This is a part of BITS Pilani , Goa Neural Networks course in this project we had to remove rain and snow from images in which rain was added through photoshop . I used a Denoising autoencoder for this project . Another alternative was using Image De-raining Conditional General Adversarial Network (ID-CGAN) inspired by paper https://arxiv.org/abs/1701.05957 . By using autoencoder I got a psnr of almost 28 on training data and 23 on testing data . The data is same as given in the paper . Testing accuracy was above 80 percent .