Face liveness detection using deep learning
Abstract
In the modern digital era as present, facial recognition is used more commonly than ever in numerous sectors of technology and science. Compared to other biometric identification techniques such as fingerprints, iris scans, etc., facial recognition systems have reduced many related issues and are more effective for human identification. With the distribution of face recognition, it now gives considerable attention on information security and system security. However, some gaps remain in the “appropriate recognition” that should be addressed, including the both subjective and objective reasons like weather or even the transmission of the detectors. In this paper, we suggest a deep neural network diagram for anti-face spoofing and liveness detection by adopting a Convolutional Neural Network (CNN) divided into feature extraction and classification stages, in which the dataset used is CelebA Spoof (2020), collected for direct and indirect face recognition. Experiments were conducted on a subset of the CelebA-Spoof dataset. The research experiment showed that the model achieved an average accuracy of 87%. With this result, the study provides a method that can improve the performance of the facial recognition technology.
Tóm tắt
Trong thời đại kỹ thuật số hiện nay, việc sử dụng nhận diện khuôn mặt đã trở nên phổ biến hơn bao giờ hết trong nhiều lĩnh vực khoa học công nghệ. Điều này đã làm giảm bớt nhiều vấn đề liên quan và hiệu quả hơn về nhận dạng người. Tuy nhiên, vẫn còn một số lỗ hổng liên quan đến vấn đề về an ninh thông tin, bảo mật trong hệ thống nhận diện khuôn mặt. Bài báo này đề xuất phương pháp chống giả mạo khuôn mặt và phát hiện sự sống dựa trên việc phát triển mô hình CNN (Mạng nơ-ron tích chập) sử dụng tập dữ liệu CelebA Spoof. Kết quả thực nghiệm của mô hình đề xuất đạt độ chính xác trung bình là 87%. Điều này có thể góp phần vào việc cải thiện hiệu suất của công nghệ nhận dạng khuôn mặt.
References
[1] D. Wen, H. Han, and A. K. Jain, “Face spoof detection with image distortion analysis,” IEEE Transactions on Information Forensics and Security, vol. 10, no. 4, pp. 746–761, Apr. 2015, doi: 10.1109/TIFS.2015.2400395.
[2] S. Kumar, S. Singh, and J. Kumar, “A comparative study on face spoofing attacks,” in Proceeding - IEEE International Conference on Computing, Communication and Automation, ICCCA 2017, Institute of Electrical and Electronics Engineers Inc., Dec. 2017, pp. 1104–1108. doi: 10.1109/CCAA.2017.8229961.
[3] Y. Chen, H. Jiang, C. Li, X. Jia, S. Member, and P. Ghamisi, “IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING 1 Deep Feature Extraction and Classification of Hyperspectral Images Based on Convolutional Neural Networks.”
[4] Y. Akbulut, A. Şengür, Ü. Budak, and S. Ekici, “Deep learning based face liveness detection in videos,” in IDAP 2017 - International Artificial Intelligence and Data Processing Symposium, Institute of Electrical and Electronics Engineers Inc., Oct. 2017. doi: 10.1109/IDAP.2017.8090202.
[5] Z. Boulkenafet, J. Komulainen, and A. Hadid, “Face Spoofing Detection Using Colour Texture Analysis,” IEEE Transactions on Information Forensics and Security, vol. 11, no. 8, pp. 1818–1830, Aug. 2016, doi: 10.1109/TIFS.2016.2555286.
[6] D. Wen, H. Han, and A. K. Jain, “Face spoof detection with image distortion analysis,” IEEE Transactions on Information Forensics and Security, vol. 10, no. 4, pp. 746–761, Apr. 2015, doi: 10.1109/TIFS.2015.2400395.
[7] Y. Liu, J. Stehouwer, A. Jourabloo, and X. Liu, “Deep Tree Learning for Zero-shot Face Anti-Spoofing.”
[8] M. Asia-Pacific Signal and Information Processing Association. Annual Summit and Conference (9th : 2017 : Kuala Lumpur, Asia-Pacific Signal and Information Processing Association, and Institute of Electrical and Electronics Engineers., Ninth Asia-Pacific Signal and Information Processing Association Annual Summit and Conference : APSIPA ASC 2017 : proceedings : 12-15 December 2017, Kuala Lumpur, Malaysia.
[9] A. Da Silva Pinto, H. Pedrini, W. Schwartz, and A. Rocha, “Video-based face spoofing detection through visual rhythm analysis,” in Brazilian Symposium of Computer Graphic and Image Processing, 2012, pp. 221–228. doi: 10.1109/SIBGRAPI.2012.38.
[10] Y. Zhang et al., “CelebA-Spoof: Large-Scale Face Anti-Spoofing Dataset with Rich Annotations,” Jul. 2020, [Online]. Available: http://arxiv.org/abs/2007.12342
[11] “The meaning and use of the area under a receiver operating characteristic (roc) curve”.
