Methods of the development of the architecture of the neural networks for identification and authentication of individuals

Abstract

This paper deals with the neural network methods of the implementation of systems of identification ofindividuals based on videos and photographs. Over the last few decades, it has been considered to be oneof the most powerful tools and has become very popular in the literature as it is able to handle a hugeamount of data. The neural network architectures used in modern biometric identification systems havebeen reviewed. Based on the research conducted in this field, an approach was developed that can improvethe accuracy of object recognition in photo and video images by increasing the quality of the attributes ofthe weights and reducing the number of the weights, as well as the number of the connections. The basis ofthe developed neural network model is a multilayer perceptron; the main system is a convolutional neuralnetwork. The neural network model has been implemented using the Python programming language withthe most popular machine learning libraries Keras and TensorFlow. In addition, we will also enumerate theparameters that affect CNN efficiency.

Keywords

• biometric identification of individuals
• three-dimensional object recognition
• convolutional neural network
• deep machine learning