Scholarlink

<p>Intelligent Transportation Systems (ITS) are pivotal in improving road safety, optimizing traffic flow, and enhancing the driving experience. However, road accidents continue to pose significant risks to passengers, pedestrians, and infrastructure, necessitating more advanced solutions. This study focuses on three key aspects of ITS: object tracking and classification, vehicle speed estimation, and lane tracking. Utilizing the You Only Look Once (YOLOv8) algorithm, a state-of-the-art deep learning model, the research evaluates its effectiveness in real-time traffic surveillance. Comparative analysis with YOLO-NAS reveals YOLOv8’s superior performance, achieving higher precision (0.85 vs. 0.75), recall (0.82 vs. 0.70), and mean Average Precision (mAP) scores (0.80 vs. 0.68), along with faster inference times. The system also accurately detects and tracks vehicles exceeding speed limits and monitors lane usage with annotated outputs for improved enforcement and safety measures. Statistical analysis of speed violations showed only 3% of vehicles exceeded 130&#xa0;km/h, with speeds ranging from 133 to 162&#xa0;km/h. Lane analysis highlighted variations in vehicle types and speeds, indicating traffic flow dynamics. These findings demonstrate the potential of YOLOv8 in enhancing ITS applications, offering a robust and efficient framework for traffic management, accident prevention, and road safety improvements in urban environments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Design and Analysis of Visual Vehicle Tracking System for Traffic Surveillance Using Deep Learning

  • 期刊论文

  • 发表日期：

    2025年2月21日

查看全文
  • Muhammad Amir Mohd Nazri,
  • Fahmi Samsuri,
  • Vigneswaran Narayanamurthy

摘要

Intelligent Transportation Systems (ITS) are pivotal in improving road safety, optimizing traffic flow, and enhancing the driving experience. However, road accidents continue to pose significant risks to passengers, pedestrians, and infrastructure, necessitating more advanced solutions. This study focuses on three key aspects of ITS: object tracking and classification, vehicle speed estimation, and lane tracking. Utilizing the You Only Look Once (YOLOv8) algorithm, a state-of-the-art deep learning model, the research evaluates its effectiveness in real-time traffic surveillance. Comparative analysis with YOLO-NAS reveals YOLOv8’s superior performance, achieving higher precision (0.85 vs. 0.75), recall (0.82 vs. 0.70), and mean Average Precision (mAP) scores (0.80 vs. 0.68), along with faster inference times. The system also accurately detects and tracks vehicles exceeding speed limits and monitors lane usage with annotated outputs for improved enforcement and safety measures. Statistical analysis of speed violations showed only 3% of vehicles exceeded 130 km/h, with speeds ranging from 133 to 162 km/h. Lane analysis highlighted variations in vehicle types and speeds, indicating traffic flow dynamics. These findings demonstrate the potential of YOLOv8 in enhancing ITS applications, offering a robust and efficient framework for traffic management, accident prevention, and road safety improvements in urban environments.