Industrial Automatic Control Systems and Controllers Annotation << Back Method of Detecting and Classifying Vehicles on Ultra-high Resolution Satellite Images V.S. Tormozov The article is devoted to the development of a method for detecting and classifying vehicles on satellite images of ultrahigh spatial resolution. As an input data method are used: a set of satellite images of ultra-high resolution, geographic information about the location and width of roads on the ground. It is planned to substantiate the choice of ultra-high resolution satellite images as a source of vehicle data. For the recognition and classification of vehicles it is planned to use convolutional neural networks. To improve the quality of work, selective introduction of high order neurons is proposed. The main tasks, problems associated with the recognition and counting of vehicles. The aim of the work is to develop a method to detect and classify vehicles with high accuracy, to conduct research to assess the quality of the method. It is planned to use the methods of digital image processing, machine learning and pattern recognition. A key element of the developed method is a second-order convolutional neural network. Keywords: transport planning; satellite pictures; ultrahigh permission; image processing; traffic flow; convolution neural network; digital image processing; machine learning; artificial intelligence; image identification. DOI: 10.25791/asu.06.2019.678 Contacts: E–mail: 007465@pnu.edu.ru Pp. 18-24.