An Automated Cheating Detection System for Online Exams

 

Executive Summary

 

The growth of online courses offered for credit heavily depends on innovations in the online proctoring industry to provide accurate, yet flexible remote examination.  Current software-based online proctoring systems do not sufficiently monitor and inhibit cheating behavior in high stake online exams as they only rely on the narrow field of vision from the test taker’s webcam.  Our new tested and patented system for automated detection of cheating behavior in online exams uses an additional small camera attached to the test taker’s headset, which enables the system to provide an unmatched level of cheating detection security and quality of proof – even for high stake online exams.

 

Description of Technology

 

To capture audio-visual cues of the exam environment and the online test taker, the MSU multimedia analytics system uses three sensors – a webcam, a wearcam that can be integrated into a test taker’s head set, and a microphone.  The wearcam enables the system to capture cues from the test taker’s field of vision. The software processes the sensed data using six components to detect cheating behavior: User verification, text detection, speech detection, gaze estimation, phone detection, and active window detection.  Depending on the exam’s context (open book etc.) the components can be flexibly reconfigured.  Testing data shows that the cheating detection rate of the MSU automated online exam proctoring system is as good as or better than the judgement of a qualified human online proctor. 

 

Key Benefits

  • Cost-effective and practical for high quality monitoring of online exams
  • Scalability: Scale up the number of simultaneously proctored online test takers as desired
  • High flexibility for online test takers
  • Enables a wide range of online proctoring services: Can be used to assist real people online proctors to increase their efficiency or to offer a low-cost automated proctoring service (e.g. record, flag, and review)
  • Excellent cheating detection rate

 

Applications

  • Online exam proctoring for online courses offered by universities and companies

 

Patent Status: 

 

US Patent No. 9,154,748

 

Licensing Rights Available

 

Full licensing rights available

 

Inventors: Stephen Hsu, Xiaoming Liu, Xiang-Yang (Alex) Liu

 

Tech ID: TEC2013-0052

 

Patent Information:

For Information, Contact:

Raymond DeVito
Technology Manager
Michigan State University
devitora@msu.edu
Keywords: