Implementación de prototipo de reconocimiento de placas de carros mediante redes neuronales convolucionales mediante librerías open source.

Abstract

The objective of this project is the application of convolutional neural networks for the study of images through the use of open source libraries. Within this project, the study of the term Deep Learning and specifically of convolutional neural networks was carried out. The objective of the extraction is to get to recognize specific characteristics of the images entered into the system. After detecting the vehicle registration of any image, optical character recognition is performed, which allows detecting and extracting the text from the detected image. Finally, the information obtained is stored separately, the image goes directly to a folder while the text to a database. The first section exposes the description of the project and the theory consulted. Within this, the terms Deep Learning, convolutional networks, open source libraries and programs that allowed the realization of the project were studied. The second section covers the methodology used for its development, this was divided in stages that allowed a better and orderly implementation of the detection model. The third section corresponds to the results obtained. The verification of the project consisted in the detection of the license plate and its text from an image. These records were stored in a folder (images) and in a database. The last section covers the conclusions that were reached after carrying out the project, in addition to the recommendations that people who carry out this project in the future can make.