A Privacy-Preserving Gait Recognition Scheme Under Homomorphic Encryption

Abstract

In recent years, machine learning and deep neural networks have achieved remarkable results and have been widely used in different domains. Affected by COVID-19, the potential of gait feature recognition in biometric authentication has gradually emerged. However, machine learning algorithms are generally demanded in terms of computing power, sometimes need the support of cloud service providers, and require raw data, which is often sensitive, most privacy-preserving approaches only encrypted the trained model, and the data collected from users are unprotected. We propose a scheme for running deep neural networks on encrypted data using homomorphic encryption to address these issues.