Effective Vehicle Tracking Algorithm for Smart Traffic Networks

A smart traffic analysis system could be used to reduce congestion, prevent accidents, as well as to control traffic flow. Such a system would need to make use of many technologies, such as computer networking, communication, image processing, object detection, and tracking. In this paper, we introduce an efficient vehicle tracking algorithm which could be used to help create a smart city traffic control system. Our algorithm is targeted to solve the multi-object tracking (MOT) problem. Our method follows a track-by-detection paradigm, i.e., it relies on vehicle detection results to decide whether a detected vehicle in sequential frames belong to the same track. Our vehicle tracker extends the intersection over union (IOU) tracker and improves upon it by fusing historical tracking information, taking into consideration the balance between tracking efficiency and effectiveness. We demonstrate the effectiveness and efficiency of our approach using the UA-DETRAC benchmark dataset. Our proposed method runs at an average speed of 1,264 frames per second. We conclude that our tracker could be useful for applications running in a real-time environment.