Advancing-Traffic-Sign-Detection-with-YOLOv9-Evaluation-on-a-Custom-Dataset

Traffic sign detection is vital for intelligent transportation systems, enhancing vehicular safety and efficiency. This study evaluates YOLOv9, the latest version of the You Only Look Once object detection model, on a custom traffic sign dataset. YOLOv9, introduced by Chien-Yao Wang, I-Hau Yeh, and Hong-Yuan Mark Liao in 2024, features Programmable Gradient Information (PGI) and the Generalized Efficient Layer Aggregation Network (GELAN) for improved accuracy. Our experiments demonstrate that YOLOv9 excels in detecting small and diverse traffic signs, outperforming previous YOLO versions and other models in precision, recall, and mean average precision (mAP). These findings underscore YOLOv9's potential to advance traffic sign detection, contributing to safer and more efficient transportation systems.

Keywords: Object detection, YOLOv9, Traffic sign recognition, Deep learning, Computer vision