Real-time Vehicle Detection, Tracking and Counting System Based on YOLOv7

Abstract

The importance of real-time vehicle detection tracking and counting system based on YOLOv7 is an important tool for monitoring traffic flow on highways. Highway traffic management, planning, and prevention rely heavily on real-time traffic monitoring technologies to avoid frequent traffic snarls, moving violations, and fatal car accidents. These systems rely only on data from timedependent vehicle trajectories used to predict online traffic flow. Three crucial duties include the detection, tracking, and counting of cars on urban roads and highways as well as the calculation of statistical traffic flow statistics (such as determining the real-time vehicles flow and how many different types of vehicles travel). Important phases in these systems include object detection, tracking, categorizing, and counting. The YOLOv7 identification method is presented to address the issues of high missed detection rates of the YOLOv7 algorithm for vehicle detection on urban highways, weak perspective perception of small targets, and insufficient feature extraction. This system aims to provide real-time monitoring of vehicles, enabling insights into traffic patterns and facilitating informed decision-making. In this paper, vehicle detecting, tracking, and counting can be calculated on real-time videos based on modified YOLOv7 with high accuracy.