SIXTHSENS AI

Home/E-Exam Proctoring System

E-Exam Proctoring System

Objectives:


Develop a real-time proctoring system to replace human proctoring during tests. Establish a remote candidate login system for online exams. Detect identity fraud and analyze cheating behavior. Monitor candidates through live streaming. Implement post-proctoring analysis, including face detection status, cheat estimation, user authentication, and log analysis.



Current Features:


The Proctor Software, powered by a combination of Mediapipe, Inception Resnet V2, and Django, ensures a secure online exam environment with the following features:

  1. Face Detection and Recognition: Utilizing Mediapipe for face detection and Inception Resnet V2 for face recognition, the system verifies users' identities during exams. It detects instances of no face or multiple faces, indicating potential cheating.

  2. Cheat Detection: Matplotlib generates estimated cheat reports based on users' cheat statuses during the exam.

  3. Online Exam Management: Django facilitates an online exam management interface where students log in for exams, authenticated by the face recognition model. An admin panel is provided for exam and student management, along with viewing cheat reports.



Additional Features:


Following the initial paper publication, the Proctor Software gained advanced capabilities:

  1. Eye Movement Tracking: Utilizing Mediapipe, the system tracks students' eye movements to detect when they look away from the screen or at another part, potentially indicating cheating behaviors.

  2. Mouth Detection: Mediapipe detects mouth movements to identify instances of talking or whispering during the exam, helping to uncover cheating incidents.

  3. Head Pose Estimation: Mediapipe estimates students' head poses to detect when they divert their gaze from the screen, providing insights into potential cheating actions.

  4. Face Spoofing Detection: By employing OpenCV, the system identifies face spoofing attempts, such as using fake faces or photographs instead of real faces.



These enhancements fortify the Proctor Software's capabilities in maintaining exam integrity, safeguarding against fraudulent behaviors, and ensuring a fair testing environment.



Access the Published Paper Here: Published Paper Link

Model Type : Video


Industry Sector: Education


For Proctor Demo Please Reach Out to shubham@sixthsens.ai











Reach out to us