We integrate Hyperspectral (point spectra) and RGB images from herbarium specimens to build single-modal and multimodal classification models and evaluate the performance of taxonomic species identifications at the level of family, order, and genus.

Each branch contains a different data modality (rgb, hyperspectral, scalograms from hyperspectral) and corresponding data utils, models, and evaluation for given data modality.