基于改进YOLOv8的混凝土裂缝检测算法

打开文本图片集

中图分类号：TU17 文献标志码：A 文章编号：1008-0562（2025）05-0590-07

Concrete crack detection algorithm based on improved YOLOv8

ZHAO Wenhua，LIU Aopeng，DU Changbo，SUNQi，KONG Jiahui （SchoolofCivilEngineering，Liaoning TechnicalUniversity，Fuxin123ooo，China）

Abstract：In order to realize the accurate and eficient detection ofconcretecracks，aconcretecrack detection model YOLO-CCD based on YOLOv8 s is proposed.The multi-scale convolution module PSConv （poly-scale convolution）is introducedtoenhance the learmingabilityofcros-scale features and improve the detection effect of small cracks and cracks in complex backgrounds.The efficient channel atention （ECA） mechanism is used to enhance the dependence between feature channels and optimize feature representation.The SIoU loss function is introduced tooptimize the bounding box regression process by comprehensively considering the geometric features，so as to improve the detection accuracy of the model.Compared with the YOLOv8 s model，the average accuracy mAP50 and mAP50-95 of the improved model are increased by 7.9% and 2.4% ，respectively. Compared with other detection methods，the model proposed in this paper has significant advantages in detection accuracy and computational eficiency.The research conclusion provides a new feasible method for concrete crack detection.

Keywords：objectdetection;YOLOv8;concretecrack;concrete crack detection;efficientchannelattention;poly scale convolution; loss function

0引言

随着城市化的不断推进，混凝土已经成为建筑工程中应用最广泛的材料之一]。（剩余12095字）