Moodle plugin for real-time face recognition in online exams for Systems Engineering and Computer Science students at Universidad Continental, Huancayo, Peru

Abstract

Facial recognition systems on learning platforms like Moodle enable fast and efficient evaluation, automating the detection of impersonation in online exams. This project implements a Moodle plugin that performs facial recognition on the student attempting to take a quiz. It consists of two parts for authentication through a webcam: the first part occurs before accessing the quiz, and the second part involves surveillance through a camera during the quiz. The plugin was developed using the PHP and JavaScript programming languages, as well as Face-api. The JavaScript API detects and recognizes faces in the web browser using Tiny Yolo v2 and FaceRecognizerNet. It compares a student's photo or image with the image from their webcam while they are taking the quiz. The Moodle plugin achieved an average error probability of 39.39% for face detection similarity distance.