Please present reference. How much it should be all area of training samples in the classification of satellite data relative to the whole study area?

Similar questions and discussions