Face RF: A Secure, Dual-Factor Automated Attendance System with Server-Side Liveness Detection

Abstract

Traditional attendance management systems in educational institutions are often plagued by inefficiencies, data errors, and susceptibility to fraudulent practices such as "proxy" attendance. To address these challenges, this paper introduces FaceRF, a secure, dual-factor automated attendance system that integrates Radio Frequency Identification (RFID) with advanced computer vision techniques. The system utilizes a custom ESP32-based hardware module to perform the initial user identification via NFC cards, followed immediately by a biometric scan using an ESP32-CAM. On the server side, the system employs a Deep Residual Network (ResNet-34) to generate 128-dimensional face encodings for high-precision identity verification. Crucially, to mitigate the vulnerability of standard face recognition to presentation attacks (spoofing), a novel server-side Liveness Detection algorithm is implemented. This algorithm utilizes dlib’s 68-point facial landmark predictor to calculate the Eye Aspect Ratio (EAR) across a burst of captured frames, effectively distinguishing between live subjects and static photographs. Experimental results demonstrate that FaceRF achieves a 0% False Acceptance Rate (FAR) and successfully rejects 95% of static photo spoofing attempts, offering a highly accurate, fast, and secure alternative to conventional attendance methods.