The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
Error code: DatasetGenerationError
Exception: CastError
Message: Couldn't cast
id: string
image: string
question: string
answer: string
gpt: string
len: int64
human: string
to
{'id': Value('string'), 'image': Value('string'), 'human': Value('string'), 'gpt': Value('string'), 'len': Value('int64')}
because column names don't match
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1872, in _prepare_split_single
for key, table in generator:
^^^^^^^^^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 609, in wrapped
for item in generator(*args, **kwargs):
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 260, in _generate_tables
self._cast_table(pa_table, json_field_paths=json_field_paths),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 120, in _cast_table
pa_table = table_cast(pa_table, self.info.features.arrow_schema)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2272, in table_cast
return cast_table_to_schema(table, schema)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2218, in cast_table_to_schema
raise CastError(
datasets.table.CastError: Couldn't cast
id: string
image: string
question: string
answer: string
gpt: string
len: int64
human: string
to
{'id': Value('string'), 'image': Value('string'), 'human': Value('string'), 'gpt': Value('string'), 'len': Value('int64')}
because column names don't match
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1342, in compute_config_parquet_and_info_response
parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 907, in stream_convert_to_parquet
builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1739, in _prepare_split
for job_id, done, content in self._prepare_split_single(
^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1922, in _prepare_split_single
raise DatasetGenerationError("An error occurred while generating the dataset") from e
datasets.exceptions.DatasetGenerationError: An error occurred while generating the datasetNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
id string | image string | human string | gpt string | len int64 |
|---|---|---|---|---|
67-2021-0995-6880-LA93-0M20-E080_10_2 | BDOrtho_67-2021-0995-6880-LA93-0M20-E080_10.png | Can you find a Path in this scene? (yes/no) | yes | 15 |
74-2020-0935-6570-LA93-0M20-E080_884_03 | BDOrtho_74-2020-0935-6570-LA93-0M20-E080_884.png | Generate a polygon-based segmentation of the Building contained within the supplied bounding box. [499, 722, 650, 896] | [551, 896, 499, 857, 591, 722, 650, 765, 551, 896] | 89 |
31-2022-0575-6285-LA93-0M20-E080_120_7 | BDOrtho_31-2022-0575-6285-LA93-0M20-E080_120.png | Label what appears in: [614, 115, 634, 115, 631, 0, 611, 0, 614, 115] | Road | 52 |
67-2021-1035-6805-LA93-0M20-E080_250_2 | BDOrtho_67-2021-1035-6805-LA93-0M20-E080_250.png | Locate and outline Orchard with a segmentation mask (polygon) | [] | 17 |
45-2020-0615-6750-LA93-0M20-E080_1121_10 | BDOrtho_45-2020-0615-6750-LA93-0M20-E080_1121.png | Given the bounding box for Shunting yard, delineate the object inside using a detailed polygon segmentation. [281, 0, 845, 701] | [533, 0, 762, 652, 845, 701, 592, 609, 487, 545, 452, 488, 281, 0, 533, 0] | 116 |
67-2021-1030-6830-LA93-0M20-E080_900_3 | BDOrtho_67-2021-1030-6830-LA93-0M20-E080_900.png | Provide the appropriate class for what appears in: [1000, 631, 1000, 1000, 0, 1000, 0, 0, 890, 0, 1000, 631] | Industrial zone | 70 |
91-2021-0645-6850-LA93-0M20-E080_1140_1 | BDOrtho_91-2021-0645-6850-LA93-0M20-E080_1140.png | Using bounding boxes, detect all instances of Bridge | [88, 640, 685, 888] | 29 |
69-2020-0840-6495-LA93-0M20-E080_992_01 | BDOrtho_69-2020-0840-6495-LA93-0M20-E080_992.png | Using the provided bounding box and label Building, segment the object within the box using a polygon mask. [0, 494, 390, 972] | [0, 972, 0, 643, 312, 494, 390, 665, 290, 714, 335, 812, 0, 972] | 103 |
34-2021-0780-6285-LA93-0M20-E080_31_3 | BDOrtho_34-2021-0780-6285-LA93-0M20-E080_31.png | Label what appears in: [997, 311, 707, 394, 578, 464, 342, 666, 0, 894, 5, 912, 354, 682, 588, 480, 715, 412, 1000, 329, 997, 311] | Watercourse | 114 |
21-2020-0845-6680-LA93-0M20-E080_849_4 | BDOrtho_21-2020-0845-6680-LA93-0M20-E080_849.png | Provide the appropriate class for what appears in: [872, 894, 1000, 1000] | Church | 33 |
13-2020-0900-6255-LA93-0M20-E080_315_8 | BDOrtho_13-2020-0900-6255-LA93-0M20-E080_315.png | Is a Path included in the picture? (yes/no) | yes | 14 |
76-2019-0500-6935-LA93-0M20-E080_1297_2 | BDOrtho_76-2019-0500-6935-LA93-0M20-E080_1297.png | Locate and outline Factory with a segmentation mask (polygon) | [257, 199, 55, 1000, 0, 1000, 0, 129, 257, 199] | 61 |
34-2021-0790-6285-LA93-0M20-E080_135_5 | BDOrtho_34-2021-0790-6285-LA93-0M20-E080_135.png | Is a Orchard part of this scene? (yes/no) | yes | 16 |
94-2021-0655-6855-LA93-0M20-E080_386_5 | BDOrtho_94-2021-0655-6855-LA93-0M20-E080_386.png | Identify what is contained within: [999, 125, 921, 121, 915, 28, 878, 30, 872, 92, 644, 109, 635, 25, 606, 0, 1000, 0, 999, 125] | Building | 101 |
95-2021-0645-6875-LA93-0M20-E080_229_1 | BDOrtho_95-2021-0645-6875-LA93-0M20-E080_229.png | Provide a concise classification for the object shown in: [348, 840, 543, 978, 555, 962, 360, 824, 348, 840] | Highway ramp | 65 |
74-2020-0975-6535-LA93-0M20-E080_281_1 | BDOrtho_74-2020-0975-6535-LA93-0M20-E080_281.png | Draw bounding boxes to denote the positions of all Bridge present | [932, 439, 994, 459] | 32 |
44-2020-0350-6730-LA93-0M20-E080_646_3 | BDOrtho_44-2020-0350-6730-LA93-0M20-E080_646.png | Determine the category of the object found in: [323, 0, 1000, 0, 1000, 641, 210, 414, 323, 0] | Stadium | 58 |
33-2021-0415-6425-LA93-0M20-E080_1211_0 | BDOrtho_33-2021-0415-6425-LA93-0M20-E080_1211.png | Provide the appropriate class for what appears in: [747, 750, 900, 876, 805, 1000, 552, 1000, 747, 750] | Bridge | 63 |
34-2021-0745-6280-LA93-0M20-E080_106_08 | BDOrtho_34-2021-0745-6280-LA93-0M20-E080_106.png | Segment the object labeled Building inside the given bounding box using a detailed polygon. [185, 676, 290, 781] | [185, 743, 243, 676, 290, 719, 232, 781, 185, 743] | 88 |
44-2020-0365-6695-LA93-0M20-E080_973_1 | BDOrtho_44-2020-0365-6695-LA93-0M20-E080_973.png | Provide a short label for: [5, 674, 1000, 158, 996, 140, 0, 656, 5, 674] | Gravel road | 55 |
59-2021-0660-7100-LA93-0M20-E080_969_5 | BDOrtho_59-2021-0660-7100-LA93-0M20-E080_969.png | Give the correct label for the object pictured in: [666, 569, 219, 600, 4, 579, 2, 599, 219, 620, 403, 617, 1000, 532, 998, 512, 666, 569] | Road | 100 |
21-2020-0795-6705-LA93-0M20-E080_914_4 | BDOrtho_21-2020-0795-6705-LA93-0M20-E080_914.png | Draw polygon-based segmentation masks around all detected Building | [369, 833, 393, 795, 402, 805, 414, 793, 443, 826, 399, 866, 369, 833], [751, 71, 735, 77, 705, 0, 939, 0, 943, 30, 842, 71, 762, 101, 751, 71] | 153 |
76-2019-0545-6935-LA93-0M20-E080_232_3 | BDOrtho_76-2019-0545-6935-LA93-0M20-E080_232.png | Does the picture contain a Road? (yes/no) | yes | 13 |
91-2021-0650-6845-LA93-0M20-E080_981_215 | BDOrtho_91-2021-0650-6845-LA93-0M20-E080_981.png | Segment the object inside the bounding box labeled Building and output its polygonal mask. [399, 73, 499, 174] | [499, 135, 451, 174, 399, 112, 446, 73, 499, 135] | 88 |
44-2020-0325-6695-LA93-0M20-E080_298_3 | BDOrtho_44-2020-0325-6695-LA93-0M20-E080_298.png | Using a bounding box, locate the described item: The heritage site positioned north of the large sports field and near the top edge of the image. | [308, 850, 430, 967] | 52 |
34-2021-0705-6255-LA93-0M20-E080_1365_3 | BDOrtho_34-2021-0705-6255-LA93-0M20-E080_1365.png | Is the Market present in this image? (yes/no) | no | 14 |
59-2021-0685-7070-LA93-0M20-E080_1215_4 | BDOrtho_59-2021-0685-7070-LA93-0M20-E080_1215.png | Assign a label to the item located in: [49, 0, 189, 0, 515, 283, 427, 384, 360, 326, 341, 348, 272, 289, 292, 266, 158, 149, 138, 171, 70, 113, 89, 91, 22, 31, 49, 0] | Building | 138 |
44-2020-0350-6700-LA93-0M20-E080_960_2 | BDOrtho_44-2020-0350-6700-LA93-0M20-E080_960.png | Label the contents of: [9, 220, 0, 230, 391, 1000, 409, 995, 9, 220] | Road | 51 |
34-2021-0785-6295-LA93-0M20-E080_532_8 | BDOrtho_34-2021-0785-6295-LA93-0M20-E080_532.png | Is a Vineyard part of this scene? (yes/no) | yes | 16 |
13-2020-0875-6260-LA93-0M20-E080_21_1 | BDOrtho_13-2020-0875-6260-LA93-0M20-E080_21.png | Assign an appropriate category to the item visible in: [1000, 688, 888, 704, 888, 644, 1000, 624, 1000, 688] | Vegetation | 67 |
95-2021-0630-6885-LA93-0M20-E080_486_1 | BDOrtho_95-2021-0630-6885-LA93-0M20-E080_486.png | Classify the visual content of: [1000, 652, 896, 532, 888, 455, 905, 421, 1000, 335, 994, 319, 889, 409, 868, 449, 878, 542, 993, 666, 1000, 652] | Road | 121 |
33-2021-0455-6390-LA93-0M20-E080_197_6 | BDOrtho_33-2021-0455-6390-LA93-0M20-E080_197.png | Assign a label to the item located in: [357, 164, 323, 235, 284, 220, 247, 235, 237, 295, 216, 317, 55, 310, 0, 280, 0, 165, 36, 186, 104, 193, 140, 179, 149, 153, 135, 101, 188, 67, 192, 0, 262, 0, 281, 57, 326, 94, 329, 131, 357, 164]. Use one of the following labels: ['Market', 'Heliport', 'Canal', 'Aquaculture', 'Shunting yard', 'Campground', 'Industrial zone', 'Water tower', 'Woodland', 'Bicycle path', 'Railroad', 'Drinking water production plant', 'Ecomuseum', 'Motorway type', 'Aqueduct', 'Indoor sports complex', 'Zoo', 'Dam reservoir'] | Woodland | 315 |
31-2022-0585-6295-LA93-0M20-E080_240_20 | BDOrtho_31-2022-0585-6295-LA93-0M20-E080_240.png | Produce a precise polygon segmentation for the Small multi-sports field instance enclosed by the given bounding box. [690, 60, 1000, 753] | [731, 146, 775, 225, 848, 124, 893, 88, 1000, 60, 1000, 753, 690, 165, 731, 146] | 125 |
13-2020-0890-6275-LA93-0M20-E080_925_4 | BDOrtho_13-2020-0890-6275-LA93-0M20-E080_925.png | Classify the visual content of: [574, 505, 5, 161, 0, 179, 562, 521, 627, 587, 641, 573, 574, 505] | Road | 74 |
92-2021-0645-6860-LA93-0M20-E080_1124_13 | BDOrtho_92-2021-0645-6860-LA93-0M20-E080_1124.png | Classify the visual content of: [769, 317, 860, 430, 812, 472, 722, 362, 769, 317]. Pick from: ['Archaeological remains', 'Highway ramp', 'Shunting yard', 'Vehicle service area', 'Reservoir', 'Salt marsh', 'Passenger stop', 'Air traffic control tower', 'Tramway', 'Tennis court', 'Drinking water production plant', 'Church', 'Water tower', 'Forestry house', 'Sand', 'Dam reservoir', 'Rest or service area', 'Subway'] | Tennis court | 163 |
59-2021-0660-7065-LA93-0M20-E080_580_17 | BDOrtho_59-2021-0660-7065-LA93-0M20-E080_580.png | Identify what is contained within: [885, 475, 939, 461, 943, 487, 890, 499, 885, 475] | Building | 58 |
92-2021-0640-6860-LA93-0M20-E080_504_3 | BDOrtho_92-2021-0640-6860-LA93-0M20-E080_504.png | Give the correct label for the object pictured in: [994, 184, 600, 455, 612, 471, 1000, 200, 994, 184] | Road | 64 |
59-2021-0710-7065-LA93-0M20-E080_388_4 | BDOrtho_59-2021-0710-7065-LA93-0M20-E080_388.png | Classify the item that appears in: [170, 0, 339, 132] | Building | 27 |
33-2021-0410-6425-LA93-0M20-E080_944_17 | BDOrtho_33-2021-0410-6425-LA93-0M20-E080_944.png | Identify the object by naming what appears in: [49, 0, 173, 100] | Building | 28 |
13-2020-0830-6290-LA93-0M20-E080_240_9 | BDOrtho_13-2020-0830-6290-LA93-0M20-E080_240.png | State the category of the object depicted in: [725, 137, 768, 218, 702, 259, 818, 438, 834, 428, 730, 265, 792, 226, 743, 127, 725, 137] | Path | 101 |
33-2021-0450-6450-LA93-0M20-E080_508_2 | BDOrtho_33-2021-0450-6450-LA93-0M20-E080_508.png | Give the correct label for the object pictured in: [305, 55, 205, 452, 213, 601, 196, 659, 141, 725, 0, 1000, 0, 859, 68, 772, 109, 567, 164, 493, 217, 77, 268, 0, 341, 0, 305, 55] | Watercourse | 143 |
77-2021-0695-6885-LA93-0M20-E080_101_6 | BDOrtho_77-2021-0695-6885-LA93-0M20-E080_101.png | Classify the visual content of: [0, 509, 132, 775] | Large sports field | 29 |
76-2019-0555-6925-LA93-0M20-E080_526_2 | BDOrtho_76-2019-0555-6925-LA93-0M20-E080_526.png | Classify the visual content of: [175, 0, 0, 225, 7, 239, 191, 6, 175, 0] | Road | 48 |
44-2020-0365-6695-LA93-0M20-E080_290_3 | BDOrtho_44-2020-0365-6695-LA93-0M20-E080_290.png | Return a bounding box around the object referred to in the following description: The building located on the right side of the road and north of the forest. | [560, 759, 729, 887] | 51 |
92-2021-0640-6870-LA93-0M20-E080_1248_1 | BDOrtho_92-2021-0640-6870-LA93-0M20-E080_1248.png | Segment every Vegetation in the frame using a polygon mask | [522, 0, 0, 333, 0, 237, 30, 250, 110, 217, 125, 188, 110, 158, 88, 150, 89, 116, 228, 0, 522, 0] | 110 |
59-2021-0645-7080-LA93-0M20-E080_1060_6 | BDOrtho_59-2021-0645-7080-LA93-0M20-E080_1060.png | Is any Building depicted in this image? (yes/no) | yes | 15 |
33-2021-0435-6430-LA93-0M20-E080_539_1 | BDOrtho_33-2021-0435-6430-LA93-0M20-E080_539.png | Use bounding boxes to highlight all objects classified as Bridge | [747, 211, 813, 265] | 32 |
13-2020-0895-6255-LA93-0M20-E080_285_2 | BDOrtho_13-2020-0895-6255-LA93-0M20-E080_285.png | Can a Road be observed in this scene? (yes/no) | yes | 15 |
78-2021-0630-6860-LA93-0M20-E080_630_7 | BDOrtho_78-2021-0630-6860-LA93-0M20-E080_630.png | Identify the type of item present in: [0, 784, 333, 689, 488, 585, 635, 562, 693, 469, 827, 461, 871, 478, 940, 539, 975, 525, 1000, 491, 1000, 829, 958, 871, 896, 872, 839, 914, 680, 954, 659, 975, 518, 1000, 0, 1000, 0, 784] | Forest | 198 |
77-2021-0685-6870-LA93-0M20-E080_1259_3 | BDOrtho_77-2021-0685-6870-LA93-0M20-E080_1259.png | Can you identify a Canal in the picture? (yes/no) | yes | 15 |
69-2020-0825-6545-LA93-0M20-E080_283_10 | BDOrtho_69-2020-0825-6545-LA93-0M20-E080_283.png | Using a bounding box, locate the described item: The building situated in the top-center region, slightly south of the road. | [31, 219, 126, 315] | 46 |
77-2021-0680-6820-LA93-0M20-E080_729_16 | BDOrtho_77-2021-0680-6820-LA93-0M20-E080_729.png | Using the provided bounding box and label Building, segment the object within the box using a polygon mask. [0, 558, 305, 973] | [0, 917, 0, 826, 39, 851, 74, 793, 0, 746, 0, 574, 12, 558, 73, 586, 12, 693, 305, 882, 217, 973, 62, 866, 21, 930, 0, 917] | 162 |
33-2021-0465-6395-LA93-0M20-E080_556_5 | BDOrtho_33-2021-0465-6395-LA93-0M20-E080_556.png | Can you detect a Road in this image? (yes/no) | yes | 15 |
59-2021-0735-7045-LA93-0M20-E080_724_00 | BDOrtho_59-2021-0735-7045-LA93-0M20-E080_724.png | Segment the contents of the bounding box labeled Building and represent the shape as a polygon mask. [0, 249, 162, 418] | [0, 347, 129, 249, 162, 287, 0, 418, 0, 347] | 83 |
21-2020-0815-6690-LA93-0M20-E080_281_8 | BDOrtho_21-2020-0815-6690-LA93-0M20-E080_281.png | Is there a Zoo in the image? (yes/no) | no | 15 |
34-2021-0755-6315-LA93-0M20-E080_1214_7 | BDOrtho_34-2021-0755-6315-LA93-0M20-E080_1214.png | Give the correct label for the object pictured in: [338, 312, 336, 341, 289, 341, 289, 314, 338, 312] | Building | 63 |
33-2021-0435-6395-LA93-0M20-E080_1082_0 | BDOrtho_33-2021-0435-6395-LA93-0M20-E080_1082.png | Apply bounding boxes to outline all detected Building in the frame | [132, 0, 222, 85] | 30 |
13-2020-0855-6280-LA93-0M20-E080_857_1 | BDOrtho_13-2020-0855-6280-LA93-0M20-E080_857.png | Identify the type of item present in: [364, 0, 328, 1000, 348, 1000, 384, 0, 364, 0] | Canal | 56 |
21-2020-0845-6720-LA93-0M20-E080_834_16 | BDOrtho_21-2020-0845-6720-LA93-0M20-E080_834.png | Give a name to the item shown in: [582, 888, 802, 1000]. Select the label from: ['Water tower', 'Building', 'Port gantry', 'Archaeological remains', 'Service track', 'Heliport', 'Fire station', 'Dam', 'Pond', 'Ferry terminal', 'Lock', 'Military enclosure', 'Ecomuseum', 'Church', 'Zoo', 'Highway ramp', 'Pumping station', 'Orchard'] | Building | 123 |
95-2021-0615-6885-LA93-0M20-E080_614_12 | BDOrtho_95-2021-0615-6885-LA93-0M20-E080_614.png | Draw a bounding box around the object referenced below: The building weakly included within the vegetation patch in the top-right corner. | [362, 830, 455, 951] | 48 |
74-2020-0980-6535-LA93-0M20-E080_149 | BDOrtho_74-2020-0980-6535-LA93-0M20-E080_149.png | Use bounding boxes to mark the following items in the image: ['Cable car station', 'Church', 'Chapel'] | [705, 783, 874, 951], [726, 91, 1000, 368], [793, 435, 912, 554], [] | 89 |
76-2019-0560-6925-LA93-0M20-E080_311_5 | BDOrtho_76-2019-0560-6925-LA93-0M20-E080_311.png | Classify the item that appears in: [338, 135, 304, 128, 308, 104, 243, 89, 256, 33, 356, 57, 338, 135] | Building | 76 |
44-2020-0380-6720-LA93-0M20-E080_308 | BDOrtho_44-2020-0380-6720-LA93-0M20-E080_308.png | Write a caption describing this image | This is an outdoor area combining recreational and infrastructural elements. A large sports field dominates the bottom of the scene, extending across the center and right side. A building is situated on the bottom-left, close to a road that runs across the top of the image. Vegetation is present on the upper-left, and a reservoir is located on the upper-right, with the road bordering both of these features. The vegetation partially surrounds the reservoir. | 110 |
34-2021-0775-6285-LA93-0M20-E080_686_6 | BDOrtho_34-2021-0775-6285-LA93-0M20-E080_686.png | Provide a concise classification for the object shown in: [534, 552, 600, 610] | Building | 33 |
67-2021-1055-6865-LA93-0M20-E080_235_2 | BDOrtho_67-2021-1055-6865-LA93-0M20-E080_235.png | Return bounding boxes for every occurrence of Refuge | [] | 11 |
69-2020-0835-6525-LA93-0M20-E080_1277_0 | BDOrtho_69-2020-0835-6525-LA93-0M20-E080_1277.png | Identify each instance of Grandstand and mark it with a bounding box | [677, 42, 1000, 680], [0, 458, 287, 1000] | 54 |
13-2020-0885-6255-LA93-0M20-E080_1117_03 | BDOrtho_13-2020-0885-6255-LA93-0M20-E080_1117.png | Given a bounding box and label Building, derive a detailed polygon mask for the object inside. [613, 385, 701, 449] | [698, 385, 701, 445, 616, 449, 613, 390, 698, 385] | 89 |
78-2021-0625-6875-LA93-0M20-E080_544_3 | BDOrtho_78-2021-0625-6875-LA93-0M20-E080_544.png | Locate, with a bounding box, the object described: The building positioned in the top-left corner, distant from the roundabout. | [28, 0, 158, 76] | 46 |
34-2021-0705-6285-LA93-0M20-E080_816_0 | BDOrtho_34-2021-0705-6285-LA93-0M20-E080_816.png | Determine and outline the object described using a bounding box: The bridge situated in the top-right corner near the forest. | [753, 266, 1000, 471] | 48 |
59-2021-0665-7075-LA93-0M20-E080_545_5 | BDOrtho_59-2021-0665-7075-LA93-0M20-E080_545.png | Can you identify a Road in the picture? (yes/no) | yes | 15 |
33-2021-0415-6420-LA93-0M20-E080_477_5 | BDOrtho_33-2021-0415-6420-LA93-0M20-E080_477.png | Give a name to the item shown in: [286, 868, 384, 918, 333, 1000, 0, 1000, 0, 780, 286, 868]. Identify the correct label from the list: ['Water treatment plant', 'Water body', 'Intersection', 'Harbor basin', 'Ground water tank or tower', 'Marsh', 'Dam', 'Toll booth', 'Air traffic control tower', 'Shooting range', 'Swimming pool', 'Industrial zone', 'Triumphal arch', 'Large sports field', 'Roundabout', 'Monument', 'Racetrack', 'Aqueduct'] | Industrial zone | 179 |
txt_45741 | null | Compute the horizontal bounding box that contains the given polygon. [710, 449, 815, 764, 593, 849, 436, 619, 478, 430, 710, 449] | [436, 430, 815, 849] | 93 |
78-2021-0635-6870-LA93-0M20-E080_1001_1 | BDOrtho_78-2021-0635-6870-LA93-0M20-E080_1001.png | Return polygonal masks that capture the boundaries of all visible Sports track | [266, 240, 183, 302, 64, 327, 0, 323, 0, 82, 153, 214, 330, 0, 464, 0, 266, 240] | 95 |
69-2020-0840-6525-LA93-0M20-E080_96_5 | BDOrtho_69-2020-0840-6525-LA93-0M20-E080_96.png | Are you able to see a Transport facility in the frame? (yes/no) | no | 18 |
67-2021-1050-6845-LA93-0M20-E080_263_4 | BDOrtho_67-2021-1050-6845-LA93-0M20-E080_263.png | Identify the type of item present in: [531, 0, 511, 1, 538, 189, 558, 187, 531, 0]. Choose the most appropriate label from: ['Stairway', 'Road', 'Fire station', 'Subway station', 'Hard runway', 'Hospital facility', 'Amusement park', 'Shunting yard', 'Windmill', 'Campground', 'Path', 'Transport facility', 'Forestry house', 'Harbor basin', 'Watercourse', 'Silo', 'Hospital', 'Stadium'] | Road | 149 |
44-2020-0355-6690-LA93-0M20-E080_750_2 | BDOrtho_44-2020-0355-6690-LA93-0M20-E080_750.png | What is the object shown in: [363, 905, 259, 932, 106, 756, 223, 653, 238, 670, 272, 640, 335, 707, 256, 757, 329, 839, 345, 824, 376, 857, 360, 871, 363, 905] | Building | 138 |
33-2021-0420-6420-LA93-0M20-E080_237_0 | BDOrtho_33-2021-0420-6420-LA93-0M20-E080_237.png | Generate a polygon mask outlining every visible instance of Forest | [546, 312, 445, 266, 382, 189, 348, 101, 332, 0, 891, 0, 949, 62, 896, 146, 713, 286, 613, 311, 546, 312] | 118 |
93-2021-0650-6875-LA93-0M20-E080_1343_0 | BDOrtho_93-2021-0650-6875-LA93-0M20-E080_1343.png | Provide a short label for: [0, 536, 105, 1000, 125, 998, 16, 530, 0, 536]. Choose one of these options: ['Aquaculture', 'Sand', 'Orchard', 'Grandstand', 'Train station', 'Port', 'Road', 'Campground', 'Hop field', 'Greenhouse', 'Port gantry', 'Shooting stand', 'Toll booth', 'Vegetation', 'Salt marsh', 'Livestock farming', 'Pumping station', 'Rest or service area'] | Road | 156 |
34-2021-0720-6250-LA93-0M20-E080_880_11 | BDOrtho_34-2021-0720-6250-LA93-0M20-E080_880.png | Classify the item that appears in: [97, 896, 4, 902, 4, 570, 86, 564, 92, 688, 89, 713, 76, 714, 97, 896] | Building | 79 |
69-2020-0830-6510-LA93-0M20-E080_1146_2 | BDOrtho_69-2020-0830-6510-LA93-0M20-E080_1146.png | Can you see a Vineyard here? (yes/no) | yes | 15 |
33-2021-0400-6425-LA93-0M20-E080_115_0 | BDOrtho_33-2021-0400-6425-LA93-0M20-E080_115.png | For every Bridge found, return a bounding box surrounding it | [684, 586, 767, 695] | 32 |
34-2021-0755-6275-LA93-0M20-E080_683_7 | BDOrtho_34-2021-0755-6275-LA93-0M20-E080_683.png | Label the contents of: [210, 235, 162, 307, 143, 292, 124, 324, 68, 287, 94, 251, 33, 208, 57, 180, 106, 216, 173, 129, 219, 165, 188, 202, 210, 235] | Building | 132 |
45-2020-0615-6755-LA93-0M20-E080_971_01 | BDOrtho_45-2020-0615-6755-LA93-0M20-E080_971.png | Using the bounding box labeled Building as guidance, segment the enclosed object with a polygon mask. [696, 796, 729, 829] | [717, 829, 696, 810, 707, 796, 729, 817, 717, 829] | 91 |
13-2020-0865-6255-LA93-0M20-E080_505_3 | BDOrtho_13-2020-0865-6255-LA93-0M20-E080_505.png | Determine the category of the object found in: [513, 693, 571, 751]. Determine the right label from the following: ['Campground', 'Fort, bunker, casemate', 'Ice rink', 'Drinking water production plant', 'Canal', 'Building', 'Road', 'Castle', 'Shooting stand', 'Highway ramp', 'Path', 'Watercourse', 'Monument', 'Power plant', 'Tramway', 'Bicycle path', 'Bicycle service area', 'Silo'] | Building | 137 |
31-2022-0500-6195-LA93-0M20-E080_952_1 | BDOrtho_31-2022-0500-6195-LA93-0M20-E080_952.png | Generate a polygon mask outlining every visible instance of Airfield | [911, 121, 518, 0, 1000, 0, 1000, 22, 911, 121] | 61 |
13-2020-0860-6285-LA93-0M20-E080_399_0 | BDOrtho_13-2020-0860-6285-LA93-0M20-E080_399.png | Using a segmentation mask (polygon), locate all instances of Building | [690, 265, 402, 0, 943, 0, 690, 265] | 50 |
76-2019-0520-6955-LA93-0M20-E080_557_5 | BDOrtho_76-2019-0520-6955-LA93-0M20-E080_557.png | Does the picture contain a Road? (yes/no) | yes | 13 |
69-2020-0840-6525-LA93-0M20-E080_824_1 | BDOrtho_69-2020-0840-6525-LA93-0M20-E080_824.png | Return a segmentation mask (polygon) wherever Building appears in the image | [670, 155, 644, 94, 671, 82, 701, 141, 670, 155], [896, 592, 946, 504, 1000, 443, 1000, 629, 896, 592] | 115 |
76-2019-0605-6945-LA93-0M20-E080_301_9 | BDOrtho_76-2019-0605-6945-LA93-0M20-E080_301.png | Provide the appropriate class for what appears in: [740, 451, 1000, 696]. Choose the label that fits best from these choices: ['Ski lift', 'Service track', 'Building', 'Heliport', 'Swimming pool', 'Penstock', 'Sports track', 'Marsh', 'Tramway', 'Waste disposal site', 'Castle', 'Motorway type', 'Aquaculture', 'Water body', 'Subway station', 'Barracks', 'Port', 'Path'] | Building | 130 |
78-2021-0635-6855-LA93-0M20-E080_311_2 | BDOrtho_78-2021-0635-6855-LA93-0M20-E080_311.png | Highlight the full extent of each Large sports field using a detailed segmentation polygon | [86, 607, 0, 869, 0, 579, 86, 607] | 51 |
31-2022-0495-6200-LA93-0M20-E080_610_7 | BDOrtho_31-2022-0495-6200-LA93-0M20-E080_610.png | Is there a Forest in the image? (yes/no) | yes | 14 |
34-2021-0740-6260-LA93-0M20-E080_396_6 | BDOrtho_34-2021-0740-6260-LA93-0M20-E080_396.png | Assign an appropriate category to the item visible in: [915, 321, 1000, 472] | Building | 33 |
21-2020-0845-6695-LA93-0M20-E080_1111_4 | BDOrtho_21-2020-0845-6695-LA93-0M20-E080_1111.png | Segment every Drinking water production plant in the frame using a polygon mask | [] | 16 |
76-2019-0570-6970-LA93-0M20-E080_576_5 | BDOrtho_76-2019-0570-6970-LA93-0M20-E080_576.png | Assign a label to the item located in: [247, 722, 244, 618, 321, 615, 324, 722, 247, 722] | Building | 61 |
67-2021-1045-6835-LA93-0M20-E080_658_0 | BDOrtho_67-2021-1045-6835-LA93-0M20-E080_658.png | Use bounding boxes to locate and identify Building throughout the image | [28, 0, 135, 79] | 28 |
45-2020-0660-6760-LA93-0M20-E080_833_5 | BDOrtho_45-2020-0660-6760-LA93-0M20-E080_833.png | Does this image feature a Castle? (yes/no) | yes | 13 |
59-2021-0655-7110-LA93-0M20-E080_443_1 | BDOrtho_59-2021-0655-7110-LA93-0M20-E080_443.png | Give the correct label for the object pictured in: [851, 338, 849, 358, 999, 372, 1000, 352, 851, 338]. Use one of the following labels: ['Port gantry', 'Intersection', 'Barracks', 'Lift system', 'Park house', 'Factory', 'Racetrack', 'Passenger stop', 'Zoo', 'Arena or ancient theater', 'Grass runway', 'Marsh', 'Path', 'Parking lot', 'Dam reservoir', 'Military structure', 'Stadium', 'Vineyard'] | Path | 164 |
33-2021-0425-6440-LA93-0M20-E080_480_4 | BDOrtho_33-2021-0425-6440-LA93-0M20-E080_480.png | Provide the appropriate class for what appears in: [539, 1000, 787, 857, 865, 1000, 539, 1000] | Castle | 54 |
59-2021-0695-7060-LA93-0M20-E080_687_16 | BDOrtho_59-2021-0695-7060-LA93-0M20-E080_687.png | Give a name to the item shown in: [754, 54, 699, 112, 650, 67, 685, 30, 653, 0, 693, 0, 754, 54]. Use one of the following labels: ['Penstock', 'Park house', 'Forest', 'Drinking water production plant', 'Path', 'Toll booth', 'Air traffic control tower', 'Church', 'Ecomuseum', 'Chapel', 'Building', 'Indoor sports complex', 'Heliport', 'Refuge', 'Sports track', 'Vineyard', 'Silo', 'Monument'] | Building | 167 |
13-2020-0845-6295-LA93-0M20-E080_40_2 | BDOrtho_13-2020-0845-6295-LA93-0M20-E080_40.png | Determine the category of the object found in: [683, 416, 728, 464]. Select the label from: ['Ecomuseum', 'Port', 'Fire station', 'Watercourse', 'Monument', 'Chairlift', 'Small multi-sports field', 'Water body', 'Bridge', 'Indoor sports complex', 'Sand', 'Rest or service area', 'Path', 'Pond', 'Lock', 'Waste disposal site', 'Arena or ancient theater', 'Shooting stand'] | Bridge | 130 |
GroundSet: A Cadastral-Grounded Dataset for Spatial Understanding with Vector Data
GroundSet is a large-scale Earth Observation dataset grounded in verifiable cadastral vector data, designed to bridge the gap in fine-grained spatial understanding for modern Multimodal Models.
The dataset is built upon high-resolution (20 cm) optical aerial orthophotos and legally verified vector data provided by the French national mapping agency (IGN). It features high semantic richness and geometric precision, enabling robust model training for complex geospatial tasks.
๐ Key Statistics
- Pretraining Scale: 3.8 million annotated objects across 510k high-resolution images.
- Finetuning Scale: 880k objects across 60k images, with 1.8M instruction queries.
- Semantic Granularity: 135 highly specific semantic categories (e.g., power plants, heritage sites, crops, etc.).
- Supported Tasks: Scene captioning, localized classification, object detection, multi-class detection, referring expression comprehension (REC), segmentation and Visual Question Answering (VQA).
GroundSet overview showing finetuning tasks
๐๏ธ Dataset Structure
The repository is organized into two primary components: a pretraining dataset featuring raw geometric annotations (including bounding boxes, lines and polygons), and a supervised fine-tuning (SFT) dataset specifically tailored for Multimodal Large Language Models (MLLMs) using an instruction-based format. For efficiency, all image and metadata pairs are stored in Parquet files.
GroundSet/
โ
โโโ pretraining/ <-- (Parquet files containing images and raw JSON annotations)
โโโ finetuning/ <-- (Parquet files containing images and raw JSON annotations)
โโโ instructions/
โโโ train/ <-- (Single JSON file with training instructions)
โโโ test/ <-- (JSONL files with testing instructions)
1. Pretraining Split
The pretraining dataset contains the full-scale release. It comprises 510,483 patches containing 3,829,755 objects across 135 unique categories. This release is intended to support broader research beyond the standard MLLM instruction tuning paradigm.
โ ๏ธ Warning: To prevent data leakage, the pretraining dataset does not contain any of the samples used in the finetuning subset.
2. Finetuning Split
This subset is tailored specifically for supervised fine-tuning and contains the visual data (60k images) and raw annotations corresponding to the instruction sets. We partition the visual data of this finetuning dataset into 14,000 images for testing and 45,988 images for training.
๐ก Note: This split houses the actual image files referenced by the Q&A pairs in the Instructions subset.
3. Instructions
The instructions subset contains the actual textual question-answer pairs required to train and evaluate vision-language models on the images from the Finetuning split:
- Train: A single JSON file containing a total of 1,845,076 instructions used to train our baseline.
- Test: JSONL files containing 72,597 evaluation instructions.
๐พ Data Format
The dataset uses Parquet files to bundle the images and their corresponding metadata. Each row in the Parquet files contains the following fields:
image: The PIL Image object (decoded from the raw bytes). The images are patches with a size of 672x672 pixels, corresponding to a spatial extent of approximately 134x134 meters.file_name: The original filename of the aerial patch.json: A stringified JSON payload containing the raw metadata and geometric annotations. Annotations can include Horizontal Bounding Boxes (HBB), Oriented Bounding Boxes (OBB) and polygonal segmentation masks.
The instruction files format the question-answer pairs and are stored directly in standard JSON or JSONL format.
๐ Qualitative Samples
The cadastral vector data provides exact boundaries for diverse and highly specific infrastructure across varied environments (urban, rural, alpine, maritime).
Qualitative samples from the GroundSet dataset showing semantic and geometric annotations
๐ป Usage Example
You can easily load and parse the dataset using the Hugging Face datasets library. Because the json column is stored as a string, you can parse it into a Python dictionary using the standard json library.
from datasets import load_dataset
import json
# 1. Load the finetuning dataset split
dataset = load_dataset("RogerFerrod/GroundSet", data_dir="finetuning", split="train")
# 2. Inspect a single sample
sample = dataset[0]
# The image is immediately accessible as a PIL object
image = sample["image"]
file_name = sample["file_name"]
# 3. Parse the JSON string into a dictionary
annotations = json.loads(sample["json"])
print(f"File: {file_name}")
print(f"Raw Annotations: {annotations}")
# --- Loading Instructions ---
# You can load the instructions similarly:
train_instructions = load_dataset("RogerFerrod/GroundSet", data_dir="instructions/train", split="train")
print(train_instructions[0])
๐ Citation
If you utilize this dataset in your research, please consider citing the original work:
@article{groundset,
title={GroundSet: A Cadastral-Grounded Dataset for Spatial Understanding with Vector Data},
author={Ferrod, Roger and Lecene, Ma{\"e}l and Sapkota, Krishna and Leifman, George and Silverman, Vered and Beryozkin, Genady and Lobry, Sylvain},
journal={arXiv preprint},
year={2026}
}
๐ Acknowledgements
This work was supported by Google under a research collaboration agreement with Universitรฉ Paris Citรฉ. The underlying GroundSet dataset leverages official data from IGN (French National Institute of Geographic and Forest Information), specifically BD ORTHOยฎ and BD TOPOยฎ, released under Open Licence 2.0.
- Downloads last month
- 45