A Closer Look at Geometric Temporal Dynamics for Face Anti-Spoofing

TL;DR

This paper introduces GAIN, a geometry-aware model utilizing facial landmarks and spatio-temporal graph convolution to improve face anti-spoofing, especially against unseen attack types and domain shifts, achieving state-of-the-art results.

Contribution

The paper proposes GAIN, a novel geometry-aware network with a cross-attention mechanism that enhances interpretability and robustness in face anti-spoofing tasks.

Findings

Achieves state-of-the-art intra- and cross-dataset performance.

Significantly outperforms existing methods on CASIA-SURF 3DMask with +10.26% AUC.

Demonstrates strong robustness against unseen spoofing types.

Abstract

Face anti-spoofing (FAS) is indispensable for a face recognition system. Many texture-driven countermeasures were developed against presentation attacks (PAs), but the performance against unseen domains or unseen spoofing types is still unsatisfactory. Instead of exhaustively collecting all the spoofing variations and making binary decisions of live/spoof, we offer a new perspective on the FAS task to distinguish between normal and abnormal movements of live and spoof presentations. We propose Geometry-Aware Interaction Network (GAIN), which exploits dense facial landmarks with spatio-temporal graph convolutional network (ST-GCN) to establish a more interpretable and modularized FAS model. Additionally, with our cross-attention feature interaction mechanism, GAIN can be easily integrated with other existing methods to significantly boost performance. Our approach achieves state-of-the-art performance in the standard intra- and cross-dataset evaluations. Moreover, our model outperforms state-of-the-art methods by a large margin in the cross-dataset cross-type protocol on CASIA-SURF 3DMask (+10.26% higher AUC score), exhibiting strong robustness against domain shifts and unseen spoofing types.