search

naver-ai / StyleMapGAN

Official pytorch implementation of StyleMapGAN (CVPR 2021)
https://www.youtube.com/watch?v=qCapNyRA_Ng
Other
458 stars 81 forks source link

Questions about metrics for reconstruction #12

Closed Annonymous-repos closed 3 years ago

Annonymous-repos commented 3 years ago

Hello, thanks for sharing your code. I'm very interested in your work and I got a question when I run the metrics for reconstruction. The output of metrics.reconstruction are 0.015(mse) and 0.214(lpips). The output seems much better than the results reported in Table 3 in the paper (mse=0.024, lpips=0.242). I would appreciate it if you could give me an explanation. image

blandocs commented 3 years ago

Thank you for your interest :)

We publish the pretrained networks with further training. The paper shows a 5M-image-trained result but we publish a 20M-image-trained model.

Please refer to this comment.

Annonymous-repos commented 3 years ago

Thanks for your reply.

Annonymous-repos commented 3 years ago

Thank you for your interest :)

We publish the pretrained networks with further training. The paper shows a 5M-image-trained result but we publish a 20M-image-trained model.

Please refer to this comment.

Would you mind providing the 5M-image-trained model?

blandocs commented 3 years ago

Thank you for your interest :) We publish the pretrained networks with further training. The paper shows a 5M-image-trained result but we publish a 20M-image-trained model. Please refer to this comment.

Would you mind providing the 5M-image-trained model?

Please refer to this commit :)

Annonymous-repos commented 3 years ago

Thanks a lot :)