• Version
• Download 4
• File Size 355.85 KB
• File Count 1
• Create Date 26 April 2023
• Last Updated 26 April 2023

Traffic Proctoring Using YOLOv5

Rajbhoj Harshal *, Yadav Vikas, Satam Pranav, Dalwai Naima, Maste Deepali

Students, Head Of Department

Department of Information Technology
Atharva College of Engineering
Mumbai, Maharashtra, India

Abstract—Motorcycle accidents are a leading cause of fatalities in traffic accidents, particularly in developing countries. Non-use of helmets by riders or passengers is a major contributing factor to fatal injuries in these accidents. To address this issue, a deep learning-based approach for automatic helmet detection of motorcyclists is presented in this research paper. The approach involves two stages. An improved YOLOv5 detector is used in the first stage to identify motorcycles, their riders, in video surveillance footage. In the second stage, the detected motorcycles are further analyzed to determine whether the riders are wearing helmets.The bettered YOLOv5 sensor is enhanced with the emulsion of triplet attention and the use of soft- NMS rather of NMS to improve its delicacy. To validate the proposed method, a new and more comprehensive motorcycle helmet dataset. This dataset is larger than existing datasets and is derived from multiple traffic monitoring sources in Indian cities

Index Terms—Helmet detection system, YOLOv5, Background subtraction, Image processing.