Multi-Frames Temporal Abnormal Clues Learning Method for Face Anti-Spoofing

TL;DR

This paper introduces EulerNet, a temporal feature fusion network for face anti-spoofing that leverages abnormal clues from continuous frames, with a new labeling method and a large-scale dataset, outperforming existing methods.

Contribution

Proposes EulerNet, a novel temporal network with differential filters and residual pyramids, and a low-cost landmark-based labeling method, along with a large real-world dataset for face anti-spoofing.

Findings

Outperforms state-of-the-art on OULU-NPU dataset

Successfully deployed in real-world systems

Uses large-scale diverse dataset with 30,000 samples

Abstract

Face anti-spoofing researches are widely used in face recognition and has received more attention from industry and academics. In this paper, we propose the EulerNet, a new temporal feature fusion network in which the differential filter and residual pyramid are used to extract and amplify abnormal clues from continuous frames, respectively. A lightweight sample labeling method based on face landmarks is designed to label large-scale samples at a lower cost and has better results than other methods such as 3D camera. Finally, we collect 30,000 live and spoofing samples using various mobile ends to create a dataset that replicates various forms of attacks in a real-world setting. Extensive experiments on public OULU-NPU show that our algorithm is superior to the state of art and our solution has already been deployed in real-world systems servicing millions of users.