Detect cheating during assessment on NTU E-learning at Nha Trang University by using computer vision

Abstract

Detecting and preventing cheating are essential for upholding academic integrity in educational institutions and online learning platforms. Whether using Moodle or any other eLearning or LMS platforms, the task of ensuring honest academic practices has grown increasingly complicated. Detecting cheating by opening another tab or window during an assessment on E-learning platforms needs to be addressed. In this research, The authors will build a dataset includes test screenshots in normal and cheating cases. The authors propose a cheating detection strategy based on computer vision, specifically YOLOv8. The goal of this strategy is to detect behaviors such as opening another tab or window during assessments on the Moodle platform. This will help ensure fairness and reduce the time and effort required by proctors. Our research results have demonstrated high accuracy and feasibility.