CVID

CVID stands for Conditional Variational Image Deraining. It is a software tool used for removing rain from images. This task can be challenging because rain intensity varies in both space and color channels. The authors of this paper have proposed a new method called CVID which uses a Conditional Variational Auto-Encoder (CVAE) to provide diverse predictions for the rainy image. They also use a spatial density estimation (SDE) module to estimate a rain density map for each image and a channel-wise (CW) deraining scheme to account for variations in rain density across different color channels. Experiments have shown that the CVID network outperforms previous deterministic methods on image deraining.