Abstract: Timely identification and efficient repair of road defects is critical to road safety and longevity. At present, the performance of deep learning models in detecting complex road defects, especially for small targets, is poor, and it is easy to miss detection and false detection, so a road defect detection model based on improved YOLO11n is proposed. Firstly, the CGBlock context guidance module is introduced to effectively extract local and global feature information, expand the receptive field, and significantly enhance the accuracy of detection. Secondly, the DRCGN cross-scale feature fusion module is designed and introduced to effectively capture the feature information of different scales, and the rich context information is fused to solve the problem of missed detection and misdetection of small target defects on the road, so as to improve the accuracy of detection. Finally, on the basis of the MPDIoU loss function, the scale factor learning mechanism is integrated to capture richer and more detailed feature representations, so that the generalization performance is improved and the convergence speed is significantly accelerated. The experimental results indicate that, compared to the original model, the improved model achieves a 3.2% increase in precision, a 1.5% improvement in recall rate, and a 3.0% enhancement in mAP@50 on the RDD2022 dataset, thereby validating the effectiveness of the proposed optimization approach..