Datasets:

fieldsoftheworld
/

ftw-q

Modalities:

Geospatial

Tags:

License:

Dataset card Files Files and versions

xet

Community

Dataset Viewer

The dataset viewer is not available for this subset.

Cannot get the split names for the config 'default' of the dataset.

Exception:    SplitsNotFoundError
Message:      The split names could not be parsed from the dataset config.
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 286, in get_dataset_config_info
                  for split_generator in builder._split_generators(
                                         ~~~~~~~~~~~~~~~~~~~~~~~~~^
                      StreamingDownloadManager(base_path=builder.base_path, download_config=download_config)
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/webdataset/webdataset.py", line 80, in _split_generators
                  raise ValueError(
                  ...<2 lines>...
                  )
              ValueError: The TAR archives of the dataset should be in WebDataset format, but the files in the archive don't share the same prefix or the same types.
              
              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/split_names.py", line 68, in compute_split_names_from_streaming_response
                  for split in get_dataset_split_names(
                               ~~~~~~~~~~~~~~~~~~~~~~~^
                      path=dataset,
                      ^^^^^^^^^^^^^
                      config_name=config,
                      ^^^^^^^^^^^^^^^^^^^
                      token=hf_token,
                      ^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 340, in get_dataset_split_names
                  info = get_dataset_config_info(
                      path,
                  ...<6 lines>...
                      **config_kwargs,
                  )
                File "/usr/local/lib/python3.14/site-packages/datasets/inspect.py", line 291, in get_dataset_config_info
                  raise SplitsNotFoundError("The split names could not be parsed from the dataset config.") from err
              datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.

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.

FTW-Q

The Fields of The World (FTW) benchmark rebuilt on one year of quarterly Sentinel-2 cloudless mosaics: the same chips, splits and field polygons as FTW, with 16-band imagery (four quarters × four bands) in place of FTW's two Sentinel-2 scenes. 70,817 chips in 25 countries.

Files

One {country}.tar.gz per country (25 archives, 23 GB in total), unpacking to:

{country}/{aoi_id}.tif         imagery
{country}/{aoi_id}_label.tif   labels
{country}/chips.parquet        aoi_id, split (train / val / test / none), year

index.parquet (GeoParquet, EPSG:4326) has one row per chip across all countries: country, aoi_id, split, year, archive and the image and label paths inside it, the chip's grid (epsg, GDAL transform, width, height), the mosaic items it was read from (q1_item to q4_item, e.g. Sentinel-2_mosaic_2021_Q1_33UUP_0_0) and their source_tile, label_license, bbox, and the chip footprint as geometry.

Imagery

Source: Copernicus Sentinel-2 Global Mosaics (quarterly first-quartile cloudless composites), read from the public mirror at s3://tge-labs/sentinel-2-quarterly-cloudless-mosaics/ (endpoint data.source.coop).
Bands: 16, ordered [B04, B03, B02, B08] for each of Q1, Q2, Q3, Q4 (band 4 * q + b).
Values: int16 surface reflectance × 10000; −32768 where a quarter has no valid observation.
Grid: 10 m, in the UTM projection of the mosaic tile, cropped to the FTW chip footprint.
Year: one per country (below), taken as the year of FTW's labels, or the nearest year with mosaics.

Labels

uint8 at 2.5 m (4× the imagery grid, same extent), rasterized from the FTW field polygons: 0 background, 1 field, 2 field boundary (polygon outlines buffered by 7 m).

How it was built

For each country and year, search the CDSE STAC API (https://stac.dataspace.copernicus.eu/v1/search, collection sentinel-2-global-mosaics) over the country's bounding box, per quarter. Assign each FTW chip, per quarter, the first returned mosaic tile whose bounding box contains the chip's centre.
Read the chip's footprint from each of the 16 band COGs (Q1–Q4 × B04, B03, B02, B08) on the tile's native grid and stack them.
Rasterize the FTW polygons intersecting the chip onto the 2.5 m grid.

Country	Year	License (FTW source labels)
Austria	2021	CC-BY-4.0
Belgium	2021	No restrictions on public access
Brazil	2020	CC-BY-4.0
Cambodia	2021	CC-BY-4.0
Corsica	2021	CC-BY-2.0
Croatia	2023	Open Data
Denmark	2021	CC0-1.0
Estonia	2021	CC-3.0
Finland	2021	CC-BY-4.0
France	2020	Open Licence
Germany	2019	DL-DE/BY-2-0
India	2017	CC-BY-4.0
Kenya	2022	GPL-2.0-or-later
Latvia	2021	CC-BY-NC-4.0
Lithuania	2021	Non-commercial use only
Luxembourg	2022	CC0-1.0
Netherlands	2022	CC0-1.0
Portugal	2021	CC-BY-NC-4.0
Rwanda	2021	CC-BY-4.0
Slovakia	2021	CC0-1.0
Slovenia	2021	CC-BY-4.0
South Africa	2018	CC-BY-NC-SA-4.0
Spain	2020	CC-BY-4.0
Sweden	2021	No restrictions on public access
Vietnam	2021	CC-BY-4.0

License

Each country's labels keep the license of their FTW source (table above). The imagery is Copernicus Sentinel data, modified, under the Legal Notice on the Use of Copernicus Sentinel Data and Service Information.

Downloads last month: 178

Total file size:

23.8 GB

Models trained or fine-tuned on fieldsoftheworld/ftw-q

fieldsoftheworld/ftw-q

Image Segmentation • Updated 20 days ago