Robust Motorcycle Helmet Detection Benchmarking

Date of Award

12-1-2023

Degree Name

M.C.S. (Master of Computer Science)

Department

Department of Computer Science

Advisor/Chair

Advisor: Tam Nguyen

Abstract

Ensuring road safety is of utmost importance, with a particular focus on detecting anomalies in traffic videos, specifically concerning helmet usage by motorcyclists. Wearing helmets significantly reduces the risk of head injuries, emphasizing the need to identify situations where riders are not wearing them. This study proposes an innovative approach that combines computer vision techniques and deep learning algorithms to detect and categorize helmets in traffic videos. By harnessing a pre-trained object detection model, we pinpoint areas of interest within video frames and subsequently analyze them to determine the presence or absence of a helmet. To enhance the accuracy of helmet detection, we employ various techniques such as image preprocessing, data augmentation, and model fine-tuning. Additionally, we leverage attention mechanisms and temporal information to bolster the algorithm's robustness. The developed system undergoes rigorous evaluation using a diverse dataset of traffic videos, covering both typical and abnormal instances of helmet usage. We assess the model's effectiveness through performance evaluation metrics like precision, recall, and F1-score. Our results unequivocally demonstrate the efficacy of the proposed approach in accurately detecting anomalies related to helmet usage in traffic videos. The implications of this research extend to road safety authorities, law enforcement agencies, and policymakers. By automating the helmet detection process, we enable the prompt identification of potential risks, facilitating timely interventions and enforcement of safety regulations. Ultimately, this study contributes significantly to the overarching goal of reducing accidents and fostering safer road environments for motorcyclists.

Keywords

Helmet Detection, Motorcycle Helmet, Motorcyclist, Helmet

Rights Statement

Copyright © 2023, author.

Link to Full Text

Share

COinS