A Brief Review of Anti-Face Recognition

The face recognition system has achieved significant success. However, the excessive power of these systems poses a significant threat to the security and privacy of individuals. It has been revealed that deep learning models are susceptible to adversarial examples. Consequently, researchers have explored the utilization of adversarial examples to safeguard images from unauthorized face recognition systems. This paper aims to provide a comprehensive review of recent approaches that employ adversarial examples to safeguard the face privacy. Specifically, the fundamental concepts associated with adversarial attacks in the context of face recognition are introduced. Subsequently, the generation of adversarial face images, utilizing adversarial example techniques for privacy preservation in both digital and physical domains, is scrutinized. Finally, the challenges for further exploration are presented.