Selected projects & software

Open-source packages

• SSL4HFM — Self-supervised learning package for hyperspectral foundation models (four models).
• FUnmix — Fast semi-supervised unmixing Python package (two methods).
• HySUPP — Hyperspectral unmixing Python package (20 methods).
• Unmixing Tutorial Package — Python package + Jupyter notebooks (incl. Colab) for seven unmixing methods.
• HyFTech — Hyperspectral shallow & deep feature extraction toolbox (Python/Matlab), 15 methods.
• HyDe — Python-based, GPU-accelerated hyperspectral denoising package.
• Hyperspectral Image Denoising Toolbox — Matlab toolbox including six hyperspectral denoising methods.
• Fast Wavelet Toolbox — Fast wavelet toolbox for 1D/2D/3D signals.

Current research directions

• Physics-aware foundation models for Earth observation
• Spectral unmixing
• Onboard hyperspectral super-resolution
• Multisensor data fusion