Improving Turn Movement Count Using Cooperative Feedback.

Area of Science:

• Computer Vision
• Traffic Engineering
• Artificial Intelligence

Background:

• Vision-based Turn Movement Count (TMC) systems struggle with accuracy in heavy traffic due to occlusions.
• Vehicle detection and tracking failures lead to miscounted or missed movements at intersections.
• Existing methods lack robustness in complex traffic scenarios.

Purpose of the Study:

• To develop a cooperative method for improving TMC accuracy under challenging traffic conditions.
• To address limitations of current vision-based TMC systems, particularly occlusions and misclassifications.
• To enhance the reliability of traffic data collection at intersections.

Main Methods:

• Proposing a cooperative method integrating contextual observations from surrounding areas.
• Implementing information sharing between observation systems at neighboring intersections.
• Utilizing a cooperative scheme to infer missing or occluded vehicle movement data.

Main Results:

• The proposed cooperative method significantly improves the accuracy of Turn Movement Count (TMC).
• Demonstrated superior performance compared to existing reference methods in challenging traffic.
• Successfully mitigated issues caused by vehicle occlusions and improved data completeness.

Conclusions:

• Cooperative observation systems enhance the accuracy and reliability of vision-based TMC.
• Information sharing is crucial for overcoming limitations in complex traffic environments.
• The proposed method offers a robust solution for accurate traffic monitoring.