A repo for learning Gan and applying them on super resolution of images
Generative adversarial networks basically model a generator and discriminator where a generator generates images and a discriminator decides if these images are passable as real or not, they work against each other. The right balance however is finding the right balance where a generator generates decent enough images and a discriminator is powerful but not very powerful too. Its an interesting field but well, training images are hard and my results were really hazy. Use the TFDS dataset, seems like the right place to get started