The Requirement Analysis of An Offline Automated Invigilation System with Gmail Alert Integration

Abstract

Examination malpractice remains a major concern for academic institutions, impacting on the fairness and credibility of evaluations. To address this, we analyze and propose an Offline Automated Invigilation System with Gmail Integration that leverages computer vision and machine learning to detect and prevent unethical behavior during offline exams. The system features three detection modules: YOLO for identifying mobile phones, Support Vector Machines (SVM) for tracking abnormal head movements, and Haar Cascade for real-time eye movement analysis. These technologies work together to monitor students, detect suspicious behavior, and capture evidence, which is then sent via Gmail alerts to examination authorities. Designed to operate without internet connectivity, the system ensures effective invigilation even in remote or resource-limited environments. By reducing human dependency and automating the detection process, this solution enhances accuracy, scalability, and integrity in offline examination settings.