search

JeremyCJM / DiffSHEG

[CVPR'24] DiffSHEG: A Diffusion-Based Approach for Real-Time Speech-driven Holistic 3D Expression and Gesture Generation
https://jeremycjm.github.io/proj/DiffSHEG/
BSD 3-Clause "New" or "Revised" License
112 stars 9 forks source link

Training code #1

Closed HaroldChen19 closed 4 months ago

HaroldChen19 commented 4 months ago

Thanks for your awesome work, and when the training code is going to be released?

JeremyCJM commented 4 months ago

Hi Harold, the training code is already in this repo. For the dataset processing, I will write a guide later. If you want to try by yourself, you can refer to the official BEAT and TalkSHOW GitHub repo for the data processing. The general process is extracting motion and audio feature clips via sliding window and save them via lmdb.

HaroldChen19 commented 4 months ago

Hi Harold, the training code is already in this repo. For the dataset processing, I will write a guide later. If you want to try by yourself, you can refer to the official BEAT and TalkSHOW GitHub repo for the data processing. The general process is extracting motion and audio feature clips via sliding window and save them via lmdb.

Thank you so much!!