OpenTalker/video-retalking

Perceptual Loss

Opened this issue · 1 comments

ducMNSD commented

I have a problem when finetuning your L-net ckpt
Could you share me your config of perceptual loss in training phase ?, I want to know vgg network you using, which layers is apply and how to normalize of input for perceptual loss ?. Thanks you very much .
24

kunncheng commented

Sorry for the late reply. We followed the perceptual loss used in this repo https://github.com/RenYurui/PIRender/blob/main/loss/perceptual.py