期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2019
卷号：97
期号：7
页码：2062-2075
出版社：Journal of Theoretical and Applied
摘要：This paper proposes and implements an efficient real-time vehicle counting (RT-VC) approach. This approach is based on the most efficient action detection and tracking methods in computer vision. YOLO is used for object detection, whereas Kalman filter with Hungarian algorithm are used for tracking. The road is divided into two zones of interest by the end-user, and any vehicle will be counted if its trajectory crosses these zones of interest. The experiments show that the proposed system is very accurate in comparison with other existing approaches. For comparative evaluation, our proposed approach obtained accuracy above 90% for most of the tested videos in the highway roads. Therefore, the proposed approach can efficiently work with many real-time surveillance systems and has the potential to be used in many real road applications.
