Secure Examination Hall Authentication Using HOG Feature Extraction and SVM-Based Robust Face Detection
Abstract
With the rapid increase in student populations, conventional examination hall verification methods such as manual identity checks, admit cards, and handwritten attendance have become inefficient, time-consuming, and prone to impersonation and human error. To address these challenges, this paper proposes an advanced, automated exam hall authentication system that integrates machine learning- based facial recognition with QR code verification. The system is designed to provide secure, contactless, and real-time student authentication using lightweight embedded hardware. A Raspberry Pi serves as the central processing unit, hosting a Flask-based web server, while OpenCV and Mediapipe are employed for face detection and recognition. During registration, student facial data are collected and encoded to create a unique digital identity. On the examination day, students scan a QR code to retrieve their room information and undergo realtime facial verification at the exam hall entrance. Upon successful authentication, seat allocation and attendance are recorded automatically. Experimental results demonstrate high accuracy, low false-match rates, and fast inference suitable for real-world deployment. The proposed system reduces crowding, minimizes administrative workload, and enhances examination security while remaining cost- effective and scalable. This approach offers a practical solution for modernizing examination management systems in educational institutions.