A Real-Time Video Surveillance for Rule Violation Detection on High Ways

Abstract

Processing of images obtained with cameras is important for monitoring and/or securing highways. Early warning is important for highway safety in cases such as accident, traffic rule violation, automatic detec-tion of people or animals entering the road. For this purpose, besides the detection of objects such as vehicles and people, other moving targets or objects falling on the road, such as boards, etc., must be detected. For this, besides object recognition, motion detection and object tracking are also needed. The purpose of motion detection is to classify pixels in an image as background and foreground. Within the scope of this study, deep learning-based machine learning method was used for the detection of vehicles and people entering the highway, and classical image processing methods were preferred for the detection of any moving object other than these. In this context, a software infrastructure has been developed in which different processes can work together and share data with each other. The developed software was tested on the Nvidia Jetson development card, and algorithm performance and speed tests were carried out on the selected GPU supported Nvidia Jetson card.

Keywords

• Traffic surveillance
• Highway security
• Vehicle detection
• Vehicle tracking

References

1. Delibasoglu, I. (2022). A real-time video surveillance for rule violation detection on highways. The Eurasia Proceedings of Science, Technology, Engineering & Mathematics (EPSTEM), 19, 15-20.