[[566, 477, 573, 494]]</box> stands visible amidst the snowy setting.\n\nLayer 9: A <ref>person</ref><box>[[488, 475, 497, 493]]</box> is also seen, their location adding to the depth of the snowy expanse.\n\nLayer 10: The <ref>snowboard</ref><box>[[469, 533, 503, 555]]</box> is nestled into the snow near a <ref>person</ref><box>[[227, 564, 257, 648]]</box>.\n\nLayer 11: A pair of <ref>skis</ref><box>[[213, 639, 251, 660]]</box> intrudes into view, resting on the snowy slope.\n\nLayer 12: A <ref>person</ref><box>[[170, 599, 185, 626]]</box> and another <ref>person</ref><box>[[416, 547, 435, 569]]</box> are seen towards the edge of the captured scene, enveloped by the snowy charm.\n\nLayer 13: <ref>Skis</ref><box>[[165, 619, 187, 628]]</box> are positioned next to the previously mentioned <ref>person</ref><box>[[416, 547, 435, 569]]</box> near another <ref>person</ref><box>[[451, 481, 491, 550]]</box>.\n\nLayer 14: Overarching the winter scene is the <ref>sky</ref><box>[[0, 167, 880, 433]]</box>, expansively reaching across the horizon and framing the picturesque snowy landscape below."
}
] |
coco/train2017/000000326863.jpg
|
00000089
| 480
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nThere are <ref>people</ref><box>[[555, 408, 612, 529]]</box> visible, one of them is on the <ref>train</ref><box>[[371, 324, 749, 808]]</box>, which has a yellow, white, and black color scheme and is driving on a <ref>railroad</ref><box>[[51, 511, 977, 874]]</box> track.\n\nLayer 2: \nAnother <ref>person</ref><box>[[469, 458, 535, 529]]</box> is driving the train. The <ref>train</ref><box>[[371, 324, 749, 808]]</box> is on the railroad track, with <ref>grass</ref><box>[[0, 513, 961, 874]]</box> beside it and some <ref>dirt</ref><box>[[81, 724, 975, 874]]</box> below.\n\nLayer 3: \nIn the background, there is a <ref>house</ref><box>[[196, 313, 310, 353]]</box> that contributes to the suburban feel.\n\nLayer 4: \nA <ref>building</ref><box>[[135, 400, 387, 555]]</box> is situated amid the <ref>trees</ref><box>[[0, 124, 999, 667]]</box>. Above everything, the <ref>sky</ref><box>[[85, 124, 566, 325]]</box> stretches widely."
}
] |
coco/train2017/000000144620.jpg
|
00000090
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the upper part of the image, the <ref>sky</ref><box>[[488, 260, 847, 360]]</box> stretches across, appearing vast and clear.\n\nLayer 2: A <ref>cell phone</ref><box>[[451, 524, 472, 539]]</box> is being held by the <ref>woman</ref><box>[[329, 297, 521, 765]]</box> who is <pred>sitting on</pred><box>[[329, 297, 521, 765]]</box> a <ref>bench</ref><box>[[496, 439, 977, 734]]</box>. Alongside her, a <ref>man</ref><box>[[67, 266, 393, 811]]</box> is also <pred>sitting on</pred><box>[[67, 266, 393, 811]]</box> the same bench. Overlooking the entire scene is a <ref>tree</ref><box>[[0, 165, 999, 739]]</box> that provides shade.\n\nLayer 3: Beneath the <ref>bench</ref><box>[[496, 439, 977, 734]]</box>, the <ref>sand</ref><box>[[0, 669, 999, 831]]</box> stretches out, emphasizing the park setting.\n\nLayer 4: Out in the distance, a <ref>fence</ref><box>[[0, 341, 561, 430]]</box> completes the scene, positioned against the backdrop."
}
] |
coco/train2017/000000258085.jpg
|
00000091
| 432
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1:\n- The <ref>door-stuff</ref><box>[[82, 162, 167, 219]]</box> is closest to the camera.\n\nLayer 2:\n- A <ref>person</ref><box>[[630, 163, 699, 370]]</box> is visible.\n- A <ref>chair</ref><box>[[0, 239, 24, 280]]</box> is present on the <ref>floor</ref><box>[[0, 188, 999, 590]]</box>.\n- There is another <ref>chair</ref><box>[[67, 229, 132, 274]]</box>.\n\nLayer 3:\n- Multiple <ref>people</ref><box>[[120, 550, 340, 828], [519, 463, 607, 595], [366, 202, 540, 582], [0, 413, 85, 534]]</box> are standing in the scene. Some are <pred>holding</pred><box>[[1, 509, 174, 725]]</box><box>[[67, 588, 185, 732]]</box> what appears to be a piece of <ref>paper</ref><box>[[67, 588, 185, 732]]</box>.\n\nLayer 4:\n- A <ref>person</ref><box>[[349, 162, 407, 333]]</box>, dressed with a <ref>tie</ref><box>[[379, 193, 393, 262]]</box> is present.\n\nLayer 5:\n- A <ref>person</ref><box>[[310, 574, 499, 829]]</box> appears farther back.\n\nLayer 6:\n- The <ref>wall</ref><box>[[12, 162, 110, 230]]</box> forms the backdrop of the scene.\n\nLayer 7:\n- A <ref>chair</ref><box>[[110, 268, 182, 347]]</box> can be seen on the floor.\n\nLayer 8:\n- Another <ref>person</ref><box>[[1, 509, 174, 725]]</box> appears in the deeper part of the room.\n\nLayer 9:\n- A second <ref>chair</ref><box>[[0, 280, 110, 366]]</box> is visible.\n\nLayer 10:\n- The <ref>rug</ref><box>[[0, 266, 195, 375]]</box> is positioned on the floor.\n\nLayer 11:\n- Finally, another <ref>chair</ref><box>[[0, 239, 24, 280]]</box> can be seen stacked further away."
}
] |
coco/train2017/000000573549.jpg
|
00000092
| 640
| 480
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: \nA <ref>person</ref><box>[[694, 680, 756, 753]]</box> is close to the camera, being the nearest among the other people. Next to them is another <ref>person</ref><box>[[532, 693, 622, 903]]</box>. Nearby, a <ref>fire hydrant</ref><box>[[246, 817, 292, 911]]</box> is present on the sidewalk. Not far from the hydrant, another <ref>person</ref><box>[[396, 688, 471, 903]]</box> appears. The scene unfolds on a <ref>pavement</ref><box>[[238, 686, 793, 999]]</box>, where someone is <pred>riding</pred><box>[[532, 693, 622, 903]]</box><box>[[508, 784, 589, 903]]</box> a <ref>bicycle</ref><box>[[508, 784, 589, 903]]</box>. A <ref>building</ref><box>[[469, 0, 874, 999]]</box> can be seen looming over the scene, and a <ref>tree</ref><box>[[124, 0, 778, 999]]</box> adds a touch of greenery.\n\nLayer 2: \nIn the second layer, more depth is added with a <ref>person</ref><box>[[670, 721, 775, 980]]</box>, who stands further back, adding to the depth of field. The <ref>pavement</ref><box>[[238, 686, 793, 999]]</box> continues to extend back into this layer with the <ref>building</ref><box>[[469, 0, 874, 999]]</box> and <ref>tree</ref><box>[[124, 0, 778, 999]]</box> still in view.\n\nLayer 3: \nFurther into the scene, a <ref>person</ref><box>[[655, 718, 784, 945]]</box> is visible, forming the backdrop of people present in the scene. Like previous layers, the <ref>pavement</ref><box>[[238, 686, 793, 999]]</box>, <ref>building</ref><box>[[469, 0, 874, 999]]</box>, and <ref>tree</ref><box>[[124, 0, 778, 999]]</box> continue to provide structure and depth.\n\nLayer 4: \nA solitary <ref>person</ref><box>[[536, 671, 568, 719]]</box> appears at this distance, adding to the dimensionality of the scene.\n\nLayer 5: \nAnother individual comes into view <ref>person</ref><box>[[508, 677, 538, 750]]</box>, likely slightly obscured by layers ahead.\n\nLayer 6: \nThe <ref>ceiling</ref><box>[[448, 360, 874, 572]]</box> of the building garnishes the structural view, with the continuous presence of the <ref>tree</ref><box>[[124, 0, 778, 999]]</box> providing a natural element.\n\nLayer 7: \nWrapping elements like the <ref>fire hydrant</ref><box>[[246, 817, 292, 911]]</box> appear in this layer, neatly lying on the <ref>pavement</ref><box>[[238, 686, 793, 999]]</box>. The solidity of the <ref>building</ref><box>[[469, 0, 874, 999]]</box> and underlying presence of the <ref>tree</ref><box>[[124, 0, 778, 999]]</box> remain constant.\n\nLayer 8: \nThe scene is completed with a glimpse of the <ref>road</ref><box>[[124, 841, 161, 902]]</box>, grounding the overture with an element extending beyond the line of sight of the setup."
}
] |
coco/train2017/000000260951.jpg
|
00000093
| 427
| 640
|
[
{
"from": "human",
"value": "<image>\nDescribe the given image as a scene graph organized by depth. Group objects by their approximate distance from the camera, starting from nearest to farthest. For each layer, specify the objects and their relationships with other objects in the same layer."
},
{
"from": "gpt",
"value": "Layer 1: In the foreground, two <ref>jockeys</ref><box>[[438, 422, 600, 599], [103, 419, 238, 580]]</box> are seen.\n\nLayer 2: Nearby, a <ref>fence</ref><box>[[407, 533, 452, 547]]</box> can be observed, possibly part of the race track's boundary. An adjacent <ref>building</ref><box>[[666, 499, 999, 572]]</box> is also present.\n\nLayer 3: Slightly further in the distance, the <ref>horses</ref><box>[[237, 485, 696, 751], [1, 469, 377, 743]]</box> ridden by the jockeys are visible, running on the <ref>grassy track</ref><box>[[0, 471, 999, 831]]</box>.\n\nLayer 4: In the background, several <ref>trees</ref><box>[[0, 208, 999, 541]]</box> provide an additional layer of depth.\n\nLayer 5: The <ref>sky</ref><box>[[0, 165, 999, 306]]</box> spans above everything, creating an open and expansive atmosphere."
}
] |
End of preview.
README.md exists but content is empty.
- Downloads last month
- 64