GANs for Natural Images and Pokemons

The focus of this project is the understanding and the uses of Generative Adversarial Networks (GANs). GANs are a class of unsupervised networks that can generate data and can be used in a variety of applications such as Text-to-Text, Image-to-Image, Image-to-Text translation etc. For our experiments, we use a basic implementation of GAN that does Image-to-Image translation i.e. it generates new images based off the original images. Furthermore, we evaluate the GAN model against 3 different datasets namely, MNIST, CFAR10, and a Pokemon dataset from Kaggle. For more details, please visit the GitHub repository.