Development of a Procedure for Improving the Effectiveness of Personal Identification Systems Based on the Analysis of Facial Images

This research aims to develop a new procedure to improve the performance of the personal identification system. This process will be based on the analysis of digital facial images using special artificial intelligence technologies. Additionally, it is emphasized that the successful development of a completely new neural network algorithm for personality identification and its special robust modification are required to achieve this goal. Furthermore, it is essential that this algorithm demonstrates increased effectiveness in identifying faces. As part of this algorithm, two key components are distinguished, namely, an algorithm aimed at determining the area where a face is located in an image, which involves the use of a standard version of a convolutional neural network. And the presented neural network algorithm is the second component, which is designed for personal identification with a specific robust modification. The studies presented have shown that an algorithm with a proposed architecture, will have around 25 times less weight parameters than the algorithms utilized at present. Also, unlike other algorithms, it was possible to improve the level of accuracy.