Skip to main content
placeholder image

Vehicle Detection, Classification and Counting on Highways - Accuracy Enhancements

Chapter


Abstract


  • In Australian urban roads, pneumatic tubes are temporarily installed over roads to determine the road usage by vehicles. This is a relatively expensive process and the data cannot be obtained for about two weeks until a manual retrieval of data. In the past, we developed a highly accurate real-time computer vision-based system which relied on back ground subtraction, morphological operations and Gaussian filtering to track centroid of vehicles and accurately determine their speeds and count them. However, in this latest research, we provide our updated system that can determine not only speeds of vehicles but also identifies them including cyclists and pedestrian. This is achieved thorough neural network implementation allowing us to determine their speeds even when they do not follow a straight-line movement. This research utilizes the YOLO family, specifically YOLOv5 for neural network implementation. Such a system is very versatile in determining the variety of traffic in intersections that could not be handled in our previous approach using centroid tracking.

Publication Date


  • 2022

Edition


Citation


  • Premaratne, P., Blacklidge, R., & Lee, M. (2022). Vehicle Detection, Classification and Counting on Highways - Accuracy Enhancements. In Unknown Book (Vol. 13395 LNAI, pp. 394-408). doi:10.1007/978-3-031-13832-4_33

International Standard Book Number (isbn) 13


  • 9783031138317

Scopus Eid


  • 2-s2.0-85137263212

Web Of Science Accession Number


Book Title


  • Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Start Page


  • 394

End Page


  • 408

Place Of Publication


Abstract


  • In Australian urban roads, pneumatic tubes are temporarily installed over roads to determine the road usage by vehicles. This is a relatively expensive process and the data cannot be obtained for about two weeks until a manual retrieval of data. In the past, we developed a highly accurate real-time computer vision-based system which relied on back ground subtraction, morphological operations and Gaussian filtering to track centroid of vehicles and accurately determine their speeds and count them. However, in this latest research, we provide our updated system that can determine not only speeds of vehicles but also identifies them including cyclists and pedestrian. This is achieved thorough neural network implementation allowing us to determine their speeds even when they do not follow a straight-line movement. This research utilizes the YOLO family, specifically YOLOv5 for neural network implementation. Such a system is very versatile in determining the variety of traffic in intersections that could not be handled in our previous approach using centroid tracking.

Publication Date


  • 2022

Edition


Citation


  • Premaratne, P., Blacklidge, R., & Lee, M. (2022). Vehicle Detection, Classification and Counting on Highways - Accuracy Enhancements. In Unknown Book (Vol. 13395 LNAI, pp. 394-408). doi:10.1007/978-3-031-13832-4_33

International Standard Book Number (isbn) 13


  • 9783031138317

Scopus Eid


  • 2-s2.0-85137263212

Web Of Science Accession Number


Book Title


  • Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Start Page


  • 394

End Page


  • 408

Place Of Publication