Adaptive Traffic Signal Timing: Leveraging YOLOv10 and Computer Vision for Real-Time Optimization

Traffic congestion poses a significant challenge in modern urban environments, particularly in densely populated cities. Traditional traffic signals, which operate on fixed timings, often require manual adjustments to respond to varying traffic densities, leading to inefficient traffic flow. This issue arises because fixed intervals do not account for the dynamic nature of traffic patterns throughout the day. While developing intelligent systems that adjust signal durations dynamically offers a potential solution, such systems involve substantial financial investment and extended implementation timelines. This paper proposes a method to optimize green light durations by analyzing footage collected from a traffic junction in Bengaluru. This analysis can be performed using existing CCTV footage or manually gathered data. Even limited sample sizes can provide valuable insights for preliminary adjustments. The approach utilizes YOLOv10, a state-of-the-art computer vision algorithm, to extract relevant traffic data. The data is processed by an algorithm designed to reallocate green light durations, which are then compared with the original timings. An experimental setup involved recording one-minute videos over three days at the selected junction and applying the proposed algorithm to assess and refine signal durations. The results demonstrate the feasibility of using such algorithms to improve traffic signal management effectively.