Normalization Perturbation: A Simple Domain Generalization Method for Real-World Domain Shifts

TL;DR

This paper introduces Normalization Perturbation, a simple yet effective method that perturbs channel statistics in CNNs to improve model generalization across diverse real-world domain shifts without needing target domain data.

Contribution

It proposes a novel normalization perturbation technique that synthesizes diverse styles by perturbing channel statistics, enhancing domain generalization in real-world scenarios.

Findings

Significantly improves model robustness against domain shifts

Effective with only a single source domain

Easy to implement and computationally efficient

Abstract

Improving model's generalizability against domain shifts is crucial, especially for safety-critical applications such as autonomous driving. Real-world domain styles can vary substantially due to environment changes and sensor noises, but deep models only know the training domain style. Such domain style gap impedes model generalization on diverse real-world domains. Our proposed Normalization Perturbation (NP) can effectively overcome this domain style overfitting problem. We observe that this problem is mainly caused by the biased distribution of low-level features learned in shallow CNN layers. Thus, we propose to perturb the channel statistics of source domain features to synthesize various latent styles, so that the trained deep model can perceive diverse potential domains and generalizes well even without observations of target domain data in training. We further explore the style-sensitive channels for effective style synthesis. Normalization Perturbation only relies on a single source domain and is surprisingly effective and extremely easy to implement. Extensive experiments verify the effectiveness of our method for generalizing models under real-world domain shifts.