Unsupervised Feature Disentanglement and Augmentation Network for One-class Face Anti-spoofing

2025-03-29

AI Summary

In single-class face anti-spoofing (FAS), poor generalization to unseen domains and unknown attacks arises from entanglement between liveness-relevant and domain-specific features. To address this, we propose an unsupervised feature disentanglement framework—the first to achieve unsupervised separation of liveness and domain features in single-class FAS. Our method introduces a dual-path augmentation strategy: (i) out-of-distribution (OOD)-guided liveness feature deviation enhancement to sharpen decision boundaries, and (ii) generative domain feature synthesis to enrich domain-invariant representation learning. These are jointly optimized via contrastive learning to promote discriminative, domain-robust representations. Extensive experiments across multiple benchmarks demonstrate that our approach significantly outperforms existing single-class FAS methods, achieving an average 12.6% reduction in equal error rate (EER). Notably, it matches state-of-the-art two-class FAS methods under cross-domain and unknown-attack scenarios, validating its superior robustness and generalization capability.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Application Domains: Misinformation & Fake News

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Security and Privacy: Large-scale security measurements

Abstract

Face anti-spoofing (FAS) techniques aim to enhance the security of facial identity authentication by distinguishing authentic live faces from deceptive attempts. While two-class FAS methods risk overfitting to training attacks to achieve better performance, one-class FAS approaches handle unseen attacks well but are less robust to domain information entangled within the liveness features. To address this, we propose an Unsupervised Feature Disentanglement and Augmentation Network ( extbf{UFDANet}), a one-class FAS technique that enhances generalizability by augmenting face images via disentangled features. The extbf{UFDANet} employs a novel unsupervised feature disentangling method to separate the liveness and domain features, facilitating discriminative feature learning. It integrates an out-of-distribution liveness feature augmentation scheme to synthesize new liveness features of unseen spoof classes, which deviate from the live class, thus enhancing the representability and discriminability of liveness features. Additionally, extbf{UFDANet} incorporates a domain feature augmentation routine to synthesize unseen domain features, thereby achieving better generalizability. Extensive experiments demonstrate that the proposed extbf{UFDANet} outperforms previous one-class FAS methods and achieves comparable performance to state-of-the-art two-class FAS methods.

Problem

• Separates liveness and domain features unsupervisedly for FAS
• Augments liveness features to handle unseen spoof attacks
• Enhances generalizability by synthesizing unseen domain features

Innovation

• Unsupervised disentanglement of liveness and domain features
• Out-of-distribution liveness feature augmentation scheme
• Domain feature augmentation for better generalizability