Online Assessment and Proctoring for Candidate Recruitment at Workplace

Abstract

In the present world, human proctoring is the most important approach to evaluation, by either providing the supervisor to visit an examination center or supervising them observably and acoustically while exam using a webcam; however, such methods are manual and costly. So, we aim to develop an automated proctoring system to detect an extensive variety of cheating behavior during an online assessment exam. The system requirement includes a working webcam and mic. This system has four components. User verification, eye-tracking, head-pose-estimation, and person and object detection. Person and object detection is done using the yolo algorithm and eye-tracking using blob algorithm and cascade classification. To develop our purposed system, we collect required data from 10 subjects who also performed various types of cheating while examination/assessment. Vast experimental outcomes indicate our proctoring system's accuracy, validity, and proficiency. By using this system for recruitment, we can increase the chances of getting deserving candidates to the organization.