Septian Enggar Sukmana, Cahya Rahmad, Muh. Shulhan Khairy, Rudy Ariyanto, Ramona Andhani
Tire damage inspection is a crucial aspect of vehicle maintenance aimed at ensuring tires are in good condition before use. Visual inspections by humans have limitations and may not always accurately determine tire suitability. This study discusses how to detect car tire damage using a Convolutional Neural Network (CNN) with the MobileNetV2 architecture via a mobile application. The parameters used in training the CNN model included the Adam optimizer, a learning rate of 0.0001, a batch size of 16, and 32 epochs. The results indicate that the CNN model effectively detected tire damage, as evidenced by the model evaluation metrics: an accuracy of 93%, precision of 940%, recall of 93%, and an F1 score of 93 %. These findings suggest that the MobileNetV2 CNN architecture can serve as an effective tool for detecting tire damage, offering a reliable alternative to human visual inspections. © 2024 IEEE.
Jurusan Teknologi Informasi, Politeknik Negeri Malang, Malang, Indonesia