Institute of Transportation, MOTC

      The vigorous development of the city will affect the deterioration of city traffic. In order to effectively improve the city traffic problem, it is necessary to effectively collect city traffic data. If the traditional vehicle detection equipment is used to collect data, it is difficult to reflect the real-time detailed operation status of the road immediately, and it must be compared manually to deal with the traffic situation, slow actions often cannot solve a critical situation. AI image recognition technology is gradually mature, not only it can collect more traffic characteristics data than the traditional detection equipment, it can also use the collected data to perform intelligent traffic management with artificial intelligence though the image recognition method of deep learning. This Study is first to investigate and plan for the AI image recognition technology and equipment for intersections and road sections based on the normal and temporary traffic flow survey operation conditions, to effectively collect the road traffic characteristics data; and then attempt to conduct traffic signal dynamic control at the intersections with the AI image recognition data, expecting the traffic flow can be more efficiently; finally, introduce the information dashboard design and planning to provide the traffic competent authorities to control the traffic characteristics at the intersections and road sections as well as the performance before and after the signal adjustment, so that the traffic competent authorities can take the correct actions to solve the problem on traffic improvement; in addition, this Study also aims at Taoyuan Airport to investigate the plan for detection of vehicle parking violations around the terminal and time detection of passengers passing through the security check points. The detection data can provide the references for the Management Units