Super-resolution aims at increasing the resolution and level of detail within an image. The current state of the art in general single-image super-resolution is held by NESRGAN+, which injects a Gaussian noise after each residual layer at training time. In this paper, we harness evolutionary methods to improve NESRGAN+ by optimizing the noise injection at inference time. More precisely, we use Diagonal CMA to optimize the injected noise according to a novel criterion combining quality assessment and realism. Our results are validated by the PIRM perceptual score and a human study. Our method outperforms NESRGAN+ on several standard super-resolution datasets. More generally, our approach can be used to optimize any method based on noise injection.
@article{arxiv.2009.12177,
title = {Tarsier: Evolving Noise Injection in Super-Resolution GANs},
author = {Baptiste Roziere and Nathanal Carraz Rakotonirina and Vlad Hosu and Andry Rasoanaivo and Hanhe Lin and Camille Couprie and Olivier Teytaud},
journal= {arXiv preprint arXiv:2009.12177},
year = {2020}
}