Enhancing Exam Security AI-Powered Safe Exam System with Secure Browser Control

Abstract

The modern world is evolving and connecting to computers on a daily basis. In recent years, e-learning has grown significantly. Maintaining the integrity of exams in the context of online learning is a major challenge that calls for creative ways to prevent cheating and protect academic justice. This problem is addressed by three different methods: First, a Smart computer vision-based system proposes automatic video summarization of abnormal behavior during online exams, aiding remote proctors in post-exam reviews. By modeling normal and abnormal student behavior patterns, this method offers promising results, potentially expanding to real-time alert generation. Second, a survey-based study examines the impact of online webcam exam proctoring on student anxiety and performance, particularly among non-white and socioeconomically diverse populations. While anxiety over being wrongly flagged exists, it doesn't directly impede exam performance, highlighting the need for nuanced support for students and faculty navigating unfamiliar technologies. Lastly, a continuous authentication system for online exams is proposed, leveraging machine learning algorithms to detect and prevent fraud through modules like registration, identity verification, live video streams, and session recording. These approaches collectively underscore the critical importance of maintaining integrity, reliability, and security in online examinations, especially amidst the unprecedented challenges posed by events like the COVID-19 pandemic..