Learn Convolutional Neural Network for Face Anti-Spoofing

TL;DR

This paper introduces a CNN-based approach for face anti-spoofing that significantly outperforms traditional handcrafted features, demonstrating improved accuracy and generalization across datasets.

Contribution

The paper proposes using deep CNNs with data pre-processing for face anti-spoofing, achieving superior performance over traditional methods and better cross-dataset generalization.

Findings

Over 70% reduction in HTER on CASIA and REPLAY-ATTACK datasets.

CNN features outperform handcrafted texture features.

Combined dataset training reduces dataset bias.

Abstract

Though having achieved some progresses, the hand-crafted texture features, e.g., LBP [23], LBP-TOP [11] are still unable to capture the most discriminative cues between genuine and fake faces. In this paper, instead of designing feature by ourselves, we rely on the deep convolutional neural network (CNN) to learn features of high discriminative ability in a supervised manner. Combined with some data pre-processing, the face anti-spoofing performance improves drastically. In the experiments, over 70% relative decrease of Half Total Error Rate (HTER) is achieved on two challenging datasets, CASIA [36] and REPLAY-ATTACK [7] compared with the state-of-the-art. Meanwhile, the experimental results from inter-tests between two datasets indicates CNN can obtain features with better generalization ability. Moreover, the nets trained using combined data from two datasets have less biases between two datasets.