PhysAug: A Physical-guided and Frequency-based Data Augmentation for Single-Domain Generalized Object Detection

TL;DR

PhysAug introduces a physics-inspired, frequency-based data augmentation method for single-domain generalized object detection, improving robustness across unseen domains by simulating real-world atmospheric variations.

Contribution

The paper proposes PhysAug, a novel physical model-based augmentation leveraging atmospheric optics and frequency spectrum analysis to enhance domain generalization in object detection.

Findings

Achieves 7.3% and 7.2% improvements over baseline on DWD and Cityscape-C datasets.

Outperforms state-of-the-art methods in S-DGOD tasks.

Enhances detector robustness without changing network architecture or loss functions.

Abstract

Single-Domain Generalized Object Detection~(S-DGOD) aims to train on a single source domain for robust performance across a variety of unseen target domains by taking advantage of an object detector. Existing S-DGOD approaches often rely on data augmentation strategies, including a composition of visual transformations, to enhance the detector's generalization ability. However, the absence of real-world prior knowledge hinders data augmentation from contributing to the diversity of training data distributions. To address this issue, we propose PhysAug, a novel physical model-based non-ideal imaging condition data augmentation method, to enhance the adaptability of the S-DGOD tasks. Drawing upon the principles of atmospheric optics, we develop a universal perturbation model that serves as the foundation for our proposed PhysAug. Given that visual perturbations typically arise from the interaction of light with atmospheric particles, the image frequency spectrum is harnessed to simulate real-world variations during training. This approach fosters the detector to learn domain-invariant representations, thereby enhancing its ability to generalize across various settings. Without altering the network architecture or loss function, our approach significantly outperforms the state-of-the-art across various S-DGOD datasets. In particular, it achieves a substantial improvement of $7.3\%$ and $7.2\%$ over the baseline on DWD and Cityscape-C, highlighting its enhanced generalizability in real-world settings.