Classificação de Vagas de Estacionamento usando Aprendizagem de Máquina

In this paper we present contributions to the classification of parking spaces monitored by video cameras, using machine learning with classification by Support Vector Machine. The contributions include the usage of a Radial Basis Function kernel and of a threshold in the decision region. Such contributions lead to better classification accuracy with little impact in the performance of an embedded implementation. The proposed method is implemented in an embedded environment, classifying up to 256 parking spaces per image with the expected accuracy.