Real-Time Vehicle Detection and Classification Using Deep Learning based Approach

Abstract

Computer vision is a key component of many technologies, including Automation in industry, robotics, character recognition, human-machine interfaces, text analytics, and motion detection technologies.  One significant area within this field is the identification of moving objects, crucial for applications like security cameras and intelligent transportation systems. This paper explores vehicle detection and classification using YOLO base algorithm, focusing on YOLOv5 due to its efficiency and speed. The study evaluates the proposed method using two datasets: Highway Traffic Videos and Vehicle Detection Image Dataset. The YOLOv5 model demonstrated high accuracy in detecting and classifying vehicles, with Confidence scores for vehicles, specifically cars and trucks, fall within the range of 0.41 to 0.86. Results indicate that YOLOv5 achieves effective vehicle detection in both low and high traffic scenarios, addressing challenges such as occlusion and low-resolution footage. The proposed model significantly enhances real-time traffic monitoring and management, highlighting its potential for intelligent transportation systems. The findings underscore YOLOv5's applicability in improving vehicle detection accuracy, performance and efficiency.