This an unofficial implementation of CycleGAN using pytorch lightning, during my research internship.
I go for pytorch lightning choice, for its flexibility and because it offer a high level interface of pytorch, which facilitate the process to go from an idea to a real implementation.
I work on how to use image to image translation GAN based methods for realism enhancement.