A method combining multiple perspectives to enhance accuracy in facial recognition problems

Abstract

This article introduces an advanced method in the field of facial recognition, using a unique technique that combines Convolutional Neural Networks (CNN) and Multilayer Perceptron (MLP) to integrate different perspectives. The highlight of this method is the application of CNN to analyze image features from multiple angles, along with MLP, to optimize the information synthesis process, thereby enhancing the accuracy of facial recognition under varying lighting conditions and angles. The main goal is to address the challenge of performance degradation in facial recognition in real-world situations, especially when there is a significant change in the viewpoint. This study details the model-building process from data collection and processing, training complex neural networks, and evaluating effectiveness through standard and experimental datasets.