Transfer Learning: Convolutional Neural Network-AlexNet Achieving Face Recognition

Abstract

Nowadays, Machine-based face recognition is becoming very commonplace, robust and dependable system that is widely employed in numerous cases for access control. As in traditional approach, face recognition needs the extraction of face features prior to classification and recognition, which affects recognition rate. We employ Face Verification checks whether the pictures are associated with a single individual, whereas Face Identification must identify a specific face from a collection of known profiles in the system. To tackle this question, this paper incorporates the CNN structure Alexnet to obtain face identification.Throughout this article, we perform facial recognition using transfer learning in a Siamese network composed of 2 comparable CNNs. A pair of 2 face picture is fed into the Siamese network as input, after which the network learns the traits of this pair of pictures.Next the network is trained using the PRelu activation function to find the ideal learning algorithm and maximal values. Then, the face was identified and categorized. Library Multi-Spectral Face Data - set and Library 2D Faceprint Database were used to test the methodology, it enhances the accuracy of face recognition when compared to algorithms trained on datasets with a particular dataset and a specific spectrum’s recognition rate peaked up to 98 percent.