In modern urban environments, traffic congestion is still a problem that brings delays, fuel consumption, and pollution in terms of travel. Conventional traffic signal control is based solely on fixed timers without adaptation to real-time conditions, which amplifies the previously mentioned problems. This paper proposes developing an intelligent traffic signal control system that adapts dynamic signal timing according to the real-time density of detections using the YOLOv5 (You Only Look Once version 5) object detection algorithm. YOLOv5 has gained fame because of the high speed at which it analyses live video feeds streaming through traffic cameras. It very efficiently differentiates between plus counts the vehicles crossing from multiple lanes. Therefore, the developed system will optimize the traffic signal switching by using vehicle density information, reducing waiting time at intersections, and offering improved traffic flow. In contrast to the static systems, the method presented here is flexible and adaptive and would suit various changes in the daily conditions around traffic. The utilization of YOLOv5 makes it possible to improve the system's accuracy in vehicle detection, even with intricate traffic patterns, without diluting it down to real-time computation. A simulation model of the proposed model was performed using real traffic footage from cities, and significant improvements over traditional traffic management methods were demonstrated. Therefore, the system shows reduced congestion with better time distribution and a more efficient distribution of green signal time with better traffic throughputs. It can, thus, be inferred that real-time object detection combined with adaptive control of traffic lights can be a new strong combination in the development of smarter urban traffic systems that can considerably reduce congestion and their associated environmental impacts.
Real Time Vehicle Density Detection for Adaptive Traffic Signal Control Using YOLOv5
06.11.2024
890600 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch