<![CDATA[This research aims to initiate an automatization process in the method of classifying vehicle types in the Jasa Marga transaction service system, which is the largest toll road operator company in Indonesia. The method used is YOLOv8 which is the latest version of the YOLO algorithm which is state-of-the-art performance in image processing. The dataset used in this study consists of vehicle images obtained from transactional data in an electronic toll collection system operating on toll roads, comprising five vehicle classification classes. In the initial stage, the images are examined and processed using pre-processing techniques such as data cleaning, image masking and data annotation. Next, the YOLOv8 model is trained using the data and tested on a separate validation dataset to measure the model's performance. Based on the results of experiments that have been carried out in this research, the performance of the YOLOv8 model without handling imbalance data resulted in an accuracy of classification of vehicle class types of 91.4%, while the performance of the model that handled imbalance data using under-sampling resulted in an increase in classification accuracy of vehicle class types to 94.4 %.]]>
Copyrights © 2024
