Implementation of Invigilation System using Face Detection and Face Recognition Techniques. A Case Study

Abstract

In recent years, face detection and face recognition techniques are improved sufficiently to make use in real-time applications and in crucial computer vision tasks. In this paper, the approaches that can use in real-time are discussed and implemented in a real-time application - Invigilation system. The web framework Django is used in designing the invigilation system and the database used is Mysql for storing the student and faculty data. The drawback of the traditional system is manual and failed to notify the wrong student attending the exam. This paper presents a method for automatic and optimised allotment into rooms and invigilators having face recognition for the student's correct prediction. Face detection is crucial for face recognition. To get the quick processing, an efficient, speed and accurate method was found by processing different images with various faces and found that the HOG method is best suited for processing a vast number of images. The face recognition model used in this paper has an accuracy of 99.38%, which is sufficient for proper identification. The cameras placed in the rooms can take the pictures and send them to the Django server. The server processes the images, the face detector extracts the faces from the image and the face recognizer compares them with the faces of the allotted from the database. The whole system can identify the wrong person and able to find the attendees list. In future, it can develop in identifying the malpractices by implementing the tracking algorithms.