Closed KVBK01 closed 10 months ago
Hi, Thanks for raising the issue. You are correct, the paper results are based on the BIWI dataset processed by following CodeTalker (https://github.com/Doubiiu/CodeTalker/tree/main/BIWI) for fair comparison with previous works. The README has been updated with the correct information now. The old pre-processing steps are for a different project.
Sorry to disturb. I see the BIWI dataprepocessing requires only to download the [faces0x.tgz] files, which is different from other preprocessing ways such as CodeTalker that requires other files like [scans0x.tgz].
I wonder if the provided preprocessing code outputs the exactly same precessed files as CodeTalker.
Thanks again for the good work.