universome / stylegan-v

[CVPR 2022] StyleGAN-V: A Continuous Video Generator with the Price, Image Quality and Perks of StyleGAN2
https://universome.github.io/stylegan-v
333 stars 36 forks source link

FVD Metrics #41

Closed leoShen917 closed 5 months ago

leoShen917 commented 6 months ago

@universome Hello, I wonder if it is necessary to normalise the data so that it is distributed between [-1,1] when calculating the FVD metrics, I see that there is a normalisation operation in the original repository but I don't find it in your code. Looking forward to your reply, thanks!

universome commented 5 months ago

Hi! The I3D torchscript model takes a [0, 255] videos in uint8 format and performs all the normalization internally.