Rustam Asnawi, Deny Budi Hertanto, Sarwo Pranoto, Ade Pangalasan, Muhammad Haikal Hanif, Muhammad Fikri, Muhammad Minto Aji Shodiq, Asnan Munandar, Thoriq Fadhlurrohman
The efficient and secure connection between trains and their couplers is of paramount importance for the safety and functionality of railway systems. This research presents a novel approach to addressing this challenge through the utilization of a pre-trained deep learning model, specifically Single Shot MultiBox Detector (SSD) with Mobilenet V1 architecture. In this research, SSD-Mobilenet V1 is used to identify and classify whether the coupler is connected or disconnected. A dataset of 2843 coupler images in .jpg extension was used for training and testing. The obtained results showed that the SSD-Mobilenet V1 model achieves high classification performance with 98% accuracy. © 2025 American Institute of Physics Inc.. All rights reserved.
Department of Electrical Engineering Education, Faculty of Engineering, Universitas Negeri Yogyakarta, Yogyakarta, Indonesia
