Tag
A lightweight set-based deep learning framework using a transformer is presented for atmospheric compensation in passive long-wave infrared hyperspectral imaging, jointly estimating transmittance, atmospheric path radiance, and downwelling spectrum from multi-range radiance measurements.
A researcher shares their struggle with achieving only ~50% accuracy using SSL methods (BYOL, MAE, VICReg) for hyperspectral crop stress classification on cabbage nitrogen deficiency detection, seeking advice on SSL techniques, feature engineering, and model architectures better suited for spectral data.