In supervised classification, how polygon training sample works?

If I select polygon samples to train my model, how the value of pixels used? Does the model take DN of each pixel or the average of each object (polygon)?

02 October 2020 4,999 2 View