A Lightweight AI-Based Access Control System for IoT-Integrated Smart Classrooms

Abstract

Innovative technologies are integrated into educational frameworks. This requires making access management systems efficient, secure, and protective of personal information. This article introduces an innovative classroom management system with IoT features. It adds simplicity and efficiency by employing edge AI facial recognition. Unlike RFID and cloud biometric systems, our method uses a quantized MobileNetV2 model on a Raspberry Pi 4, ensuring private, real-time identity verification through local processing. The system architecture comprises imaging, external face recognition, face feature image file description, and logic making, all integrated into a single dynamic computer system, allowing rapid offline operation. Test results confirmed the system's 96.8% recognition accuracy, 850 ms average response time, and low energy consumption. Compared with the conventional techniques, the proposed system provided better speed, affordability, flexible expansion, and security. This paper introduces an adaptable, economical, and privacy-focused approach for classroom environments in resource-limited settings. The findings justify the use of edge AI for secure access and attendance automation in educational institutions.