Design and implementation of an encrypted facial recognition attendance management system

Abstract

Attendance management is of paramount importance in organisational, academic, and industrial contexts because it ensures an accurate record of presence and participation. Traditional attendance management methods are often inefficient and susceptible to manipulation, leading to the adoption of biometric solutions. While facial recognition offers a contactless and efficient alternative, the storage of unencrypted biometric data poses significant privacy and security risks. This study presents the design and implementation of an Encrypted Facial Recognition Attendance Management System that integrates deep learning-based identification with robust cryptographic protection. The system utilises a DeepFace CNN architecture for feature extraction and implements AES-256 encryption to secure biometric templates during storage and transmission. A dataset of 25 cadets was enrolled, and attendance was monitored across six different academic courses. Experimental results demonstrate a recognition accuracy of 95.0%, with a False Acceptance Rate (FAR) of 2.67% and a False Rejection Rate (FRR) of 7.33% at an optimal Euclidean threshold of 7.0. The system achieved an average processing latency of 26.9 ms, with an encryption overhead of approximately 8 ms, ensuring real-time performance. Comparison with recent literature confirms that the proposed system provides a superior balance between computational efficiency and high-level data privacy.