Motorcycle accidents contribute substantially to morbidity and mortality. Motorcyclists are injured when their motorcycles fall. The impact of a motorcycle fall can cause brain injuries and many bone fractures; therefore, it is beneficial if we can identify fallen motorcycles early. The study proposed a method to identify motorcycle falls using You Only Look Once (YOLO) and a machine learning classifier. The study used YOLO to detect a motorcycle. Then we studied the performance of three machine learning classifiers to classify the detectable motorcycles into two categories: those that do not fall over and those that do. MobileNet, ResNet50, and VGG16 classifiers could achieve accuracies of 98.75%, 97.50%, and 100%, respectively, for motorcycle fall classification.
A Hybrid YOLO-VGG16 for Fallen Motorcycle Detection
05.03.2025
1143524 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Hybrid-driven motorcycle and hybrid kit of motorcycle
Europäisches Patentamt | 2022
|