Traffic Surveillance to detect wrong-way vehicles by image processing
No Thumbnail Available
Date
2023-07
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Department of Electronic and Telecommunication Engineering
Abstract
Wrong-way driving is one of the throughout the process causes of road accidents and traffic
blocks all over the worldwide. By detecting wrong-way vehicles, the frequency of accidents
may be minimized and traffic congestion can be alleviated. With the rising popularity of real time traffic management systems and thanks to the availability of cheaper cameras, the
surveillance video has become a big source of data. In this study, we present an autonomous
wrong-way vehicle identification system using on-road security camera footage. Our system
operates in three stages: the recognition of vehicles from the video frame by utilizing the You
Only Look Once (YOLO) method, track each vehicle in a specific zone of interest using
the centroid tracking technique, and identify the wrong-way of driving vehicles. YOLO is
incredibly precise in object detection and the centroid tracking method can track any moving
item efficiently. Experimenting with several traffic recordings indicates that our proposed
system can recognize and identify any wrong-way vehicle in varied light and weather
situations. The system is very simple and straightforward to apply.
Description
Submitted By Ariful Islam T181023
Keywords
vehicle, YOLO, centroid, wrong-way, computer vision, vehicle tracking