Red light and wrong parking violation detection system based on deep learning
Abstract
Nowadays, smart city is a rising trend; therefore, building a system that can automatically detect violations to reduce the pressure on traffic surveillance is necessary. In this paper, the authors have proposed a novel method that can help to detect various traffic violations such as going through red lights or wrong parking by using the YOLOv3 neural network to recognize violating vehicles, then giving information about the position of the vehicles identified by tracking object which can be used to classify the traffic violations. The proposed method is evaluated on a Da Nang traffic data set and the experiment has yielded promising results with an accuracy of 94%in morning dataset. In other conditions, the results are in the range of 40% and 80%.
Moraref