Is the Released Models Trained on Whole Video Clip?

YuanxunLu / LiveSpeechPortraits

Live Speech Portraits: Real-Time Photorealistic Talking-Head Animation (SIGGRAPH Asia 2021)

MIT License

1.2k stars 209 forks source link

Is the Released Models Trained on Whole Video Clip? #94

Open sjy234sjy234 opened 11 months ago

sjy234sjy234 commented 11 months ago

Is the Released Models Trained on Whole Video Clip? For example, there are 14575 frames in May's clip. Did you use all the frames or make a train/test splits ?