Review of Masked Face Recognition Based on Deep Learning

Abstract

With the widespread adoption of face masks due to global health crises and heightened security concerns, traditional face recognition systems have struggled to maintain accuracy, prompting significant research into masked face recognition (MFR). Although various models have been proposed, a comprehensive and systematic understanding of recent deep learning (DL)-based approaches remains limited. This paper addresses this research gap by providing an extensive review and comparative analysis of state-of-the-art MFR techniques. We focus on DL-based methods due to their superior performance in real-world scenarios, discussing key architectures, feature extraction strategies, datasets, and evaluation metrics. This paper also introduces a structured methodology for selecting and reviewing relevant works, ensuring transparency and reproducibility. As a contribution, we present a detailed taxonomy of MFR approaches, highlight current challenges, and suggest potential future research directions. This survey serves as a valuable resource for researchers and practitioners seeking to advance the field of robust facial recognition in masked conditions.