isp-uv-es/bigearthnet-v2-reference-maps-taco

BigEarthNet v2 CORINE reference maps

This is a repackaging, not a new dataset. It is BigEarthNet v2 CORINE reference maps by reBEN / BigEarthNet authors (TU Berlin), Copernicus CORINE Land Cover, converted to TACO. Pixel values and labels are kept as released except where the description below says otherwise. All credit belongs to the original authors: if you use it, please cite them and follow their licence.

original dataset · paper · licence: CDLA-Permissive-1.0

Repackaged into TACO by the Image and Signal Processing Group (ISP), Universitat de València, within the ELLIOT project.

Citation

Please cite the original work:

@inproceedings{clasen2024reben,
  title     = {{reBEN}: Refined {BigEarthNet} Dataset for Remote Sensing Image Analysis},
  author    = {Clasen, Kai Norman and Hackel, Leonard and Burgert, Tom and Sumbul, Gencer and Demir, Beg{\"u}m and Markl, Volker},
  booktitle = {IGARSS 2025 - 2025 IEEE International Geoscience and Remote Sensing Symposium},
  pages     = {1264--1268},
  year      = {2025},
  doi       = {10.1109/IGARSS55030.2025.11242834}
}

About the data

Per-pixel CORINE Land Cover for all 549488 BigEarthNet v2 patches at 120x120: the segmentation label the release ships in a separate archive.

549,488 samples · splits: test 137,367 · train 272,544 · validation 139,577 · tasks: semantic-segmentation, classification

Packaged as TACO v3.

Full description

Label layer. No imagery is stored, because the Sentinel-1 and Sentinel-2 pixels for these patches are already in bigearthnet-v2-txt; join on patch_id. That container covers 464044 of these patches, the BigEarthNet.txt-annotated subset, so 85444 patches are here and not there, and in_bigearthnet_v2_txt says which per row.

Classes. The mask indexes the 44 raw CLC level-3 codes present in the corpus, plus a 45th entry no_reference_map that is the ignore index, rather than the release's 19 classes, so the twelve codes the 19-class nomenclature drops survive: roads, ports, airports, quarries, dumps, construction, green urban, sport/leisure, bare rock, burnt areas, intertidal flats and unclassified. clc_to_ben19 collapses them when wanted.

• That mapping is not published by the release. It is derived from it by constraint satisfaction against the per-patch 19-class labels, then validated on 30,000 held-out patches with 30,000 exact matches and zero mismatches.
• The 19-class multi-hot is recomputed per sample and asserted equal to the release's own label list.

Getting started

git clone https://github.com/OscarPellicer/taco
pip install -e "taco/python[ml]"     # builds the native reader: needs CMake and Ninja

Read it straight from the Hub:

from huggingface_hub import snapshot_download
from taco.ml import Dataset, plot_sample

path = snapshot_download("isp-uv-es/bigearthnet-v2-reference-maps-taco", repo_type="dataset")
ds = Dataset(path)                  # the repository is the container: parts + .tacocat
plot_sample(ds[0])

or from a local copy:

ds = Dataset("bigearthnet-v2-reference-maps")
sample = ds[0]                      # {slot name: SlotValue}, arrays decoded
sample["patch_id"].array.shape

Metadata without decoding anything:

import taco
taco.read("bigearthnet-v2-reference-maps/.tacocat")              # one Arrow table, levels joined

Samples

What a sample contains

role	slot	holds	modality	detail
input	patch_id	text	text
target	land_cover	mask	label_raster	45 classes
target	labels_19	class_multihot		19 classes

Licence

CDLA-Permissive-1.0

Terms of use:

• The CDLA-Permissive-1.0 licence text must accompany redistributed copies, and recipients are bound by it.
• Existing attribution notices must be kept, and modified files must be marked as modified (CDLA-Permissive-1.0).
• Credit the original authors and the data sources listed below.

Required credits:

• BigEarthNet v2.0 (reBEN), Clasen et al., Remote Sensing Image Analysis Group, Technische Universität Berlin
• CORINE Land Cover: © European Union, Copernicus Land Monitoring Service 2018, European Environment Agency (EEA)

Providers: reBEN / BigEarthNet authors (TU Berlin), Copernicus CORINE Land Cover

Acknowledgements

TACO was designed by César Aybar and is specified at https://asterisk.coop/taco/spec/.

Built by Oscar Pellicer within the ELLIOT project at the Image and Signal Processing Group (ISP), Universitat de València.