Pervasive Intelligent Traffic Surveillance System is a prototype system that uses existing roadside camera traffic monitoring streaming videos to calculate and collect traffic flow information. By using deep learning technology, it is possible to calculate traffic data from low frame rate traffic flow video streams and store the traffic data to provide long-term traffic monitoring and recording functions. This system is combined with SAGE2(Scalable Amplified Group Environment 2), and GIS (Geographic Information System), which is visualized on a TV wall and used as a traffic flow monitoring and data analysis platform interface, so the platform can display multiple traffic videos simultaneously to improve the efficiency of traffic monitoring.

This system is to realize the real-time traffic flow monitoring system by combining the image recognition network of YOLOv3 and the multiple objects tracking algorithm. Through this system, it is possible to automatically monitor the streaming image information of multiple roadside cameras, and to provide more relevant traffic flow information to the units that will pass through the application of abnormal event detection and violation detection.

The intelligent traffic monitoring system developed by this technology can be combined with control ability to become a traffic management system in the future, and further plan or control the traffic flow of each road segment; long-term collection of domestic road traffic dataset can strengthen the detection speed and accuracy of deep learning, and improve the current network; in the future, it will be possible to combine traffic flow information and traffic conditions into vehicle navigation systems or automated driving systems to provide the best route planning.