YOLOv11 기반 교통안전표지 검출 모델의 성능 비교 = (Performance Comparison of YOLOv11-Based Traffic Signs Detection Models)

다국어 초록 (Multilingual Abstract)

Recently, with the introduction of intelligent technology and the development of artificial intelligence-based transportation systems, the establishment of intelligent transportation infrastructure is increasing in domestic and abroad. Specifically, object detection technology is a key element of an intelligent transportation system and seeks to improve the safety and efficiency of the system. This study compared and analyzed the performance of the latest object detection models YOLOv10 and YOLOv11 to verify the efficient management of traffic safety facilities and the applicability of the intelligent transportation system. According to the protection zone integrated sign standard, 49 types of traffic safety signs were set as detection targets and various image data were collected to construct a total of 41,270 datasets. In addition, the quality of the dataset was improved and the model's detection performance was optimized through stepwise data expansion and repetitive learning. As a result of the experiment, it was found that the YOLOv11 model showed better performance in terms of object detection performance and reliability, and the precision and reproducibility were improved at the same time compared to the previous model.