Recoloring Grayscale Images using GAN

In this paper, we intend to develop an efficient method for Colorizing Grayscale Images such that the output images are close to their real colors. The main objective is to produce a precise image should be produced, which means that the output image should be as close to its natural color as possible. In colorization, there is no aim to recover the ground truth color; rather, the aim is to create a plausible colorization that is useful to the user, even if it differs from the ground truth color. In order to solve this problem, we use Convolutional Neural Networks (CNN) and a variety of deep learning techniques. Instead of using vectors drawn at random from the probability distribution, we employ Conditional Generative Adversarial Networks (GAN). This will enable it to extract different features and understand the correlations between them in order to predict colored images.