RESEARCH COMPUTER VISION TECHNOLOGY FOR TRAFFIC MONITORING AND VEHICLE TRACKING IN REAL-TIME

Abstract

This paper presents an advanced system for traffic monitoring and management, which leverages YOLOv7 for image processing. The system is capable of accurately
counting traffic, identifying potential traffic hazards, and detecting objects using the Haar cascade algorithm, while also calculating vehicle speeds. By utilizing YOLOv7 the
system can accurately identify over 85 types of vehicles and objects, tracking their direction of movement and creating image traces on the screen for easy monitoring and
surveillance (80 types of vehicles and objects and 5 types of objects for custom data). Moreover, the traffic flow count results are highly precise, incorporating two
parameters: the traffic density at a node and the total number of vehicles within a specific period, ensuring accurate determination of traffic levels at each intersection.
Additionally, the system can measure the speed of cars on highways using the Haar Cascade algorithm and OpenCV, enabling real-time monitoring of vehicle speed and
saving images of the vehicle with the current speed display. Overall, this system is designed to support the monitoring and management of vehicles on highways, enhancing
traffic safety and efficiency.