Improving Super-Resolution Methods via Incremental Residual Learning

TL;DR

This paper introduces Incremental Residual Learning, a framework that enhances super-resolution CNNs by sequentially adding residual branches to better model high-frequency details and improve translation from low to high resolution.

Contribution

It proposes a novel IRL framework that improves existing super-resolution networks by incrementally learning residuals, leading to state-of-the-art performance with minimal additional training time.

Findings

Consistent performance improvements on benchmark datasets.

Achieves state-of-the-art super-resolution results.

Only about 20% increase in training time.

Abstract

Recently, Convolutional Neural Networks (CNNs) have shown promising performance in super-resolution (SR). However, these methods operate primarily on Low Resolution (LR) inputs for memory efficiency but this limits, as we demonstrate, their ability to (i) model high frequency information; and (ii) smoothly translate from LR to High Resolution (HR) space. To this end, we propose a novel Incremental Residual Learning (IRL) framework to address these mentioned issues. In IRL, first we select a typical SR pre-trained network as a master branch. Next we sequentially train and add residual branches to the main branch, where each residual branch is learned to model accumulated residuals of all previous branches. We plug state of the art methods in IRL framework and demonstrate consistent performance improvement on public benchmark datasets to set a new state of the art for SR at only approximately 20% increase in training time.