search

liuh127 / NTIRE-2021-Dehazing-Two-branch

Official PyTorch implementation of Two-branch Dehazing, wining runner-up awards in NTIRE 2021.
MIT License
51 stars 7 forks source link

question about Adversarial loss function #13

Open akashdhedhi opened 2 years ago

akashdhedhi commented 2 years ago

Why do you use sigma and N when creating the adversarial loss function? What does N stand for?

and you use (1-fake out) directly in your code; why not use the log value?