Extract 6d pose sequence and train audio2pose model?

Zejun-Yang / AniPortrait

AniPortrait: Audio-Driven Synthesis of Photorealistic Portrait Animation

Apache License 2.0

4.65k stars 582 forks source link

Extract 6d pose sequence and train audio2pose model? #178

Open roudimit opened 5 months ago

roudimit commented 5 months ago

Hi, thanks for the open-source! I ran /scripts/preprocess_dataset.py but I don't think the 6d pose sequences were extracted, can you please let me know how to do this? Do you use Mediapipe or some other tool?

Is audio2pose also trained on the same 1 hour of data used to train audio2mesh?