isp-uv-es/s2-100k-taco

S2-100K (SatCLIP pretraining corpus)

This is a repackaging, not a new dataset. It is S2-100K (SatCLIP pretraining corpus) by Klemmer et al. (SatCLIP), ESA Copernicus / Sentinel-2, 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: MIT

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{klemmer2025satclip,
  title     = {SatCLIP: Global, General-Purpose Location Embeddings with Satellite Imagery},
  author    = {Klemmer, Konstantin and Rolf, Esther and Robinson, Caleb and Mackey, Lester and Ru{\ss}wurm, Marc},
  booktitle = {Proceedings of the AAAI Conference on Artificial Intelligence},
  volume    = {39},
  number    = {4},
  pages     = {4347--4355},
  year      = {2025},
  doi       = {10.1609/aaai.v39i4.32457}
}

About the data

100000 Sentinel-2 L2A patches of 256x256 px at 10 m (12 bands, reflectance x 1e4), sampled uniformly at random from every Planetary Computer scene between 2021-01-01 and 2023-05-17 that intersects land and is under 20% cloud. The sample is globally uniform rather than drawn over a chosen region.

100,000 samples · splits: test 5,244 · train 89,913 · validation 4,843 · tasks: similarity-search

Packaged as TACO v3.

Full description

Target. None: the corpus is unlabelled. Each patch carries its longitude and latitude, which is what SatCLIP trains an image encoder against.

Splits. The release publishes none. The split here is 1-degree spatial blocks by hash (train 89913, val 4843, test 5244); the intended use is to train on all of it.

Getting started

git clone --recursive https://github.com/OscarPellicer/taco
pip install -e "taco/python[ml]"     # builds the reader: C++23, CMake, Ninja, pkg-config, libcurl >= 7.83, OpenSSL >= 3

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/s2-100k-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("s2-100k")
sample = ds[0]                      # {slot name: SlotValue}, arrays decoded
sample["s2"].array.shape

Metadata without decoding anything:

import taco
taco.read("s2-100k/.tacocat")              # one Arrow table, levels joined

Samples

What a sample contains

role	slot	holds	modality	detail
input	s2	raster	multispectral	12 band(s), unit 1, scaled

Licence

MIT

Terms of use:

• Keep the MIT licence notice with any copy.

Required credits:

• S2-100K: Klemmer et al. (SatCLIP)
• Contains modified Copernicus Sentinel data

Providers: Klemmer et al. (SatCLIP), ESA Copernicus / Sentinel-2

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.