<![CDATA[Faded road markings can reduce driver visibility and increase the risk of traffic accidents. This study aims to develop an automatic detection system to identify the level of road marking fading using a deep learning approach based on YOLOv8. The dataset consists of 2,049 road marking images categorized into two classes clear and faded and trained with four data augmentation variations: no augmentation, horizontal flip, saturation + exposure, and a combination of 90° rotation, grayscale, saturation, and brightness. The research process includes data collection, image preprocessing, model training, and evaluation using accuracy, precision, recall, and F1-score metrics. Experimental results show that the model trained with saturation + exposure augmentation achieved the best performance, with an accuracy of 86%, precision of 86%, recall of 86%, and an F1-score of 85%. These findings demonstrate that illumination- and color-based augmentation variations are effective in improving the model’s generalization capability under diverse environmental conditions. This study is expected to serve as a foundation for developing automatic road marking monitoring systems to enhance transportation safety and efficiency.]]>
Copyrights © 2025
