Skip to content
Review Open access

Secure and Privacy-Preserving Lightweight Face Recognition: A Comprehensive Review of Architectures, Threats, and Future Directions

2026 · International research journal of innovations in engineering and technology · 0 citations

Abstract

Facial Recognition Systems (FRS) are fast being deployed on edge devices, intelligent surveillance and real-time authentication platforms. But this growth is compounding critical privacy concerns, biometric data breaches and susceptibility to adversarial manipulation. New FRS designs are confronted with a intricate trilemma of preserving privacy, lowering computational expenses, and enhancing recognition accuracy. This review paper summarizes the recent developments (2020-2026) of robust, lightweight and secure face recognition architectures with privacy preservation. We study cutting-edge loss functions and quantization techniques aimed at scaling recognition models to millions of identities and on limited hardware. In addition, we explore various visual privacy mechanisms from the hardware level that is, using optics, the cryptographic enclave, thumbnail-preserving encryption and decentralized federated learning paradigms. In addition, we assess the biometric security risks like template inversion attacks, cancelable biometrics and life-cycle adversarial defense. Lastly, we discuss new frontiers in the field of security, such as safety in Large Vision-Language Models (LVLMs) and dynamic user-revocable privacy paradigms. This paper offers researchers and practitioners an integrated roadmap towards the development of resilient biometric vision systems that are compliant with privacy requirements.

Read PDF

We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.