Towards Data-Centric Face Anti-Spoofing: Improving Cross-domain Generalization via Physics-based Data Synthesis
Rizhao Cai, Cecelia Soh, Zitong Yu, Haoliang Li, Wenhan Yang, Alex Kot
TL;DR
The paper tackles cross-domain generalization in face anti-spoofing by shifting from model-centric to data-centric strategies. It introduces FAS-Aug, a physics-based augmentation that simulates the capturing and recapturing processes to synthesize diverse spoofing artifacts, and Spoofing Attack Risk Equalization (SARE) to prevent overfitting to non-invariant cues by minimizing the domain-wise empirical risk variance $L_{SARE} = Var\{R_1, R_2, ..., R_m\}$. The methods are tested with Vision Transformer backbones, achieving strong cross-domain performance on MICO and MICY benchmarks and demonstrating complementary gains when combined with traditional augmentation. The work also discusses data maintenance implications, showing that simply adding more real-face examples does not guarantee improvements, and highlights broader applicability to other security-related tasks. The proposed approach is implemented as a plug-in data transform and is available publicly at GitHub, enabling practical adoption in FAS research and applications.
Abstract
Face Anti-Spoofing (FAS) research is challenged by the cross-domain problem, where there is a domain gap between the training and testing data. While recent FAS works are mainly model-centric, focusing on developing domain generalization algorithms for improving cross-domain performance, data-centric research for face anti-spoofing, improving generalization from data quality and quantity, is largely ignored. Therefore, our work starts with data-centric FAS by conducting a comprehensive investigation from the data perspective for improving cross-domain generalization of FAS models. More specifically, at first, based on physical procedures of capturing and recapturing, we propose task-specific FAS data augmentation (FAS-Aug), which increases data diversity by synthesizing data of artifacts, such as printing noise, color distortion, moiré pattern, \textit{etc}. Our experiments show that using our FAS augmentation can surpass traditional image augmentation in training FAS models to achieve better cross-domain performance. Nevertheless, we observe that models may rely on the augmented artifacts, which are not environment-invariant, and using FAS-Aug may have a negative effect. As such, we propose Spoofing Attack Risk Equalization (SARE) to prevent models from relying on certain types of artifacts and improve the generalization performance. Last but not least, our proposed FAS-Aug and SARE with recent Vision Transformer backbones can achieve state-of-the-art performance on the FAS cross-domain generalization protocols. The implementation is available at https://github.com/RizhaoCai/FAS_Aug.