Sistema de control para distribución de combustibles mediante la identificación de placas en vehículos utilizando visión artificial

Abstract

The recognition of vehicle plates by its acronym in English ALPR is an important advance in the field of artificial vision, to identify and read the characters of a license plate, the applications based on this technology are applied in various fields such as security and control. The present work develops a system of identification and control of vehicles in the fuel supply by means of the Jetson Nano 4GB development card with the use of convolutional neural networks and artificial vision, for the creation of the neural network was supported by CNN Mobilnet v2 modifying and retraining it in Google Colab reducing the complexity of the network optimizing the inference processes to later send the model to the Jetson Nano card that contains a series of Nvidia's own libraries such as Jetson Utilies which is integrated by TensorRT, CUDA , VPI among others. The control process is carried out by creating files with a .CVS extension in which the vehicle license plate is stored in text format to send the information to a server that is in charge of checking if there is repetition and, if there is, sending an alert via email. The implementation of the system in controlled environments, its effectiveness was verified with results greater than 90% in the detection and control of vehicles.