With the increasing urbanization and the popularity of traffic video surveillance, and the rapid development of object detection algorithms, object detection of traffic events through video analysis has become possible. This paper proposes the design and implementation of a traffic event detection system based on video analysis. Firstly, the video stream is processed, mainly for video access and display, and the compression of neural networks achieves the network acceleration. Secondly, the structured information of vehicles is obtained by nighttime vehicle detection and joint detection and tracking. Finally, the car's driving behavior is analyzed through video calibration and video analysis. The system has been used online in many places in China and has achieved remarkable results.
Most of the existing trackers have high performance and good effect, but it is difficult to have fast speed and can not meet the real-time requirements. Therefore, the development of target tracking system with high accuracy and good robustness has become an urgent task for researchers. This paper proposes to add an Region Proposal Network (RPN) module based on the structure of the traditional Siamese Network (SiamFC), calculate the template and branch of the Siamese sub network in advance in the reasoning stage, and decompose the tracking task into task nodes that are detected first and then matched for real-time online tracking. The experimental results on VOT2015, VOT2018 and OTB2015 data sets show that compared with other trackers, the model based on SiamRPN proposed in this paper shows better performance in achieving the dual goals of high accuracy and low robustness.