Real-time high-precision traffic flow monitoring with DAS

Distributed Acoustic Sensing (DAS), with its high sensitivity, long-distance monitoring capabilities, and distributed sensing features, plays a significant role in intelligent transportation systems. However, achieving real-time, accurate, and cost-effective traffic monitoring remains a challenge due to the high sensitivity required and the complex field backgrounds. This paper proposes a real-time Rotated YOLO (R-YOLO) enhanced method for high precision monitoring traffic flow with DAS, including vehicle tracking, type recognition, and moving speed estimation. By monitoring four days of railway data and two days of road data, our model achieves a tilted detection accuracy of 96.7% for traffic flow within 13 seconds per frame, while the average speed estimation error across five railway datasets is only 4.28%. These metrics outperform traditional YOLO detection methods, demonstrating the promising potential of our model for application in traffic monitoring.