Sistema assíncrono para reconhecimento automático de placas de motos

Abstract

Nowadays, the increase in urbanization has allowed the development of new research and technologies that allow better management of cities. Among the new technologies under development is the automatic recognition of license plates. This area is important for it’s application in various activities such as traffic control and application of tickets in vehicles. Most works developed in Brazil do not support the recognition of license plates of two-wheelers, because of this, this work presents a system for automatic recognition of motorcycle license plates. During the literature search, a new pipeline was found to carry out the process of recognizing the license plates of the target vehicles of the system. Each of the steps in this pipeline were implemented as services that send and receive data in the form of asynchronous messages managed by a broker. This was done to decouple and allow asynchronous communication between these components. For the development of the motorcycle and license plate detectors, the YOLOv5 object detection model and the python language were used. To validate the viability of this system, tests were carried out in the transport of frames from the videos that were processed by the system.