Recognition Method of Tunnel Lining Defects Based on Deep Learning

The defect identification of tunnel lining is a task with lot tasks and time-consuming work, currently, it mainly relies on manual operation. This paper takes the ground-penetrating radar image internal defects as research object, chooses popular VGG16, ResNet34 convolutional neural network (CNN) to build automatic recognition model for comparative study, proposes an improved defect-recognition model. In this paper, SGD Adam training algorithms are used update parameters, PyTorch depth framework train network. test results show that has faster convergence speed, higher accuracy rate, shorter time than VGG16 using algorithm can achieve 99.08% accuracy. 99.25%, at same, reduce parameter amount by 4.22% compared network, which better identify in lining. shows deep learning method provide new ideas defects.