Feature-Based Style Randomization for Domain Generalization

• Published in: IEEE transactions on circuits and systems for video technology Vol. 32; no. 8; pp. 5495 - 5509
• Main Authors: Wang, Yue, Qi, Lei, Shi, Yinghuan, Gao, Yang
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.08.2022; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Adaptation models; Data augmentation; Data models; Domain generalization; Domains; Feature extraction; Modules; Random noise; Randomization; style randomization; Task analysis; Training; Training data
• ISSN: 1051-8215; 1558-2205
• DOI: 10.1109/TCSVT.2022.3152615

As a recent noticeable topic, domain generalization (DG) aims to first learn a generic model on multiple source domains and then directly generalize to an arbitrary unseen target domain without any additional adaption. In previous DG models, by generating virtual data to supplement observed source domains, the data augmentation based methods have shown its effectiveness. To simulate the possible unseen domains, most of them enrich the diversity of original data via image-level style transformation. However, we argue that the potential styles are hard to be exhaustively illustrated and fully augmented due to the limited referred styles, leading the diversity could not be always guaranteed. Unlike image-level augmentation, we in this paper develop a simple yet effective feature-based style randomization module to achieve feature-level augmentation, which can produce random styles via integrating random noise into the original style. Compared with existing image-level augmentation, our feature-level augmentation favors a more goal-oriented and sample-diverse way. Furthermore, to sufficiently explore the efficacy of the proposed module, we design a novel progressive training strategy to enable all parameters of the network to be fully trained. Extensive experiments on three standard benchmark datasets, i.e. , PACS, VLCS and Office-Home, highlight the superiority of our method compared to the state-of-the-art methods.