Reconhecimento de placas de veículos utilizando redes neurais artificiais

Abstract

Recognizing objects in images is a common routine for intelligent living beings, however, the computational approach capable of reproducing this characteristic is complex and requires a high processing capacity. The development of artificial neural networks has proven to be an effective and reliable solution to computer vision problems. The detection of license plates is one of these problems where we try to classify in images the place where each object is located in order to extract relevant information. The objective of this work is to train a deep neural network capable of classifying Brazilian standard license plates even if not trained with this type of data, to extract relevant information using character reading techniques and to analyze the classification performance of this model.