Use of computer vision in assessment of air quality in city transportation system

An approach to the formation of a description of the city's transport system is considered in order to identify the most polluted road spans with vehicle exhaust gases. The quantitative characteristics of pollution are determined in accordance with the standard Russian GOST R 56162-2019. When estimating emissions from road transport, information presented on the web in graphical form about traffic jam is used. In addition, the information on the number of cars, average speed and type of transport obtained by analyzing the video stream from web cameras by the YOLO neural network (version 4) is also used. The binding of pollution results to the transport system is formalized using the transport system ontology. The results can be used in dynamic models of pollution in urbanized area