Enhanced Super-Resolution Training via Mimicked Alignment for Real-World Scenes

TL;DR

This paper introduces a plug-and-play module that improves real-world image super-resolution by aligning low-resolution inputs with high-resolution images during training, effectively handling dataset misalignments without affecting inference.

Contribution

A novel alignment-mimicking module that enhances super-resolution robustness to dataset misalignments, compatible with various models and removable during inference.

Findings

Effective on synthetic and real-world datasets

Improves robustness across CNN and Transformer models

No additional parameters during inference

Abstract

Image super-resolution methods have made significant strides with deep learning techniques and ample training data. However, they face challenges due to inherent misalignment between low-resolution (LR) and high-resolution (HR) pairs in real-world datasets. In this study, we propose a novel plug-and-play module designed to mitigate these misalignment issues by aligning LR inputs with HR images during training. Specifically, our approach involves mimicking a novel LR sample that aligns with HR while preserving the degradation characteristics of the original LR samples. This module seamlessly integrates with any SR model, enhancing robustness against misalignment. Importantly, it can be easily removed during inference, therefore without introducing any parameters on the conventional SR models. We comprehensively evaluate our method on synthetic and real-world datasets, demonstrating its effectiveness across a spectrum of SR models, including traditional CNNs and state-of-the-art Transformers. The source codes will be publicly made available at https://github.com/omarAlezaby/Mimicked_Ali .