Abstract: We propose to leverage image super sampling, image in-painting, image colorization and image denoising networks as priors to address image restoration problems which is a well-studied problem in computer vision and image processing, as well as a test bed for low-level image modelling algorithms. Our techniques results in removal of undesirable objects from the images without leaving traces like artifacts ghosts. We have implemented different types of autoencoders that are trained andtested using dataset named “Labeled Faces in the Wild” contain 13,233 images of faces collected from the web for accomplishment of project. Image restoration is for restoring true images from their observed but degraded versions, it is usually used for preprocessing observed images so that successive image processing and examination become more devoted. The main advantage of our project is it can be used andimplemented in various fields whereas application ranges from medical and investigation imaging, to forensic science, etc. This article introduces some fundamental methods as mentioned above.Keyword:-Autoencoder, Convolutional Neural Network (CNN/ ConvNet), ground truth, Deconvolutional , Super Sampling, Low Resolution (LR), image noise, degradation, masked image, upsampling, maxpooling