We describe our solution for the PIRM Super-Resolution Challenge 2018 where we achieved the 2nd best perceptual quality for average RMSE<=16, 5th best for RMSE<=12.5, and 7th best for RMSE<=11.5. We modify a recently proposed Multi-Grid Back-Projection (MGBP) architecture to work as a generative system with an input parameter that can control the amount of artificial details in the output. We propose a discriminator for adversarial training with the following novel properties: it is multi-scale that resembles a progressive-GAN; it is recursive that balances the architecture of the generator; and it includes a new layer to capture significant statistics of natural images. Finally, we propose a training strategy that avoids conflicts between reconstruction and perceptual losses. Our configuration uses only 281k parameters and upscales each image of the competition in 0.2s in average.
@article{arxiv.1809.10711,
title = {Multi-Scale Recursive and Perception-Distortion Controllable Image Super-Resolution},
author = {Pablo Navarrete Michelini and Dan Zhu and Hanwen Liu},
journal= {arXiv preprint arXiv:1809.10711},
year = {2019}
}
Comments
In ECCV 2018 Workshops. Won 2nd place in Region 3 of PIRM-SR Challenge 2018. Code and models are available at https://github.com/pnavarre/pirm-sr-2018