Recently, traffic Incidents have a significant negative impact on traffic flow safety. To realize a more accurate and faster incident detection mechanism, various types of AVID(Automatic Vehicle Incident Detection) algorithms have been proposed and ANNs (Artificial Neural Networks) are currently known to have the potential to achieve significantly good performance in terms of detection and false alarm rates, as well as operational improvements. Moreover, recently DNNs (Deep Neural Networks)

more »

... received huge attention due to their excellent performance on a wide range of recognition and classification tasks. In this study, DNNs (Deep Neural Networks) have been used to explore the performance of learning algorithm for automatic incident detection on freeways. The incident detection performance of the DNNs was compared with a general Back-Propagation (BP) model and the best network topology has been proposed for each model.