Estimativa de fluxo de trânsito utilizando processamento de vídeo.

Abstract

The design of smart cities assumes that public infrastructure systems such as energy, water, highways, among other types, can be better designed, executed and managed from information collected using integrated technology. Among the technologies that can be implemented, the use of surveillance cameras is already a reality today. In many cities in Brazil and around the world, the evaluation of the flow of vehicles on roads is carried out through the analysis of video cameras. Therefore, it is possible to use artificial intelligence technology to automate this analysis and thus assist managers in public policy decisionmaking. In this work, an automated form of vehicle flow analysis is presented, using a YOLO convolutional neural network to analyze the number of vehicles traveling through a specific road, automatically counting each type of vehicle in a given period.