Comparative Study of Liveness Detection Methods for Mitigating Facial Spoofing Attacks

V. Christy, H. Chandramouli Department of Computer Science and Engineering, East Point College of Engineering and Technology, Visvesvaraya Technological University, Bengaluru, Karnataka, India.

Abstract: Despite the success of FRS in security-sensitive scenarios, they are still susceptible to presentation attacks, including printed photos, replayed video and 3D masks, which require a strong liveness detection system. This study categorises approaches to face liveness detection into deep learning-based methods and provides a technical review and comparative analysis. Texture-based methods mainly rely on Convolutional Neural Networks (CNNs) to detect spatial spoofing artefacts. In contrast, temporal-based methods rely on Recurrent Neural Networks (RNNs) and 3D convolutional neural networks (3D-CNNs) to capture dynamic facial motion patterns. The hybrid framework combines spatial and temporal representations to enhance detection robustness. To evaluate the performance, the standard databases Replay-Attack, CASIA-MFSD, and OULU-NPU are used, and the standard metrics Attack Presentation Classification Error Rate (APCER), Bonafide Presentation Classification Error Rate (BPCER), and Average Classification Error Rate (ACER) are used. Comparative results show that texture-based methods achieve ACERs below 3% in controlled intra-dataset tests. In comparison, hybrid methods consistently achieve good performance in cross-dataset generalisation (ACER < 2%, APCER < 2.5%, and BPCER < 2%). Temporal-based approaches exhibit greater resistance to replay attacks but incur higher computational complexity. The analysis demonstrates that hybrid deep learning architectures offer the best balance of detection accuracy, attack robustness, and generalisation, making them a powerful choice for designing robust liveness detection systems for real-world biometric security applications.

---

Keywords: Facial Anti-Spoofing; Liveness Detection; Presentation Attack Detection; Comparative Analysis; Deep Learning (DL); Biometric Security; Facial Recognition Systems (FRS).

---

Received on: 19/07/2025Revised on: 28/09/2025Accepted on: 29/10/2025Published on: 03/03/2026

---

---