An Unbiased Benchmark for Domain Generalization Face Anti-Spoofing

• Raul Almeida UFPR
• Bruno Kamarowski UFPR
• Bernardo Biesseck UFPR
• Luiz Coelho único
• Roger Granada único
• David Menotti UFPR

Resumo

Domain Generalization for Face Anti-Spoofing (DGFAS) is an area of growing interest due to its importance in fraud prevention in Face Recognition Systems. Current benchmarks used for DG-FAS in evaluating state-of-the-art methods allow verification of test set performance during training, which causes bias towards test data. Consequently, practitioners cannot properly translate research conclusions to real-world applications, since there is no access to labels for production data in practice. We propose as an alternative an unbiased benchmark where a validation dataset is used so that the model’s generalization capability is evaluated without compromising restricted data and the scientific rigor of research. Our experiments show that model performance benefits from current biased benchmarks and that introducing a new validation dataset makes for more challenging and scientifically rigorous benchmarks that also better represent real-world performance. We additionally experiment with training on current standard benchmarks and testing on WFAS, a recent in-the-wild large FAS dataset with more attack types than the standard datasets for DG-FAS, and similarly observe poor generalization capabilities for state-of-the-art models.

Referências

Z. Yu, Y. Qin, X. Li, C. Zhao, Z. Lei, and G. Zhao, “Deep learning for face anti-spoofing: A survey,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023.

Z. Zhang, J. Yan, S. Liu, Z. Lei, D. Yi, and S. Z. Li, “A face anti-spoofing database with diverse attacks,” in 2012 5th IAPR International Conference on Biometrics (ICB), 2012, pp. 26–31.

B. M. Le and S. S. Woo, “Gradient alignment for cross-domain face anti-spoofing,” in CVPR, 2024, pp. 188–199.

Y. Sun, Y. Liu, X. Liu, Y. Li, and W.-S. Chu, “Rethinking domain generalization for face anti-spoofing: Separability and alignment,” in CVPR, 2023.

Q. Zhou, K.-Y. Zhang, T. Yao, X. Lu, R. Yi, S. Ding, and L. Ma, “Instance-aware domain generalization for face anti-spoofing,” in CVPR, 2023.

P. Zhang, X. Guo, Y. Zhang, J. Yang, W. Zhang, and Q. Li, “Domain generalization via shuffled style assembly for face anti-spoofing,” in CVPR, 2022, pp. 16 066–16 075.

D. Wang, J. Guo, Q. Shao, H. He, Z. Chen, C. Xiao, A. Liu, S. Escalera, H. J. Escalante, Z. Lei, J. Wan, and J. Deng, “Wild face anti-spoofing challenge 2023: Benchmark and results,” in CVPRW, 2023.

Q. Zhou, K.-Y. Zhang, T. Yao, X. Lu, S. Ding, and L. Ma, “Test-time domain generalization for face anti-spoofing,” in CVPR, 2024, pp. 175–187.

R. Cai, Z. Yu, C. Kong, H. Li, C. Chen, Y. Hu, and A. C. Kot, “S-adapter: Generalizing vision transformer for face anti-spoofing with statistical tokens,” IEEE Transactions on Information Forensics and Security, vol. 19, pp. 8385–8397, 2024.

Z. Li, H. Li, K.-Y. Lam, and A. C. Kot, “Unseen face presentation attack detection with hypersphere loss,” in ICASSP, 2020, pp. 2852–2856.

I. Chingovska, A. Anjos, and S. Marcel, “On the effectiveness of local binary patterns in face anti-spoofing,” in 2012 BIOSIG - Proceedings of the International Conference of Biometrics Special Interest Group (BIOSIG), 2012, pp. 1–7.

D. Wen, H. Han, and A. K. Jain, “Face spoof detection with image distortion analysis,” IEEE Transactions on Information Forensics and Security, vol. 10, no. 4, pp. 746–761, 2015.

Z. Boulkenafet, J. Komulainen, L. Li, X. Feng, and A. Hadid, “Oulu-npu: A mobile face presentation attack database with real-world variations,” in 2017 12th IEEE International Conference on Automatic Face & Gesture Recognition (FG 2017), 2017.

Y. Liu, A. Jourabloo, and X. Liu, “Learning deep models for face anti-spoofing: Binary or auxiliary supervision,” in CVPR, June 2018.

H. Li, W. Li, H. Cao, S. Wang, F. Huang, and A. C. Kot, “Unsupervised domain adaptation for face anti-spoofing,” IEEE Transactions on Information Forensics and Security, vol. 13, no. 7, pp. 1794–1809, 2018.

K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in CVPR, 2016, pp. 770–778.

R. Wightman, “Github: Pytorch image models,” [link], 2019.