Camila Castro

Masking and Mixing Strategy Cuts Label Needs in Hyperspectral Land-Cover Mapping

Researchers in China report a contrastive-learning framework that classifies hyperspectral images accurately with only five labeled samples per class.