conv neural network - What is the Loss Function in the original U-net paper?

Question

In their paper, the authors state that they calculate the energy function by a pixel-wise softmax over the final feature map in combination with the cross entropy loss function. Then they defined the pixel-wise softmax p_{k}(x) and said that the cross entropy at each position penalises the deviation of p_{l(x)}(x) from 1 by 1

Now I don't understand if 1 is the loss function or just part of the cross entropy formula. If it is only part of the cross entropy formula, what would the full loss function look like in the binary case?

question from:https://stackoverflow.com/questions/65541038/what-is-the-loss-function-in-the-original-u-net-paper

Answer 1

Waitting for answers