重庆科技大学学报（自然科学版）2025，Vol.27Issue(4)：69-76,109,9.DOI:10.19406/j.issn.2097-4531.2025.04.008
基于改进YOLOv11n的下水管道裂缝识别算法
Crack Recognition Algorithm for Sewer Pipes Based on Improved YOLOv11n
摘要
Abstract
To address the issue of insufficient detection accuracy of cracks in sewer pipes,a crack recognition algo-rithm based on an improved YOLOv11n for sewer pipelines is proposed.Firstly,the C3K2-MSCB2 module is de-signed to effectively resolve the information bottleneck problem in feature extraction and significantly reduce the pre-cision loss.Then,the coordinate attention mechanism is introduced to optimize the traditional SPPF layer,thereby enhancing the model's recognition performance.Finally,the Concat module in the feature fusion network is re-placed with the BiFPN module,ensuring the stability of detection accuracy while reducing the model complexity.The experimental results indicate that,compared to the YOLOv11n model,the improved model has increased its PmA@50 of increased from 82.9%to 87.3%,and PmA@50∶95 from 45.2%to 49.1%,demonstrating enhanced accura-cy in object detection.In addition,the confidence level of the improved model has also shown significant improve-ment,thoroughly validating its stability and reliability in complex environments.关键词
目标检测/下水管道/C3K2-MSCB2模块/坐标注意力机制/BiFPN模块Key words
object detection/sewer pipes/C3K2-MSCB2 module/coordinate attention mechanism/BiFPN module分类
信息技术与安全科学引用本文复制引用
邱丹丹,郑少峰..基于改进YOLOv11n的下水管道裂缝识别算法[J].重庆科技大学学报（自然科学版）,2025,27(4):69-76,109,9.基金项目
福建省科技厅引导性项目"基于AI的管道特种机器人系列产品研发"(2021H0025) （2021H0025）
福建省教育厅研究生教研项目"基于'新工科'实验室的产学研用深度融合研究"(FBJG20220116) （FBJG20220116）
