search

OpenTalker / StyleHEAT

[ECCV 2022] StyleHEAT: A framework for high-resolution editable talking face generation
MIT License
627 stars 77 forks source link

Dataset preprocess #2

Closed youngAt19 closed 2 years ago

youngAt19 commented 2 years ago

Hi, would you share the processed dataset mentioned in your paper or the preprocessing code? Thanks.

youngAt19 commented 2 years ago

I want to reproduce the model in your paper and I try to download the voxceleb following this repo, but it seems that some videos are not available now. (it comes out likes Can not load video EdFbFdf0K7w, broken link) Could you share the voxceleb dataset used in your experiment by google cloud, baidu netdisk or other software?

Luh1124 commented 2 years ago

Hi, would you share the processed dataset mentioned in your paper or the preprocessing code? Thanks.

I have this voxceleb dataset and you can email me

FeiiYin commented 2 years ago

I want to reproduce the model in your paper and I try to download the voxceleb following this repo, but it seems that some videos are not available now. (it comes out likes Can not load video EdFbFdf0K7w, broken link) Could you share the voxceleb dataset used in your experiment by google cloud, baidu netdisk or other software?

We use the this repo to pre-process the dataset. Some download links expired and raise the error Can not load video EdFbFdf0K7w, broken link. For 3DMM parameters extraction, we use Deep3DFaceRecon_pytorch to extract. You can follow PIRenderer to pre-process the dataset for now.

youngAt19 commented 2 years ago

Get it :). Thanks

youngAt19 commented 2 years ago

Hi, would you share the processed dataset mentioned in your paper or the preprocessing code? Thanks.

I have this voxceleb dataset and you can email me

Thank you. I have sent you an email, please check it.