{"success":true,"data":{"author":"Kossmann, Wilhelm et al. (TU Dortmund University); TACO repackaging by ISP, Universitat de Valencia","bucketAccessType":"public","bucketName":"isp-uv-es/seasonet-sampled-taco","bucketPrefix":"","category":"Earth Observation & Geospatial","cloudProvider":"huggingface","createdAt":"2026-11-16T14:11:25.529Z","description":"SeasoNet (IGARSS 2022) is a seasonal scene classification, segmentation and retrieval dataset of Sentinel-2 patches over Germany labelled with the 33-class CORINE-based Digital Land Cover Model LBM-DE2018 of the Federal Agency for Cartography and Geodesy (BKG). The full release has 1,759,830 patches of 120x120 pixels at 10 m. This repository is a location-based subsample: one in ten of the 90,832 locations that have all four seasons, chosen by a hash of the coordinates, giving 36,936 patches at 9,142 locations imaged in spring, summer, autumn and winter plus a snow acquisition where one exists. The 12 Sentinel-2 bands are kept on their native 10, 20 and 60 m grids, together with the release's cloud, snow and label-reliability masks. The official splits are kept (25,973 train, 3,703 validation, 7,260 test) and no location crosses splits. About 4.5 GB in TACO format, read with tacoreader. Original on Zenodo under CC BY 4.0; cite Kossmann et al. 2022.","displayTitle":"SeasoNet (location sample): Seasonal Sentinel-2 Land Cover Segmentation over Germany (TACO)","doi":"10.1109/IGARSS46834.2022.9884079","id":"e7dff03d-c393-5c3f-8e24-6edfb278f373","license":"CC-BY-4.0","licenseEvidenceUrl":"https://zenodo.org/records/5850307","objectCountEstimate":12,"probeStatus":"ok","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"isp-uv-es/seasonet-sampled-taco\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"isp-uv-es/seasonet-sampled-taco\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=isp-uv-es, license:cc-by-4.0, size_categories:10K<n<100K, downloads=523"],"format":"mixed","generatedAt":"2026-11-16T14:11:31.988Z","layout":{"example":"assets/logos.png","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"assets/logos.png","truncatedAt":512}],"sampleObjects":["assets/logos.png","assets/sample_0.png","assets/sample_1.png","assets/sample_2.png","assets/sample_3.png"],"sidecars":[{"bytes":"---\nlicense: cc-by-4.0\npretty_name: \"SeasoNet (5% location sample)\"\nsize_categories:\n- 10K<n<100K\ntask_categories:\n- image-segmentation\ntags:\n- earth-observation\n- remote-sensing\n- taco\n---\n\n<p align=\"center\"><img src=\"assets/logos.png\" width=\"546\" alt=\"Image and Signal Processing Group, Universitat de València · ELLIOT\"></p>\n\n# SeasoNet (5% location sample)\n\n> **This is a repackaging, not a new dataset.** It is **SeasoNet (5% location sample)** by TU Dortmund University (Koßmann et al.), BKG (LBM-DE2018), 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.\n>\n> [original dataset](https://zenodo.org/records/5850307) · [paper](https://doi.org/10.1109/IGARSS46834.2022.9884079) · licence: CC-BY-4.0\n>\n> Repackaged into TACO by the [Image and Signal Processing Group (ISP)](https://isp.uv.es/), Universitat de València, within the [ELLIOT](https://elliot-ai.eu) project.\n\n## Citation\n\nPlease cite the original work:\n\n```bibtex\n@inproceedings{kossmann2022seasonet,\n  title     = {{SeasoNet}: A Seasonal Scene Classification, Segmentation and Retrieval Dataset for Satellite Imagery over {Germany}},\n  author    = {Ko{\\ss}mann, Dominik and Brack, Viktor and Wilhelm, Thorsten},\n  booktitle = {IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium},\n  pages     = {243--246},\n  year      = {2022},\n  doi       = {10.1109/IGARSS46834.2022.9884079}\n}\n```\n\n## About the data\n\n36936 Sentinel-2 patches at 9142 locations in Germany, each with a 33-class CORINE land-cover map and each location imaged in spring, summer, fall and winter, plus a snow acquisition where one exists.\n\n**36,936 samples** · splits: `test` 7,260 · `train` 25,973 · `validation` 3,703 · tasks: `semantic-segmentation`\n\nPackaged as [TACO v3](https://asterisk.coop/taco/spec/).\n\n<details>\n<summary>Full description</summary>\n\n**Sampling.** A subset of the 1.76 M-patch release. Locations, not patches, are sampled: 1 in 10 of the 90832 locations that have all four seasons, chosen by a hash of the coordinate. Sampling by location is what preserves the seasonal axis.\n\n**Splits.** No location spans more than one official split, so the split here is the release's own: train 25973, val 3703, test 7260.\n\n**Bands.** The 12 bands are kept on their three native grids, and the release's cloud, snow and difficulty masks are carried.\n\n</details>\n\n## Getting started\n\n```bash\ngit clone --recursive https://github.com/OscarPellicer/taco\npip install -e \"taco/python[ml]\"     # builds the reader: C++23, CMake, Ninja, pkg-config, libcurl >= 7.83, OpenSSL >= 3\n```\n\nRead it straight from the Hub:\n\n```python\nfrom huggingface_hub import snapshot_download\nfrom taco.ml import Dataset, plot_sample\n\npath = snapshot_download(\"isp-uv-es/seasonet-sampled-taco\", repo_type=\"dataset\")\nds = Dataset(path)                  # the repository is the container: parts + .tacocat\nplot_sample(ds[0])\n```\n\nor from a local copy:\n\n```python\nds = Dataset(\"seasonet-sampled\")\nsample = ds[0]                      # {slot name: SlotValue}, arrays decoded\nsample[\"s2_10m\"].array.shape\n```\n\nMetadata without decoding anything:\n\n```python\nimport taco\ntaco.read(\"seasonet-sampled/.tacocat\")              # one Arrow table, levels joined\n```\n\n## Samples\n\n![sample](assets/sample_0.png)\n![sample](assets/sample_1.png)\n![sample](assets/sample_2.png)\n![sample](assets/sample_3.png)\n![sample](assets/sample_4.png)\n\n## What a sample contains\n\n| role | slot | holds | modality | detail |\n|---|---|---|---|---|\n| input | `s2_10m` | raster | optical | 4 band(s), unit 1, scaled |\n| input | `s2_20m` | raster | multispectral | 6 band(s), unit 1, scaled |\n| input | `s2_60m` | raster | optical | 2 band(s), unit 1, scaled |\n| input | `clouds` | mask | label_raster | 2 classes |\n| input | `snow` | mask | label_raster | 2 classes |\n| input | `reliable_easy` | mask | label_raster | 2 classes |\n| input | `reliable_medium` | mask | label_raster | ","encoding":"utf8","key":"README.md","name":"README.md","size":4711,"truncated":true}],"sourceObjectKey":"assets/logos.png","starterCell":"from huggingface_hub import hf_hub_download\npath = hf_hub_download(repo_id=\"isp-uv-es/seasonet-sampled-taco\", filename=\"assets/logos.png\", repo_type=\"dataset\")\nprint(path)","starterSamplePath":"assets/logos.png","totalObjects":12},"summary":"SeasoNet (IGARSS 2022) is a seasonal scene classification, segmentation and retrieval dataset of Sentinel-2 patches over Germany labelled with the 33-class CORINE-based Digital…","title":"SeasoNet (location sample): Seasonal Sentinel-2 Land Cover Segmentation over Germany (TACO)","updatedAt":"2026-11-16T16:59:45.866Z","visibility":"public","websiteUrl":"https://huggingface.co/datasets/isp-uv-es/seasonet-sampled-taco"}}