Deep Learning Models for Robust Facial Liveness Detection

View on arXiv ← Back to list

Authors: Oleksandr Kuznetsov, Emanuele Frontoni, Luca Romeo, Riccardo Rosati, Andrea Maranesi, Alessandro Muscatello

Published: 2025-08-12 17:19:20+00:00

AI Summary

This research introduces novel deep learning models for robust facial liveness detection, addressing deficiencies in current anti-spoofing techniques. By integrating texture analysis and reflective properties, the models distinguish authentic faces from replicas with high precision, achieving 99.9% average accuracy on a combined dataset.

Abstract

In the rapidly evolving landscape of digital security, biometric authentication systems, particularly facial recognition, have emerged as integral components of various security protocols. However, the reliability of these systems is compromised by sophisticated spoofing attacks, where imposters gain unauthorized access by falsifying biometric traits. Current literature reveals a concerning gap: existing liveness detection methodologies - designed to counteract these breaches - fall short against advanced spoofing tactics employing deepfakes and other artificial intelligence-driven manipulations. This study introduces a robust solution through novel deep learning models addressing the deficiencies in contemporary anti-spoofing techniques. By innovatively integrating texture analysis and reflective properties associated with genuine human traits, our models distinguish authentic presence from replicas with remarkable precision. Extensive evaluations were conducted across five diverse datasets, encompassing a wide range of attack vectors and environmental conditions. Results demonstrate substantial advancement over existing systems, with our best model (AttackNet V2.2) achieving 99.9% average accuracy when trained on combined data. Moreover, our research unveils critical insights into the behavioral patterns of impostor attacks, contributing to a more nuanced understanding of their evolving nature. The implications are profound: our models do not merely fortify the authentication processes but also instill confidence in biometric systems across various sectors reliant on secure access.

Key findings

The best performing model (AttackNet V2.2) achieved 99.9% average accuracy when trained on combined data. Cross-dataset performance significantly improved with combined training, demonstrating robust generalization. The study found no statistically significant difference in performance between the various CNN architectures tested.

Approach

The authors propose several Convolutional Neural Network (CNN) architectures (LivenessNet and AttackNet variants), incorporating skip connections and different activation functions to improve feature extraction and robustness. They utilize a combined dataset training strategy to enhance generalization across various spoofing attack types and environmental conditions.

Datasets

3DMAD, Replay-Attack, MSSpoof, CSMAD, and a custom dataset.

Model(s)

LivenessNet, AttackNet V1, AttackNet V2.1, AttackNet V2.2

Author countries

Italy, Ukraine

← Previous