РАЗРАБОТКА МЕТОДИКИ ОПРЕДЕЛЕНИЯ РАЦИОНАЛЬНОГО ВЫБОРА ДЛИТЕЛЬНОСТИ РАЗРЕШАЮЩЕГО СИГНАЛА СВЕТОФОРНОГО РЕГУЛИРОВАНИЯ НА ОСНОВАНИИ НЕЙРОННОЙ СЕТИ

Abstract

The article discusses a method for recognizing a person's face based on convolutional neural networks to develop a method for determining the rational choice of the duration of the permissive traffic light signal based on a neural network. The advantage of using convolutional neural networks is that they provide robustness to changes in face recognition conditions, as well as the ability to process a large database. The topology of the used convolutional neural network and the methodology for its training are substantiated.A fragment of a description of the program code is given that adjusts the duration of the permissive traffic light signal for pedestrian flows in real time. Methods are indicated to increase the recognition accuracy of a convolutional neural network.