KLASIFIKASI KONDISI BAN KENDARAAN MENGGUNAKAN ARSITEKTUR VGG16

Abstract

Tyres are the main component that a vehicle needs to work with reducing vibration due to uneven road surfaces, protecting the wheels from wear to provide stability between the vehicle and the ground helping to improve acceleration to facilitate travel while driving. Wear ensures stability between the vehicle and the ground helps improve acceleration for easy movement and driving. Caused including components that are often used, tires can experience damage such as the appearance of cracks in the tires. Cracks in tires can be triggered by factors such as age or the cause of the road that has been exceeded. Detection of tire cracks at this time is still carried out conventionally, where users see directly the state of the tire whether the tire is in good condition or cracked. Conventional methods are important because they maintain tire quality and rider safety. The Conventional Method certainly has weaknesses because vehicle users must have good vision and the ability to distinguish normal tires or cracked tires, but this method is considered less effective because it still uses human labor, causing the risk of human error (human negligence) which can hinder the process of identifying tire cracks. Based on this problem, this study will develop a deep learning model that can classify cracked tires using the VGG16 architecture. In this study, the model was created using 8 scenarios by changing the value of epochs, to get the best parameters in making the model. The results of the 8 scenarios carried out in this study are the best scenario obtained in scenarios 1,3,4 which get 98% accuracy in model testing