zsyOAOA / S2VD

Semi-supervised Video Deraining with Dynamical Rain Generator (CVPR, 2021, Pytorch)

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

About the derainer network

sfwang20 opened this issue · comments

Hello! Thanks for your excellent work.
I wonder about the network architecture of the derainer f(-;w) in Figure 2. of your paper:
Why does it need to minus original image after the Pixel-Shuffle layer ?
Thank you.

Note that the Pixel-shuffle layer outputs the rain layer. The original rainy image minuses the output rain layer to obtain the rain-free image.