Advancing Cross-Domain Generalizability in Face Anti-Spoofing: Insights, Design, and Metrics

TL;DR

This paper introduces a new approach for face anti-spoofing that improves cross-domain generalization by using a probabilistic ensemble of backbones, a refined video-level metric, and uncertainty measurement, leading to state-of-the-art results.

Contribution

It proposes a novel ensemble method from a Bayesian perspective and a new metric for better analysis of model stability and generalization in face anti-spoofing.

Findings

Ensembled backbone improves robustness and accuracy.

The new metric quantifies bias and variance in predictions.

Proposed method outperforms existing methods on multiple datasets.

Abstract

This paper presents a novel perspective for enhancing anti-spoofing performance in zero-shot data domain generalization. Unlike traditional image classification tasks, face anti-spoofing datasets display unique generalization characteristics, necessitating novel zero-shot data domain generalization. One step forward to the previous frame-wise spoofing prediction, we introduce a nuanced metric calculation that aggregates frame-level probabilities for a video-wise prediction, to tackle the gap between the reported frame-wise accuracy and instability in real-world use-case. This approach enables the quantification of bias and variance in model predictions, offering a more refined analysis of model generalization. Our investigation reveals that simply scaling up the backbone of models does not inherently improve the mentioned instability, leading us to propose an ensembled backbone method from a Bayesian perspective. The probabilistically ensembled backbone both improves model robustness measured from the proposed metric and spoofing accuracy, and also leverages the advantages of measuring uncertainty, allowing for enhanced sampling during training that contributes to model generalization across new datasets. We evaluate the proposed method from the benchmark OMIC dataset and also the public CelebA-Spoof and SiW-Mv2. Our final model outperforms existing state-of-the-art methods across the datasets, showcasing advancements in Bias, Variance, HTER, and AUC metrics.