{"success":true,"data":{"author":"Cesar Aybar et al., Image and Signal Processing group, University of Valencia (TACO Foundation)","bucketAccessType":"public","bucketName":"tacofoundation/cloudsen12","bucketPrefix":"","category":"Earth Observation & Geospatial","cloudProvider":"huggingface","createdAt":"2026-11-14T14:15:15.795Z","description":"CloudSEN12+ is the largest dataset of expert-labelled pixels for cloud and cloud shadow semantic segmentation in Sentinel-2 imagery, extending CloudSEN12 with twice as many reviewed labels and refined earlier annotations. Labels classify clear, thick cloud, thin cloud and cloud shadow, with high-quality, scribble and unlabelled patches at 509 x 509 and 2000 x 2000 pixels (padded to 512 and 2048). Version 1.1 provides three modes: cloudsen12-l1c (Level-1C), cloudsen12-l2a (Level-2A from Google Earth Engine) and cloudsen12-extra with Copernicus DEM elevation, ESA WorldCover, Sentinel-1 VV/VH backscatter and cloud masks from QA60, Sen2Cor, s2cloudless, Cloud Score+, UnetMobV2 and SEnSeIv2 for comparison. The listing (measured 2026-11-14) shows 15 .part.taco files of up to 20 GB each, about 248 GB in total, in the TACO format; load them with tacoreader (tacoreader.load(\"tacofoundation:cloudsen12-l1c\")) and read rasters with rasterio. GeoTIFFs are also on ScienceDataBank. The publishers release it under CC0, placing it in the public domain; cite Aybar et al. 2024 (Data in Brief).","displayTitle":"CloudSEN12+ v1.1: Expert-Labelled Cloud and Cloud Shadow Detection for Sentinel-2","id":"fa6a445b-beb9-5f41-894e-e9c134906448","license":"CC0-1.0","licenseEvidenceUrl":"https://huggingface.co/datasets/tacofoundation/cloudsen12","objectCountEstimate":16,"probeStatus":"ok","referenceUrl":"https://www.sciencedirect.com/science/article/pii/S2352340924008163","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"tacofoundation/cloudsen12\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"tacofoundation/cloudsen12\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=tacofoundation, license:cc0-1.0, downloads=9219"],"format":"mixed","generatedAt":"2026-11-14T14:15:21.120Z","layout":{"example":"cloudsen12-extra.0000.part.taco","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"5758a4b60ea804000000b7061a000000000003000000000000005bbd28a8040000008b3300000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000002379386c0a0000000000b520000000000000010000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000049492a00080000001100000103000100000000020000010103000100000000020000020103000100000010000000030103000100000050c300000601030001000000010000001501030001000000010000001c01030001000000020000003d0103000100000002000000420103000100","encoding":"hex","key":"cloudsen12-extra.0000.part.taco","truncatedAt":512}],"sampleObjects":["cloudsen12-extra.0000.part.taco","cloudsen12-extra.0001.part.taco","cloudsen12-extra.0002.part.taco","cloudsen12-l1c.0000.part.taco","cloudsen12-l1c.0001.part.taco"],"sidecars":[{"bytes":"---\nlicense:\n  - cc0-1.0\nlanguage:\n- en\ntags:\n  - clouds\n  - sentinel-2\n  - image-segmentation\n  - deep-learning\n  - remote-sensing\npretty_name: cloudsen12plus\nviewer: false\n---\n\n<div style=\"text-align: center; border: 1px solid #ddd; border-radius: 10px; padding: 15px; max-width: 250px; margin: auto; background-color: #f9f9f9;\">\n\n  [![Dataset Image](https://tacofoundation.github.io/taco.png)](https://cloudsen12.github.io/)\n  \n<b><p>This dataset follows the TACO specification.</p></b>\n</div>\n\n\n# [cloudsen12plus](https://cloudsen12.github.io/)\n\n**Website:** https://cloudsen12.github.io/\n\n**version:** 1.1.2 \n\n****The largest dataset of expert-labeled pixels for cloud and cloud shadow detection in Sentinel-2****\n\n\nCloudSEN12+ version 1.1.0 is a significant extension of the CloudSEN12 dataset, which doubles the number of \nexpert-reviewed labels, making it, by a large margin, the largest cloud detection dataset to \ndate for Sentinel-2. All labels from the previous version have been curated and refined, enhancing\nthe dataset's truestworthiness. This new release is licensed under CC0, which puts it in the public\ndomain and allows anyone to use, modify, and distribute it without permission or attribution.\n\nThe images are padded from 509x509 to 512x512 and 2000x2000 to 2048x2048 to ensure that the \npatches are divisible by 32. The padding is filled with zeros in the left and bottom sides of the \nimage. For those who prefer traditional storage formats, GeoTIFF files are available in our \n[ScienceDataBank](https://www.scidb.cn/en/detail?dataSetId=2036f4657b094edfbb099053d6024b08&version=V1)\nrepository.\n\n`CloudSEN12+` v.1.1.0 offers three distinct modes, tailored for diverse research and application needs:\n\n- **`cloudsen12-l1c`**: Patches derived from Sentinel-2 Level-1C imagery, including high-quality labels, scribble annotations, and unlabeled data.\n\n- **`cloudsen12-l2a`**: Similar to cloudsen12-l1c but based on Sentinel-2 Level-2A data as processed by Google Earth Engine.\n\n- **`cloudsen12-extra`**: A supplementary collection of metadata to enhance contextual understanding of landscapes. Cloud masks from multiple sources have **NOT** been normalized to align with the CloudSEN12 class schema. This mode includes:\n  - **`elevation:`** Elevation data (meters) sourced from the Copernicus DEM GLO-30.\n  - **`lc10:`** ESA WorldCover 10m v100 land cover product.\n  - **`vv:`** Normalized Sentinel-1 Global Backscatter Model Land Surface (VV polarization).\n  - **`vh:`** Normalized Sentinel-1 Global Backscatter Model Land Surface (VH polarization).    \n  - **`cloudmask_qa60:`** Cloud mask from Sentinel-2 Level-1C.\n  - **`cloudmask_sen2cor:`** Cloud mask from Sentinel-2 Level-2A.\n  - **`cloudmask_s2cloudless:`** Cloud mask generated by Sentinel Hub Cloud Detector.\n  - **`cloudmask_cloudscore_cs_v1:`** Cloud mask generated by [Pasquarella et al. 2023](https://openaccess.thecvf.com/content/CVPR2023W/EarthVision/html/Pasquarella_Comprehensive_Quality_Assessment_of_Optical_Satellite_Imagery_Using_Weakly_Supervised_CVPRW_2023_paper.html).\n  - **`cloudmask_cloudscore_cs_cdf_v1:`** Cloud mask generated by [Pasquarella et al. 2023](https://openaccess.thecvf.com/content/CVPR2023W/EarthVision/html/Pasquarella_Comprehensive_Quality_Assessment_of_Optical_Satellite_Imagery_Using_Weakly_Supervised_CVPRW_2023_paper.html).\n  - **`cloudmask_unetmobv2_v1:`** Cloud mask generated by [Aybar et al. 2022](https://www.nature.com/articles/s41597-022-01878-2).\n  - **`cloudmask_unetmobv2_v2:`** Cloud mask generated by [Aybar et al. 2024](https://www.sciencedirect.com/science/article/pii/S2352340924008163).\n  - **`cloudmask_sensei_v2:`**  Cloud mask generated by [Alistair Francis 2024](https://ieeexplore.ieee.org/document/10505181).\n\n- Changelog:\n\n    - Version 1.1.2:\n      - BUG FIX: cloudmask_qa60 now are in uint16 rather than uint8 to avoid overflow.\n    \n    - Version 1.1.1:\n      - BUG FIX: VV and VH now are in int16 rather than uint16 to avoid overflow.\n  \n    - Version 1.1.0:\n      - We save all GeoTIFF files with discard_lsb=2 to improve the compression ratio.\n  ","encoding":"utf8","key":"README.md","name":"README.md","size":21275,"truncated":true}],"sourceObjectKey":"cloudsen12-extra.0000.part.taco","starterCell":"from huggingface_hub import hf_hub_download\npath = hf_hub_download(repo_id=\"tacofoundation/cloudsen12\", filename=\"cloudsen12-extra.0000.part.taco\", repo_type=\"dataset\")\nprint(path)","starterSamplePath":"cloudsen12-extra.0000.part.taco","totalObjects":16},"summary":"CloudSEN12+ is the largest dataset of expert-labelled pixels for cloud and cloud shadow semantic segmentation in Sentinel-2 imagery, extending CloudSEN12 with twice as many…","title":"CloudSEN12+ v1.1: Expert-Labelled Cloud and Cloud Shadow Detection for Sentinel-2","updatedAt":"2026-11-14T17:26:00.775Z","visibility":"public","websiteUrl":"https://cloudsen12.github.io/"}}