Articles | Volume 18, issue 10
https://doi.org/10.5194/essd-18-7301-2026
https://doi.org/10.5194/essd-18-7301-2026
Data description article
 | 
06 Oct 2026
Data description article |  | 06 Oct 2026

HyBEAR: a hyperspectral benchmark for bare soil detection

Agata M. Wijata, Bogdan Ruszczak, Adriana Niepala, Michal Gumiela, Krzysztof Smykala, Nicolas Longépé, and Jakub Nalepa

Data sets

HyBEAR 🐻 A. Wijata et al. https://doi.org/10.5281/zenodo.17607897

Model code and software

HyBEAR 🐻 A. Wijata et al. https://doi.org/10.5281/zenodo.17607897

Download
Short summary
The bare soil areas detection is an important step for Earth observation and Precision Agriculture applications. The HyBEAR is a novel large-scale dataset of high-res hyperspectral images, with 2 m ground sampling distance, and manual bare soil annotations verified by experts. It includes the largest-to-date dataset (108 mln pixels), the validation procedure, and initial results. The entire parcel's pixel-level labels enable developing the methods for entire fields with no vegetation detection.
Share
Altmetrics
Final-revised paper
Preprint