Video Interpretation for Cost-Effective Remote Proctoring to Prevent Cheating

Abstract

In the rising era of globalization and digitization, remote education continues in gaining popularity and reach. Efficiently proctoring online remote examination is an important limiting factor to sustain the integrity of the exam as well as provide unprejudiced results. Currently human proctoring is the customer perspective to maintain integrity, either manually with the help of a test taker or by overseeing them visually through webcams. Online exams provide the examiner the choice, to choose the environment and the tools they wish to use during the exam. In response to this, our research proposes an application to detect fraudulent activities during online examination in real-time through the video recorded by the webcam of the examiner’s system. The application provides four features that continuously estimate the integrity of the exam: (1) User verification for checking impersonation by the examiner. (2) Multiple people together solving the exam. (3) Absence of examiner. (4) Detecting the use of mobile phones. The extensive experiment depicts accuracy of our cost-effective remote proctoring.