As a result of the Coronavirus pandemic, many of us are working or learning at home. This is true for school students as well as college students. Almost all the universities in India and abroad assess students through online tests and there are various ways in which students can cheat these tests. In order to ensure quality and integrity in the present scenario, efficient online testing is essential.By considering various parameters, we propose a system for monitoring the attentiveness of candidates, using an artificial intelligence model. The proposed scheme tracks head movements in two directions (left, right, up, and down), including mouth movements, which can be converted into a trust score based on predefined threshold values selected by the authority who conducts the assessment. The proposed scheme can be evaluated experimentally on video samples recorded for this purpose. The thresholds for the two parameters can be adjusted independently, avoiding false results for automation tool for correction of answers scripts We take a average of highest similarity scores attained by every line in student answer with that if teacher answer and call it semantic score. Then we take an IOU of important phrases in student and teacher answer and call this score the syntactic score. Now we consider a weighted average of the semantic and syntactic scores as the final score attained by a student.

Share this project:

Updates