# Suppressing Model Overfitting for Image Super-Resolution Networks

## TL;DR

This paper introduces MixUp and a learned degradation data synthesis method to effectively reduce overfitting in large image super-resolution models trained on limited data, improving generalization.

## Contribution

It proposes a novel combination of MixUp and synthetic data generation with learned degradation to suppress overfitting in large super-resolution networks.

## Key findings

- Achieved second place in NTIRE2019 Real SR Challenge.
- Effectively reduces overfitting with limited training data.
- Enhances model generalization in real-world scenarios.

## Abstract

Large deep networks have demonstrated competitive performance in single image super-resolution (SISR), with a huge volume of data involved. However, in real-world scenarios, due to the limited accessible training pairs, large models exhibit undesirable behaviors such as overfitting and memorization. To suppress model overfitting and further enjoy the merits of large model capacity, we thoroughly investigate generic approaches for supplying additional training data pairs. In particular, we introduce a simple learning principle MixUp to train networks on interpolations of sample pairs, which encourages networks to support linear behavior in-between training samples. In addition, we propose a data synthesis method with learned degradation, enabling models to use extra high-quality images with higher content diversity. This strategy proves to be successful in reducing biases of data. By combining these components -- MixUp and synthetic training data, large models can be trained without overfitting under very limited data samples and achieve satisfactory generalization performance. Our method won the second place in NTIRE2019 Real SR Challenge.