AI-Based Intelligent Traffic Signal Control System Using Real-Time Vehicle Density Detection

Traffic congestion has become a severe problem in the urban areas due to the increase in vehicles numbers at an extremely fast rate. Majority of the traffic signal systems presently in operation use a pre-fixed timing cycle without accounting the real-time number of vehicles approaching. This results in wasted signal time, extra wait time, high fuel consumption, and excessive pollution. There are number of ways proposed by sensors and traditional image processing, but these methods often require a high investment, low accuracy, and low performance when implemented on real time conditions.

This research is intended to be a smart traffic signal controller system which works using an artificial intelligence and a real-time detecting system of vehicle density. The system utilizes deep learning techniques, like YOLO for identifying and counting vehicles from the CCTV cameras by real-time videos. Lane based vehicle density is used for adjusting the timing for all the signals dynamically to smooth traffic flow. The suggested system has no need for installing extra sensors, is economical, easily scalable and efficiency.