The existing traffic control system is not entirely prepared to handle severe traffic congestion, particularly in India, where old systems such as fixed-timing signals and police personnel manually controlling traffic at crossings are still used. While several researchers have investigated the idea of automated traffic control using AI, these systems have not been widely adopted in India. One key issue in implementing these models is the inability to accurately detect the variety of vehicles found in Indian traffic, such as auto rickshaws, motorcycles, and heavier trucks, which are frequently neglected by previous researchers' systems. This study aims to fill these gaps by developing a new approach for calculating length using one single bounding box incorporating vehicles inside and allowing for real-time adjustment of signal timers in response to changing traffic conditions. This study also incorporates features like accident detection, road clearance for emergency vehicles, and predicting peak traffic hours, which were previously researched separately and not combined. By researching them together, this research anticipates developing a new system, resulting in a more efficient and adaptive traffic system. Testing found that the previous YOLO model had a detection rate of 28.57%, whereas this new method had a detection rate of 89.29% in the vehicle density estimate. As a result, this approach provides a more adaptable traffic control solution, with considerable benefits for Indian society.
AI Based Traffic Flow Prediction for Smart Urban Mobility
20.12.2024
652835 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
2-dimensional flow prediction based intelligent urban traffic-flow prediction method
Europäisches Patentamt | 2016
|