Integration Of Face Recognition Model to a Biometric Security System

Abstract

With the global COVID-19 pandemic at large, physical contact is supposed to be minimized. This raises the need for a non-invasive security system that could differentiate between individual human beings. Biometrics is the automatic recognition of an individual using certain distinguishing traits. It proves to be a solution for security systems in such times. However, computers cannot see objects as humans do, they see objects as binaries which requires further discoveries to build a system that is accurate enough to be implemented in a security system. Viola and Jones did research on such methods. They had found that computers are able to see patterns, but it requires a classifier which is a model to help computers understand certain features that makes a human face. Viola and Jones used what they call as the cascade classifiers with Haar features used by the computer to analyze an object and helps it determine whether the object is face or not. This research aims to create a multi-class classification (face or non-face and face in the database or not) facial recognition system with the help of machine learning to create a model which will implemented in the access systems. The OpenCV libraries are used in this research as there lots of facial recognition and processing functions which will render the research for designing machine learning based access systems easier. The research shows that the model formed has an accuracy of 74.8%, a precision of 74.02%, and a sensitivity of 100%.