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.

Sign up for GitHub

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

Jump to bottom

Questiones about training results. #27

Open
Seanseattle opened this issue Jun 29, 2021 · 2 comments
Open

Questiones about training results. #27

Seanseattle opened this issue Jun 29, 2021 · 2 comments

Comments

@Seanseattle
Copy link

Seanseattle commented Jun 29, 2021 •
edited
Loading

Thanks for your wonderful work and opening the source codes. I have implemented the training codes. To make the background stable, I predict the mask and merge the generated image with the target image as

SimSwap/models/networks.py

Line 559 in 39e0c2e

class Generator_Adain_Mask(nn.Module):
However, it seems hard to train, the generated image is totally similar to the target image and doesn't swap the face. I have trained 500K iterations. Do you have any idea to solve this problem?
image
The generated images are in the third row.
@NNNNAI
Copy link
Collaborator

NNNNAI commented Jul 2, 2021

Thanks for your attention to our work. I recomment you to use the default mode "opt.model == 'pix2pixHD" to reprodeuce the base performance. Because I couldn't see your code, I couldn't locate the specific problem. You can first try the default mode to see if the results can be produced. If the results still cannot be produced, please let me know and I will help you further solve the problem. Many thanks~.

@JosephPai JosephPai mentioned this issue Jul 7, 2021
Closed
@jo-dean
Copy link

jo-dean commented Aug 20, 2021

hi，can you share your training code?
i'm appreciate for receiving your reply！

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
3 participants
@Seanseattle @NNNNAI @jo-dean