Face Recognition Using Convolutional Neural Network

One of the important areas of computer vision is the face recognition (FR) issue. FR is used to detect faces seen on dispersed cameras throughout a network. There are two categories of face recognition problems. The first category, called the face identification issue involves recognizing faces based on samples, per person. One method of face recognition known as Single Sample Per Person (SSPP) is the type commonly referred to as "current face recognition approaches." There is research that supports both methods and deep learning techniques. This research paper introduces a CNN model that aims to expedite network convergence and minimize the duration of training. As a step we incorporate a normalization layer, into the CNN framework. To accommodate the normalized data, we include an activation function, within the existing activation function allowing for modifications. Finally, we utilize max pooling to ensure that the feature information remains stable and intact while still preserving its details. The experimental findings demonstrate that our approach outperforms the leading technique and addresses the issue of results, between the training and test datasets resulting in a lower rate of recognition.