Deteksi Pengendara Motor Berhelm dan Tidak Berhelm Berbasis Algoritma ZFNet dan MobileNet Single Shot Detector

Abstract

In an effort to improve driving safety, the development of a detection system for helmeted and unhelmeted motorcyclists is important. This study aims to implement the ZFNet and MobileNet SSD algorithms to distinguish helmeted and unhelmeted motorcyclists based on image analysis. The method relies on a deep learning approach to extract features from motorcyclist images and perform classification based on the presence of helmets. Performance evaluation is performed using a variety of image datasets. Experimental results show satisfactory accuracy, with ZFNet and MobileNet SSD achieving 86% accuracy. This system has the potential to be implemented in traffic safety systems to support efforts to prevent riding accidents caused by improper helmet use.