{"success":true,"data":{"author":"Hu et al. (Technical University of Munich, DLR); TACO repackaging by ISP, Universitat de Valencia","bucketAccessType":"public","bucketName":"isp-uv-es/mdas-taco","bucketPrefix":"","category":"Earth Observation & Geospatial","cloudProvider":"huggingface","createdAt":"2026-11-16T14:11:25.865Z","description":"MDAS (Earth System Science Data, 2023) is a multimodal remote sensing benchmark over the city of Augsburg, Germany, with every modality acquired or simulated for the same day, 7 May 2018. It covers three co-registered sub-areas with eight sensor products at six ground sampling distances: 0.2 m 3K aerial RGB and a 0.3 m DSM, 2.2 m HySpex airborne hyperspectral imagery (368 bands), 10 m and 30 m EeteS-simulated EnMAP (242 bands), EeteS-simulated Sentinel-2, real Sentinel-2 L2A (12 bands) and Sentinel-1 GRD. OpenStreetMap reference rasters give 19-class land use, 5-class water and binary building masks, supporting super-resolution, spectral unmixing, land cover segmentation and data fusion experiments. This TACO copy keeps every product on its native grid, cuts each sub-area into 30 tiles, and assigns whole sub-areas to train, validation and test. About 6.6 GB, read with tacoreader. The data are published by TUM under CC BY-SA 4.0, with the OSM layers under ODbL; cite Hu et al. 2023.","displayTitle":"MDAS: Multimodal Augsburg Benchmark (HySpex, EnMAP-simulated, Sentinel-1/2, aerial RGB, DSM) (TACO)","doi":"10.5194/essd-15-113-2023","id":"409d2501-eef3-5648-b358-8a19039e47d6","license":"CC-BY-SA-4.0","licenseEvidenceUrl":"https://essd.copernicus.org/articles/15/113/2023/","objectCountEstimate":9,"probeStatus":"ok","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"isp-uv-es/mdas-taco\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"isp-uv-es/mdas-taco\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=isp-uv-es, license:cc-by-sa-4.0, size_categories:1K<n<10K, format:imagefolder, downloads=77"],"format":"mixed","generatedAt":"2026-11-16T14:11:33.810Z","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-sa-4.0\npretty_name: \"MDAS (Augsburg multimodal data set)\"\nsize_categories:\n- n<1K\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# MDAS (Augsburg multimodal data set)\n\n> **This is a repackaging, not a new dataset.** It is **MDAS (Augsburg multimodal data set)** by Technical University of Munich / DLR (Hu et al.), OpenStreetMap contributors, 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://essd.copernicus.org/articles/15/113/2023/) · [paper](https://doi.org/10.5194/essd-15-113-2023) · licence: CC-BY-SA-4.0, ODbL-1.0\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{hu2023mdas,\n  title   = {MDAS: a new multimodal benchmark dataset for remote sensing},\n  author  = {Hu, Jingliang and Liu, Rong and Hong, Danfeng and Camero, Andr{\\'e}s and Yao, Jing and Schneider, Mathias and Kurz, Franz and Segl, Karl and Zhu, Xiao Xiang},\n  journal = {Earth System Science Data},\n  volume  = {15},\n  number  = {1},\n  pages   = {113--131},\n  year    = {2023},\n  doi     = {10.5194/essd-15-113-2023}\n}\n```\n\n## About the data\n\nThree co-registered sub-areas of Augsburg, Germany, acquired 7 May 2018. Converted from the original GeoTIFFs without resampling: every leaf keeps its native grid.\n\n**90 samples** · splits: `test` 30 · `train` 30 · `validation` 30 · tasks: `semantic-segmentation`, `super-resolution`\n\nPackaged as [TACO v3](https://asterisk.coop/taco/spec/).\n\n<details>\n<summary>Full description</summary>\n\n**Products.** Eight sensor products at six ground sampling distances:\n- 0.2 m 3K aerial RGB and 0.3 m 3K DSM\n- 2.2 m HySpex hyperspectral, 368 bands\n- 10 m and 30 m EeteS-simulated EnMAP, 242 bands\n- 10 m EeteS-simulated Sentinel-2, Sentinel-2 L2A (12 bands) and Sentinel-1 GRD\n\nThree OpenStreetMap reference rasters come with them: 19-class land use, 5-class water and binary buildings.\n\n**Bands.** Hyperspectral channel centres and FWHM are taken from the release's own ENVI headers.\n\n**Omissions.** The 2.2 m EeteS EnMAP product exists for sub_area_1 only and is not carried. The 3K RGB's constant alpha channel is dropped.\n\n**Legends.** Land use keeps its native legend with no canonical mapping, being land use rather than land cover. Water is additionally mapped onto `ml:lc11`.\n\n**Splits.** The release publishes none. Each sub-area is cut into 30 tiles, and `split` assigns whole sub-areas: sub_area_3 is train, sub_area_1 validation and sub_area_2 test.\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/mdas-taco\", \"mdas.zip\", repo_type=\"dataset\")\nds = Dataset(path)\nplot_sample(ds[0])\n```\n\nor from a local copy:\n\n```python\nds = Dataset(\"mdas.zip\")\nsample = ds[0]                      # {slot name: SlotValue}, arrays decoded\nsample[\"rgb_3k\"].array.shape\n```\n\nMetadata without decoding anything:\n\n```python\nimport taco\ntaco.read(\"mdas.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| role | slot | holds | modality | detail |\n|---|---|---|---|---|\n| input | `rgb_3k` | raster | optical | 3 band(s), requantised ","encoding":"utf8","key":"README.md","name":"README.md","size":5742,"truncated":true}],"sourceObjectKey":"assets/logos.png","starterCell":"from huggingface_hub import hf_hub_download\npath = hf_hub_download(repo_id=\"isp-uv-es/mdas-taco\", filename=\"assets/logos.png\", repo_type=\"dataset\")\nprint(path)","starterSamplePath":"assets/logos.png","totalObjects":9},"summary":"MDAS (Earth System Science Data, 2023) is a multimodal remote sensing benchmark over the city of Augsburg, Germany, with every modality acquired or simulated for the same day, 7…","title":"MDAS: Multimodal Augsburg Benchmark (HySpex, EnMAP-simulated, Sentinel-1/2, aerial RGB, DSM) (TACO)","updatedAt":"2026-11-16T16:08:23.634Z","visibility":"public","websiteUrl":"https://huggingface.co/datasets/isp-uv-es/mdas-taco"}}