基于改进U-net的大坝表面混凝土裂缝图像分割方法
Segmentation method of dam surface crack image based on improved U-net model
赵普 1谷艳昌 2张大伟 3吴云星2
作者信息
- 1. 南京水利科学研究院,江苏 南京 210029
- 2. 南京水利科学研究院,江苏 南京 210029;水利部大坝安全管理中心,江苏 南京 210029
- 3. 江苏省连云港市水利局,江苏 连云港 222000
- 折叠
摘要
大坝表面混凝土裂缝的检测与识别对大坝安全具有重要意义,为此开展了基于深度学习方法的混凝土裂缝织别方法研究.针对裂缝图像具有复杂拓扑结构和正负样本不平衡等特点,在典型U-net中嵌入了ASPP和CBAM优化模块,同时以Dice + BCE混合损失函数代替了单一交叉熵损失函数.结果表明:所创建的改进U-net 在自制实例大坝裂缝图像数据集上表现优异,交并比 IoU 和 F1 分数分别为 47.05%和62.99%,对比典型U-net分别提高了5.41%和5.19%,对比PSPNet分别提高了 3.05%和 3.31%.改进的U-net在裂缝分割任务中像素分类更精确,多尺度信息更丰富,具有良好的泛化能力和鲁棒性,可为大坝混凝土结构表面裂缝检测与识别提供更优的手段.
Abstract
The detection and identification of surface cracks on dams is of great significance for dam safety,so we study dam surface cracks detection based on deep learning.In view of the complex topological structures and imbalance of positive and nega-tive samples of the crack images,the ASPP and CBAM optimization modules were embedded in the typical U-net model,and a Dice +BCE hybrid loss function was used to replace the single cross entropy loss function.The improved U-net model performed well on a self-made instance dam crack image dataset,with IoU being 47.05%and F1 being 62.99%respectively.Compared with typical U-net model,it had increased by 5.41%and 5.19%,and compared with PSPNet model,it had increased by 3.05%and 3.31%respectively.The improved U-net model provides more accurate pixel classification and richer multi-scale in-formation in crack segmentation tasks,providing a better means for detecting and identifying surface cracks in dam concrete struc-tures.
关键词
混凝土裂缝检测/深度学习/语义分割/U-net模型优化/大坝安全Key words
concrete crack detection/deep learning/semantic segmentation/improved U-net model/dam safety引用本文复制引用
基金项目
国家自然科学基金(51979175)
南京水科院基本科研业务费科研创新团队建设项目(Y722003)
南京水科院基本科研业务费重点项目(Y721005)
南京水科院基本科研业务费重点项目(Y721003)
出版年
2024
国家科技期刊平台
NSTL
维普
万方数据