Performance Analysis of Machine Learning Models for Cheating Detection in Online Examinations

The use of online meeting platforms became common due to worldwide pandemic emergency preparedness. These systems are used by the majority of businesses and academic institutions for online assessments. However, it raises the issue of adequacy in the procedure. These assessments serve as a catalyst and provide learners with constructive feedback, which motivates them to do better. But because of the pandemic, quality evaluation is crucial using these online meeting platforms. Internet meetings now require the incorporation of anti-cheating analysis to elevate the process of quality evaluation. We created the components of such an anti-cheating system and established the ideal pipeline. We evaluated this pipeline and its features based on a collection of private examination footage. The cheating state of the overall video and the presence of specific cheating occurrences are used to label the dataset. The suggested strategy yields a reasonably high level of accuracy in identifying instances of examination cheating, according to numerical studies on our dataset.