Learning One Class Representations for Face Presentation Attack Detection using Multi-channel Convolutional Neural Networks

TL;DR

This paper introduces a novel face presentation attack detection framework using a one-class classifier with MCCNN, effectively generalizing to unseen attacks by learning robust bonafide representations.

Contribution

It proposes a new MCCNN-based one-class learning approach with a novel loss function for robust face PAD, improving generalization to unseen attacks.

Findings

Superior performance on WMCA dataset against unseen attacks

Effective detection on MLFP and SiW-M datasets with RGB channels

Robustness demonstrated across multiple attack types

Abstract

Face recognition has evolved as a widely used biometric modality. However, its vulnerability against presentation attacks poses a significant security threat. Though presentation attack detection (PAD) methods try to address this issue, they often fail in generalizing to unseen attacks. In this work, we propose a new framework for PAD using a one-class classifier, where the representation used is learned with a Multi-Channel Convolutional Neural Network (MCCNN). A novel loss function is introduced, which forces the network to learn a compact embedding for bonafide class while being far from the representation of attacks. A one-class Gaussian Mixture Model is used on top of these embeddings for the PAD task. The proposed framework introduces a novel approach to learn a robust PAD system from bonafide and available (known) attack classes. This is particularly important as collecting bonafide data and simpler attacks are much easier than collecting a wide variety of expensive attacks. The proposed system is evaluated on the publicly available WMCA multi-channel face PAD database, which contains a wide variety of 2D and 3D attacks. Further, we have performed experiments with MLFP and SiW-M datasets using RGB channels only. Superior performance in unseen attack protocols shows the effectiveness of the proposed approach. Software, data, and protocols to reproduce the results are made available publicly.