{"success":true,"data":{"author":"Weikmann, Paris, Bruzzone (University of Trento); TACO repackaging by ISP, Universitat de Valencia","bucketAccessType":"public","bucketName":"isp-uv-es/timesen2crop-taco","bucketPrefix":"","category":"Agriculture & Environment","cloudProvider":"huggingface","createdAt":"2026-11-16T14:11:25.567Z","description":"TimeSen2Crop (IEEE JSTARS, 2021) is a pixel-level crop type classification benchmark of 1,135,511 labelled Sentinel-2 pixel time series over Austria, drawn from 15 Sentinel-2 tiles that cover the country's agro-ecological zones. Labels come from Austrian agricultural parcel declarations for the agronomic year starting September 2017 and span 15 crop classes, with per-class counts checked against the release's description PDF. Each series has nine bands (B2, B3, B4, B5, B6, B7, B8A, B11, B12) gap-filled to a daily 365-step cadence. In this TACO copy one sample is one tile, with pixels packed into the raster grid, so a leaf must be reshaped to time by band by pixel rather than treated as an image. The release's five cross-validation folds are tile partitions and are carried as a test_fold field; legend and band names were checked against the release's PDF. About 11.5 GB, read with tacoreader. Original on Zenodo under CC BY 4.0; cite Weikmann et al. 2021.","displayTitle":"TimeSen2Crop: One Million Labelled Sentinel-2 Pixel Time Series for Crop Type Classification, Austria (TACO)","doi":"10.1109/JSTARS.2021.3073965","id":"f9761107-4d54-5d7f-9fc5-af6b770682c1","license":"CC-BY-4.0","licenseEvidenceUrl":"https://zenodo.org/records/4715631","objectCountEstimate":9,"probeStatus":"ok","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"isp-uv-es/timesen2crop-taco\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"isp-uv-es/timesen2crop-taco\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=isp-uv-es, license:cc-by-4.0, size_categories:n<1K, downloads=68"],"format":"mixed","generatedAt":"2026-11-16T14:11:30.952Z","layout":{"example":"assets/logos.png","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"89504e470d0a1a0a0000000d49484452000002220000008c0806000000413dc29d00009fbb49444154789cecfd77986c5775e78d7fd7dafb84ca9d6e0e0a57010524a1608288064cce2667f03018637bb0fd9b796dbf1ef0cc388dc7f1c7181b838d8d6dd2788c49c6649011c2422040a09ca51bfa76aaae74c2de6bbd7fecaa9baf7415407dafcee779eae9eeea3aa74eed3a55fb7bd65eebbb808a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8a8ac32002981944742fffff311f58454545454545c589cda1e2e34041424460e67b7c7c45454545454545c53d32110f3c1616fbfe1e8b8c288a70f1a3ce6da449b26f1b73800089e318173feadc461c47076d37112acc74d0f35454545454543c5ca866be7b8188a0aab0c6c0797fd8ffb76fdd6c5ff7ca175e7afa8e931e79f7ce3db77cfd9b575df5f52bbe3dbfd2ed69a35ec3a58fb970ee89975e72d1f66d9b4ebbe996db7ff8c10f7fe2df6ebdedaef2d0fd1863e0bddff77c1515151515150f072a2132e6a068842a1480310cef05975c785ef3a79fffcc9fbce5b63b6eb9f196db77de78cb6dabbbf72cb8c7fdc485532f7ff1339f17c751321c64fd24896b499ad4f7ec5db8637ecfddf3b3731b66b76cda78525114f928cf87f57ad274ce97fff8cf5ff8f417bf72f9c2bab9693e","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: \"TimeSen2Crop\"\nsize_categories:\n- n<1K\ntask_categories:\n- image-classification\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# TimeSen2Crop\n\n> **This is a repackaging, not a new dataset.** It is **TimeSen2Crop** by University of Trento (Weikmann, Paris, Bruzzone), 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/4715631) · [paper](https://doi.org/10.1109/JSTARS.2021.3073965) · 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@article{weikmann2021timesen2crop,\n  title   = {TimeSen2Crop: A Million Labeled Samples Dataset of Sentinel 2 Image Time Series for Crop-Type Classification},\n  author  = {Weikmann, Giulio and Paris, Claudia and Bruzzone, Lorenzo},\n  journal = {IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing},\n  volume  = {14},\n  pages   = {4699--4708},\n  year    = {2021},\n  doi     = {10.1109/JSTARS.2021.3073965}\n}\n```\n\n## About the data\n\n1,135,511 independent Sentinel-2 pixels, each a 365-day 9-band series, over 15 tiles covering Austria, 15 crop classes and one agronomic year from September 2017 interpolated to a daily cadence. One sample is one Sentinel-2 tile. What it adds is a gap-filled full-year daily series over a million labelled examples.\n\n**15 samples** · splits: `test` 3 · `train` 9 · `validation` 3 · tasks: `classification`\n\nPackaged as [TACO v3](https://asterisk.coop/taco/spec/).\n\n<details>\n<summary>Full description</summary>\n\n**The rasters are not images.** A leaf's height and width pack pixels rather than describe ground: neighbouring cells are unrelated places. Read a leaf as reshape(T, C, -1)[..., :n_pixels]; never crop, resample or convolve it.\n\n**Legend and bands.** Both come from the release's description PDF rather than the MONSTER benchmark copy this is converted from, which ships integers and no names, and both are verified against the pixels: the fifteen per-class counts match the PDF's table exactly. Bands are B2, B3, B4, B5, B6, B7, B8A, B11, B12 -- B8A rather than B8.\n\n**Tiles and folds.** Tile identities were recovered from the PDF the same way; index 0 merges the release's two 33UVP folders, the 2017/18 and 2019 agronomic years. The release's five cross-validation folds are exactly tile partitions, so `test_fold` is a property of the sample and no index list has to travel with it.\n\n**Georeferencing.** That copy publishes no pixel coordinates and no acquisition calendar, so no CRS and no per-frame dates 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\nimport os\nfrom huggingface_hub import hf_hub_download, snapshot_download\nfrom taco.ml import Dataset, plot_sample\n\npath = hf_hub_download(\"isp-uv-es/timesen2crop-taco\", \"timesen2crop.zip\", repo_type=\"dataset\")\nds = Dataset(path)\nplot_sample(ds[0])\n```\n\nor from a local copy:\n\n```python\nds = Dataset(\"timesen2crop.zip\")\nsample = ds[0]                      # {slot name: SlotValue}, arrays decoded\nsample[\"s2\"].array.shape\n```\n\nMetadata without decoding anything:\n\n```python\nimport taco\ntaco.read(\"timesen2crop.zip\")              # 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| rol","encoding":"utf8","key":"README.md","name":"README.md","size":4821,"truncated":true}],"sourceObjectKey":"assets/logos.png","starterCell":"from huggingface_hub import hf_hub_download\npath = hf_hub_download(repo_id=\"isp-uv-es/timesen2crop-taco\", filename=\"assets/logos.png\", repo_type=\"dataset\")\nprint(path)","starterSamplePath":"assets/logos.png","totalObjects":9},"summary":"TimeSen2Crop (IEEE JSTARS, 2021) is a pixel-level crop type classification benchmark of 1,135,511 labelled Sentinel-2 pixel time series over Austria, drawn from 15 Sentinel-2…","title":"TimeSen2Crop: One Million Labelled Sentinel-2 Pixel Time Series for Crop Type Classification, Austria (TACO)","updatedAt":"2026-11-16T16:57:46.306Z","visibility":"public","websiteUrl":"https://huggingface.co/datasets/isp-uv-es/timesen2crop-taco"}}