Road Defect Detection Algorithm Based on Improved YOLOv11n

Road surface defect detection is a key technology to ensure the safe operation of transport infrastructure, and its detection accuracy directly affects the road maintenance efficiency and traffic safety. Aiming at the deficiencies of existing detection methods in multi-scale feature adaptability, leakage detection and false detection, a road defect detection model based on improved YOLOv11n is proposed. Firstly, a new structure C3K2-SC is constructed by optimising the feature extraction structure of the backbone network, coupling the lightweight module StarBlock and the attention mechanism CAA into the Bottleneck module, which enhances the contextual information of the target while reducing the number of model parameters and improves the detection capability of small targets. Secondly, in order to enhance the multi-scale feature extraction and fusion capability, hypergraph computation is introduced into the neck network, so that it can more accurately correlate the relationship between the target and its surroundings, and reduce the misdetection and omission of detection. Experimental results show that the Precision, recall, F1 Score, mAP@50, mAP@70 and mAP@ <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$50: 95$</tex> of the improved model are improved by 4.8 %, <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{2. 8 \%}, \mathbf{3. 7 \%, } \mathbf{2. 5 \%, } \mathbf{2 \%}$</tex> and <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{2. 8 \%}$</tex> respectively compared with the benchmark model on the RDD2022 China regional road defect dataset. By balancing the calculation parameters and semantic information, the model provides an effective technical solution for road inspection and maintenance, and has important engineering application value for prolonging the service life of roads.