Use of in-the-wild images for anomaly detection in face anti-spoofing

TL;DR

This paper investigates using in-the-wild images for training anomaly detection models in face anti-spoofing, demonstrating improved generalization to unseen databases with convolutional autoencoders.

Contribution

It introduces a novel training protocol utilizing in-the-wild images for one-class face anti-spoofing detection, enhancing cross-database performance.

Findings

Increased AUC on unseen databases with in-the-wild training data.

Autoencoder-based approach effectively distinguishes real faces from spoofs.

Challenges remain in selecting optimal operating points for deployment.

Abstract

The traditional approach to face anti-spoofing sees it as a binary classification problem, and binary classifiers are trained and validated on specialized anti-spoofing databases. One of the drawbacks of this approach is that, due to the variability of face spoofing attacks, environmental factors, and the typically small sample size, such classifiers do not generalize well to previously unseen databases. Anomaly detection, which approaches face anti-spoofing as a one-class classification problem, is emerging as an increasingly popular alternative approach. Nevertheless, in all existing work on anomaly detection for face anti-spoofing, the proposed training protocols utilize images from specialized anti-spoofing databases only, even though only common images of real faces are needed. Here, we explore the use of in-the-wild images, and images from non-specialized face databases, to train one-class classifiers for face anti-spoofing. Employing a well-established technique, we train a convolutional autoencoder on real faces and compare the reconstruction error of the input against a threshold to classify a face image accordingly as either client or imposter. Our results show that the inclusion in the training set of in-the-wild images increases the discriminating power of the classifier significantly on an unseen database, as evidenced by a large increase in the value of the Area Under the Curve. In a limitation of our approach, we note that the problem of finding a suitable operating point on the unseen database remains a challenge, as evidenced by the values of the Half Total Error Rate.