Plan to add StyleGAN2 baseline for experiments about pytorch-studiogan HOT 7 CLOSED

Hi!

Thank you so much for your interest in my project.

Recently, many researchers use the generator architecture used in StyleGAN_V2 to verify their method. I also believe that implementing the generator architecture with PyTorch will be helpful for deep learning lovers. However, I am preparing a submission for ICCV and do not have enough time to implement and verify the model.

After finishing the work, I will add the generator model of StyleGAN_v2.

Thank you.

Sincerely,

Minguk

from pytorch-studiogan.

Thanks, @mingukkang for your great work! We will keep this open till the performance (FID) gets verified.

from pytorch-studiogan.

Finally we add StyleGAN2 code!

Please refer to ReadME and ./src/configs/CIFAR10/StyleGAN2.yaml for more details.

from pytorch-studiogan.

Thanks for your herculean effort!! The StudioGAN library is better than ever. Can we also obtain styleGAN evaluation metrics on the same validation set for CIFAR-10 as well?

from pytorch-studiogan.

The synthesis results reported by ReACGAN paper are from the current version of StudioGAN.

I will upload the logs, checkpoints, and configurations for the experiments soon.

Thank you.

from pytorch-studiogan.

Thank you.

We add configurations, checkpoints, and logs for StyleGAN2.

Please refer to the ReadME file for more details.

Best,

Minguk

from pytorch-studiogan.

Definitely, you can.

Use -ref "test" instead of -ref "train".

Thank you.

from pytorch-studiogan.