For various applications it is useful to automatically divide vehicles passing a point into several categories. Forth te method described the sampled signal of an inductive loop detector, called the footprint, is used. This signal is normalized to become the input of neural network that is used to separate the passing vehicles into five categories. This paper describes the choices that were made in the selection of the footprint transformation and the selection of the neural network to be used. A set of vehicle measurements with an accompanying video tape was used to train the network and to test the performance of a trained network. This research leads to the conclusion that the developed classification system, based on a 3-layer perceptron with backpropagation, gives useable results. Further research is necessary to investigate the possibilities of other types of networks and the influence of footprint transformation methods.
Classification of vehicles using neural networks
Die Einstufung von Fahrzeugen durch die Benutzung neuronaler Netzwerke
1993
22 Seiten, 9 Bilder, 3 Tabellen, 7 Quellen
Conference paper
English
Classification of vehicles using neural networks
Automotive engineering | 1993
|Classification of vehicles using neural networks
British Library Conference Proceedings | 1993
|Neural networks for intelligent vehicles
Tema Archive | 1993
|