Closed jarvisWang0903 closed 5 years ago
I'm sorry that the L1 loss in the paper is a type. We will update a new version after CVPR. If you find any other inconsistency between the code and the paper, please ignore the paper. The released code is what I use to produce all the results in the paper. Thanks for pointing out the issue for us.
Thanks for your considerate response and thanks for your open source code! :smile:
In the paper you say that L1 Loss is used for l_per , instead, you use mse loss in your code [cycle _gan_model.py def compute_semantic_loss] ,which one is correct ?