NVIDIA / vid2vid

Pytorch implementation of our method for high-resolution (e.g. 2048x1024) photorealistic video-to-video translation.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

The question about the "flowNet(real_B, real_B_prev) "

ZhouYiiFeng opened this issue · comments

Dear author,
Thanks for your sharing. I was confused with the order of the input to the "flowNet". In the code of FlowNet2.0 it writes that FlowNet2.forward(im1,im2). According to the dataloader, FlowNet2.forward(im1,im2) means that the im1 is the previous frame and the im2 is current frame. So is there any problem? or my understanding is wrong?
I really need your help. Thank you very much.

The warp means uses back-ward flow to advect the previous frame toward the current frame.