how to make/create ground truth image file for training Hyperspectral image data?

Question

Need to train hyperspectral data, how can I build a training dataset from scracth and a ground truth array/image to be used for the same? Spectral python doesn't support training by png files.

Answer 1

We need to manually create masks for particular classes on any 2D image and convert it into a .raw ENVI file with a single band.

Ground truth image need not be a hypercube necessarily.