Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

questions about style transfer #13

Open
suke27 opened this issue Mar 16, 2020 · 3 comments
Open

questions about style transfer #13

suke27 opened this issue Mar 16, 2020 · 3 comments

Comments

@suke27
Copy link

suke27 commented Mar 16, 2020

Hi, i have a question for style transfer. currently style transfer almost use BN/IN and gram matrix in train process, my question is that, is there any solution which use full convolution layer solution for style transfer?

@ycjing
Copy link
Owner

ycjing commented Mar 16, 2020

Hi @suke27 ,

I don't think so. Normalization layer is critical for the success of style transfer, as analyzed in our AAAI oral paper: https://arxiv.org/abs/1911.06953

@ycjing
Copy link
Owner

ycjing commented Apr 9, 2020

Hi @suke27,

Does it solve your problem?

@suke27
Copy link
Author

suke27 commented Apr 9, 2020

no, i want to remove BN or IN from my layer. because i can not do clip forward in case of BN and IN

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants