基于深度学习的混凝土结构表面裂缝检测

Surface crack detection of concrete structures based on deep learning

混凝土结构裂缝的结构安全问题是目前人们关注的热点,而裂缝是影响建筑物安全的重要因素之一。所以提出了1种基于深度学习的混凝土结构表面裂缝检测方法。首先运用了数据增强的方式丰富了裂缝图像数据集,再利用深度学习技术,选用DeconvNet反卷积网络模型进行模型的训练学习。将图像拼接技术与深度学习相结合应用在混凝土结构裂缝检测中,并且对图像中的裂缝特征进行分析,测量出裂缝的角度、宽度和长度。结果显示,测试集的识别率达到了71.17%,准确率达到了97.92%,使用数据增强增加了模型的泛化能力,使模型具有更好的识别效果,能够对混凝土结构表面裂缝进行完整还原。

The problem of structural safety of concrete structure crack is a hot topic at present,and crack is one of the important factors affecting the safety of buildings.Therefore,it proposes a method of surface crack detection of concrete structure based on deep learning.Firstly,the data enhancement method is used to enrich the crack image data set,and then the DeconvNet deconvolution network model is used to carry out the training and learning of the model.The image splicing technology is combined with deep learning to detect cracks in concrete structures,and the crack characteristics in the image are analyzed to measure the angle,width and length of cracks.The results showed that the recognition rate of the test set reached 71.17%,and the accuracy rate reached 97.92%.The data enhancement increased the generalization ability of the model,so that the model had better recognition effect and could completely restore the surface cracks of the concrete structure.