基于改进YOLOv8n模型的隧道裂缝检测算法研究OA

With the continuous aging of tunnel structures,tunnel crack detection is of great significance for ensuring the safe operation of tunnels.Aiming at the environmental characteristics of extremely low illumination,low contrast,and high noise in tunnels,this paper proposes a tunnel crack detection method based on the improved YOLOv8n model.Firstly,the attention mechanism module is introduced into the backbone network to enhance the feature extraction capability of the model and the attention to key information,reduce the influence of irrelevant background,and improve the robustness of the neural network.Secondly,aiming at the complex shape changes of cracks,the deformable convolution module DCNv2 is added to the neck structure to flexibly deal with targets of different scales and improve detection accuracy.Finally,the Shape-IoU loss function is introduced to promote more accurate target positioning and improve the overall detection efficiency and accuracy.Experiments show that the improved model reaches 0.919 and 0.860 in mean Average Precision(mAP)and F1-score respectively,which are improved by 2.57%and 3.61%compared with the original YOLOv8n model,and can effectively meet the actual needs of tunnel crack detection.