Real Time Cheating Detection Pipeline for Online Exams
Ahmet Dadak, Ufuk Uyan, Mahiye Uluyağmur Öztürk · 2022 30th Signal Processing and Communications Applications Conference (SIU) · 2022
Remote exams and job interviews have gained popularity due to pandemics and the advantage of remote working conditions. Most companies and institutions use online platforms for recruitment and examination. However, one of the critical problems of remote exam systems is that the exams cannot be held in a reliable environment. In this work, we present a cheating detection system for online exams. This system provides real-time detection of suspicious events such as another person, electronic device usage, candidate absent status by accessing the microphone and camera stream of the candidate’s device through the browser. The cheating detection system can detect other people than candidates, electronic device usage, voice activity, and candidate absence status using state-of-the-art deep learning algorithms. When a suspicious event is detected, its moments are recorded as a video. The advantages of the real time cheating detection pipeline are as follows: 1) The system provides real time suspicious activity detection and produce instant alarm for proctors, 2) the system does not require an external program installation, 3) the number of user the system will serve at the same time has become scalable since it is run through the browser.