Discriminative Feature Learning framework for Face Recognition using Deep Convolution Neural Network

Abstract

Face Recognition in today’s era became a widespread application in biometric technology and fo-rensic sciences to identify the individual against the crime involved in the scene. A biometric is an au-thentication framework that perceives an individual by validation of a physiological and behavior character owned by an individual. The only thought-provoking physiological biometric technique used in non-physical contact mode for recognition is the Face Recognition Technique. Algorithms implementing such practices can extract features from facial images and map these features in contradiction to the known database subjects information. It then compares to find out a match that can aid to verify individual identity. Face recognition challenges contain visual interaction issues like Pose, Illumination, Expression, Aging, Occlusions of the face. In this paper, a robust face recognition algorithm is proposed to overcome the problems mentioned above by extracting facial features using an Alex-net-based deep learning neural network. Detection rate and Recognition accuracy using this are achieved to 98.3% and 98%, respectively. Accordingly, Alex-net became the most effective discriminating technique with a high recognition rate in Deep Convolution Neural Network.