VISION-BASED ADAPTIVE TRAFFIC SIGNAL CONTROL FOR DYNAMIC URBAN TRAFFIC
Priyanka S Rumane, Nilesh R. Wankhade, S B Bagal, Rupali B Pekhale, Hemant R Mankar, Leena Deshpande
Abstract: Traffic jamming at crossroads of urban centres has been a key problem because of the prevalence of fixed-time based traffic signal management systems which cannot cater to real time changes in traffic demand. In order to overcome this weakness, this paper presents a vision-based adaptive traffic control signal framework, which adapts the green signal duration dynamically depending on the estimation of the lane-elevated traffic density. The suggested system involves a trained vision-based vehicle detector model that approximates the number of vehicles coming to each lane of a signalized intersection. Adaptive green signal durations are then calculated using an analytical vehicle discharge time model and optimization of clearance of queued vehicles is done without breaching operational constraints. The system is adaptive dedicated to a lane, meaning that the lane that has the most traffic demand is served in each control cycle and conflict-free signal cycles are executed. Experimental analysis in simulated traffic conditions proves that the given approach leads to a significant decrease in average waiting time of vehicles and increase in intersection throughput in comparison with traditional fixed-time traffic signals regulation. Along with the quantitative performance improvements, qualitative evidence has ensured that traffic sensing, decision-making, and signal executions have been successfully integrated in real-time using a visual monitoring dashboard. The suggested structure will remove the use of invasive physical sensors and provide a scalable and viable solution in smart traffic management of urban areas. The findings place emphasis on the promise of adaptive signal control systems based on vision in order to increase the efficiency and efficiency of traffic flows and minimize congestion in heterogeneous traffic systems.
Keywords: Adaptive traffic signal control; Computer vision; Traffic density estimation; intelligent transportation systems; Vision-based traffic management
DOI: 10.24874/PES08.03.010
Recieved: 05.05.2026 Revised: 24.06.2026 Accepted: 21.07.2026
UDC:
Reads: 19 
