Vehicle Classification Using Neural Networks and Image Processing

Abstract

Vehicle classification is getting important especially in security systems, surveillance, transportation congestion reduction, and accident prevention. However, it is difficult to classify the traffic objects due to the poor quality of images from videos. Hence, image processing techniques are applied to increase the accuracy of the result. The aim of this study is to propose a vehicle classification scheme where YOLO v5 algorithm and Faster R-CNN algorithm are being implemented separately into vehicle classification, followed by comparison of result between these two algorithms. In this study, vehicles are classified into five classes, namely motorcycle, car, van, bus and lorry. The labeled dataset is being split into training set and validation set and then trained under algorithm YOLO v5 and Faster R-CNN separately. Experimental results show that YOLO v5 performs better with the mean average Precision, Precision, and Recall rate up to 0.91, 0.81, and 0.86, respectively