Dear all,
when reading through the segmentation tutorial I noticed that the logits, corresponding to pixels that are predicted to be a particular class, are summed. Since logits are unbounded, I was wandering why the output is not normalized by means of a softmax prior summation.
Many thanks in advance!