deblur-gan icon indicating copy to clipboard operation
deblur-gan copied to clipboard

the Normalization

Open zengqi0730 opened this issue 5 years ago • 0 comments

In official paper and pytorch code, instance normalization is proposed after conv while batch normalization is used in this keras code. Is there any influence on the final result? Have you test the difference between them?

zengqi0730 avatar May 14 '19 10:05 zengqi0730