Helmet Detection for Motorcycle Riders and Passengers in Siriraj Hospital Area Using Deep Learning

Abstract

Siriraj Hospital has a large number of patients, some of whom use motorcycles without helmets, leading to serious injuries when accidents occur. To address this issue, the hospital implemented a policy in 2020 requiring everyone in the Siriraj Faculty of Medicine to wear a helmet. However, visual inspection by staff has limitations and is prone to errors, making it difficult to enforce the policy. This research presents a system for detecting helmet use by riders and passengers within Siriraj Hospital. The system uses the YOLOv8 deep learning model for object detection, trained on a custom dataset collected from CCTV cameras. This approach outperforms manual observation and achieves high accuracy, with a precision of 0.842, recall of 0.811, mAP of 0.858, and F1-Score of 0.826 for YOLOv8x and a precision of 0.868, recall of 0.790, mAP of 0.859, and F1-Score of 0.827 for YOLOv8l The system has the potential to improve helmet compliance and reduce motorcycle-related injuries at Siriraj Hospital.