Adversarial Style Augmentation for Domain Generalized Urban-Scene Segmentation

TL;DR

This paper introduces AdvStyle, an adversarial style augmentation method that enhances domain generalization in semantic segmentation by generating challenging stylized images, leading to improved performance on unseen real-world domains.

Contribution

The paper proposes a novel adversarial style augmentation technique that dynamically generates hard stylized images, improving domain generalization in semantic segmentation and image classification.

Findings

AdvStyle significantly improves segmentation performance on unseen real domains.

The method achieves state-of-the-art results on synthetic-to-real benchmarks.

AdvStyle is easy to implement and adaptable to different models.

Abstract

In this paper, we consider the problem of domain generalization in semantic segmentation, which aims to learn a robust model using only labeled synthetic (source) data. The model is expected to perform well on unseen real (target) domains. Our study finds that the image style variation can largely influence the model's performance and the style features can be well represented by the channel-wise mean and standard deviation of images. Inspired by this, we propose a novel adversarial style augmentation (AdvStyle) approach, which can dynamically generate hard stylized images during training and thus can effectively prevent the model from overfitting on the source domain. Specifically, AdvStyle regards the style feature as a learnable parameter and updates it by adversarial training. The learned adversarial style feature is used to construct an adversarial image for robust model training. AdvStyle is easy to implement and can be readily applied to different models. Experiments on two synthetic-to-real semantic segmentation benchmarks demonstrate that AdvStyle can significantly improve the model performance on unseen real domains and show that we can achieve the state of the art. Moreover, AdvStyle can be employed to domain generalized image classification and produces a clear improvement on the considered datasets.