Stylized Synthetic Augmentation further improves Corruption Robustness

TL;DR

This paper introduces a novel augmentation pipeline combining synthetic images with neural style transfer, significantly enhancing corruption robustness in image classifiers despite reduced FID scores.

Contribution

It demonstrates that stylized synthetic data, when combined with existing augmentations, improves corruption robustness beyond previous methods.

Findings

Achieves state-of-the-art robustness on CIFAR-10-C, CIFAR-100-C, TinyImageNet-C

Stylized synthetic data complements rule-based augmentations effectively

Method improves robustness despite lower FID scores

Abstract

This paper proposes a training data augmentation pipeline that combines synthetic image data with neural style transfer in order to address the vulnerability of deep vision models to common corruptions. We show that although applying style transfer on synthetic images degrades their quality with respect to the common Frechet Inception Distance (FID) metric, these images are surprisingly beneficial for model training. We conduct a systematic empirical analysis of the effects of both augmentations and their key hyperparameters on the performance of image classifiers. Our results demonstrate that stylization and synthetic data complement each other well and can be combined with popular rule-based data augmentation techniques such as TrivialAugment, while not working with others. Our method achieves state-of-the-art corruption robustness on several small-scale image classification benchmarks, reaching 93.54%, 74.9% and 50.86% robust accuracy on CIFAR-10-C, CIFAR-100-C and TinyImageNet-C, respectively