Sen4AgriNet: Multi-Year, Multi-Country Sentinel-2 Crop Classification Benchmark (Catalonia, France)

Provider
Sykas, Sdraka, Zografakis, Papoutsis (Orion Lab, National Observatory of Athens / NTUA)
License
MIT (source)
Size
10,016 files
Format
NetCDF (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/orion-ai-lab/S4A)
Access
Free and public. No account with the data provider is needed.
Category
Agriculture & Environment
Source
https://arxiv.org/abs/2204.00951

About this dataset

Sen4AgriNet (IEEE JSTARS, 2022) is a Sentinel-2 time series benchmark for crop type classification, parcel segmentation and parcel counting, labelled from farmer declarations in the Land Parcel Identification System (LPIS) of Catalonia (2019 and 2020) and France (2019). Sentinel-2 L1C images with under 10 percent cloud over the selected tiles were split into 900 non-overlapping patches per tile; each patch holds 366x366 pixels for the 10 m bands, 183x183 for 20 m and 61x61 for 60 m, keeping all 13 spectral bands at native resolution. Labels follow a harmonised FAO Indicative Crop Classification and CORINE based taxonomy with 9 groups and 168 classes and sub-classes (161 crop related), and a second mask gives a unique id per parcel. Samples are NetCDF patches under data/ named year_tile_patch (for example 2019_31TCF_patch_10_14) with year, country code and tile fields, loadable with the S4A.py loading script. About 281 GB. The authors release Sen4AgriNet under the MIT licence, with the Catalan and French LPIS inputs under the Open Data Commons Attribution licence; cite Sykas et al. 2022.

How to load this dataset in Python

from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="orion-ai-lab/S4A", filename="S4A.py", repo_type="dataset")
print(path)

Data structure

File types in sampled files: NetCDF

Files

Example file: S4A.py

Analyze this dataset without downloading it

On Scigantic, Sen4AgriNet: Multi-Year, Multi-Country Sentinel-2 Crop Classification Benchmark (Catalonia, France) opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Agriculture & Environment datasets