Desenvolvimento de sistema automatizado para entrada de veículos em locais privados com visão computacional

Abstract

This work consists in the implementation of a system for automating access to closed places, where the system is capable of detecting the presence of the vehicle and reading its license plate and the control and supervision are operated remotely, the devices are connected over the Internet using a VPN (Virtual Private Network). An IP camera was used, which is responsible for detecting the movement of the vehicle and sending the image for processing through the FTP (File Transport Protocol) to the remote server which will process the image. On this computer, the software was developed to check if there is a vehicle in the correct processing position, using the YOLO object detection algorithm (You Only Look Once), after which Openalpr software and Google OCR (Optical Character Recognition) were used to extract the characters from the license plates. The algorithm checks if the card is registered, and if the answer is affirmative, it sends the command to open the gate via the UDP (User Datagram Protocol) to the controller, a task performed by the ESP8266. The vehicle data being stored in a database for future consultations. The system proved to be functional with an appropriate performance for the application, thus allowing automated and controlled access control in private places.