Comparative Analysis of Facial Recognition Models to Record Classroom attendance with one click

Abstract

Human face recognition is the hottest topic in this era due to its importance in identity authentication and security checking. It has grown a momentous spot among all biometric-based systems. It can be used for verification and scrutiny to prove the identity of a person and sense individuals, singly. Great success has been achieved recently on face recognition comparison by using different models. But in this research, the HOG (Histogram Oriented Gradient) and Vgg (Visual Geometry Group) algorithms are selected as these algorithms are common and still widely accessible on the internet, and these are usually used out of the box for various applications. In this paper, we examine the accuracy and time efficiency between the HOG (Histogram Oriented Gradient) and machine learning model. Also, this paper conferred these pre-trained models and outlines which results in high-tech performance without the need for any prevailing hardware. Their features are observed and the result of each model is estimated. Their comparative analysis of the output of both of these models with the execution time and the number of repetitions of the practical algorithms is offered. It also constructed two simple applications for face recognition-based algorithms. By using these application testing data is taken from the live stream/camera in the form of image and compared through training images stored in the faces database and their time and accuracy are noted during this process and their performance is evaluated based on time and accuracy.