Face Detection and Recognition Techniques Analysis

Abstract

The face is one of the foremost widely physical attributes used in biometrics systems for human identification. Biometric authentication has gained a lot of interest and attention recently. The field of biometrics can range from complicated tasks such as video police investigations to simple ones such as access to social media accounts or unlocking mobile phones. There are several face recognition and detection algorithms. They are based mostly on two types of computer vision schools. The first type of school is the traditional techniques which use descriptive hand-designed features to capture the characteristics and then machine learning strategies to learn classifiers. On the other hand, deep learning algorithms perform the separated two tasks in one stage where the network can learn the features and the classifiers. The face verification technique contains two stages, the first stage is face detection, and the second stage is face recognition. Face detection addresses the issue of the face's localization within the entire image to determine the bounding box of individual faces. Face recognition measures the similarity of the localized faces in the image with stored templates in a particular database to identify the personality of the detected faces. This paper provides a comprehensive study and analysis of face detection and recognition techniques. Each stage is analyzed by studying the suggested algorithms in the literature and providing strengths and weaknesses. Additionally, we discuss the challenges and the problems which make the detection and recognition tasks very complicated.