Published Paper Details:

The project's aim is to create a secure and user-friendly environment for students who don't have access to a campus to take tests online from a variety of places. The research uses a variety of machine learning models to relieve the administrative load of detecting student misconduct on a wide scale. The model verifies and authenticates the applicants taking the test, and it checks for any irregularities on a regular basis. This is achieved by identifying video and audio that the model analyses in order to verify the student's honesty. The system's continuous analysis of the inputs allows it to verify the candidate's honesty, guaranteeing academic integrity in e-learning. From the viewpoint of the test taker, the system is both economical and simple to use, as it simply involves the usage of inexpensive web cameras and a microphone. User verification, audio processing, gaze detection, number of person detection, item and phone detection are the five main components that continually evaluate the essential behaviour clues. We create higher-level features to classify if the test taker is cheating at any point throughout the exam by merging the continuous estimate components and using a temporal sliding window.