Super-Resolution with Generative Adversarial Networks/ ESRGAN-PLUS
In this paper, we provide a GAN architecture inspired of ESRGAN. The changes are dedicated to generator and discriminator. Where we update both structures by adding and removing som layers. For training and testing, run train-test.py. for easy acces, run ESRGNA-PLUS.py. The results have ways to come much better.
In this image we can see the updated architecture of RRDBs
The whole structure of Generator is depicted as top:
If you are any question, do not hesitate and contact me by email: [email protected] & [email protected]