Efficient and Privacy-preserving Distributed Face Recognition Scheme via FaceNet
Xiaoyu Kou, Ziling Zhang, Yuelei Zhang, Linlin Li · 2021
In recent years, with the development of deep learning techniques, face recognition has draw numerous attention in both academic and industrial. Meanwhile, it is also widely deployed in smart home and brings great conveniences in people’s life. However, due to the sensitivity of biometric data, face recognition is still confronted with several crucial challenges including face feature data disclosure. In this paper, based on random matrix, BLS short signature and FaceNet, we propose an efficient and privacy-preserving face recognition scheme for smart home. Specifically, the scheme includes two main algorithms: face templates encryption algorithm and privacy-preserving similarity computation algorithm. With the proposed two algorithms, face recognition is achieved without revealing face feature data. Security analysis proves that the face feature data is well protected. Moreover, extensive experiments are carried out with LFW dataset, and the experiment results demonstrate that our scheme is indeed efficient and precise.