The timely detection of road traffic accidents is of great significance for improving rescue efficiency and reducing casualties. Addressing this critical issue, this paper proposes an improved method based on the YOLOv11 algorithm, aimed at enhancing the accuracy of road traffic accident detection. Building on the existing YOLO series algorithms, we have introduced the Asymptotic Feature Pyramid Network (AFPN) mechanism to optimize YOLOv11, making it more suitable for the scenario of road traffic accident detection. Experimental results show that the improved YOLOv11n algorithm has increased the detection accuracy by 4.1 points in the mAP \(_{50-95}\) metric compared to the original algorithm, significantly enhancing the accuracy of traffic accident detection, which aids in timely rescue and reduces the mortality rate of traffic accidents. This study provides a new perspective and method for the development of road traffic accident detection technology.