Abstract:The crack is a common pavement failure problem. A lack of periodic maintenance will result in extending the cracks and damage the pavement, which will affect the normal use of the road. Therefore, it is significant to establish an efficient intelligent identification model for pavement cracks. The neural network is a method of simulating animal nervous systems using gradient descent to predict results by learning a weight matrix. It has been widely used in geotechnical engineering, computer vision, medicine, and other fields. However, there are three major problems in the application of neural networks to crack identification. There are too few layers, extracted crack features are not complete, and the method lacks the efficiency to calculate the whole picture. In this study, a fully convolutional neural network based on ResNet-101 is used to establish an intelligent identification model of pavement crack regions. This method, using a convolutional layer instead of a fully connected layer, realizes full convolution and accelerates calculation. The region proposals come from the feature map at the end of the base network, which avoids multiple computations of the same picture. Online hard example mining and data-augmentation techniques are adopted to improve the model's recognition accuracy. We trained and tested Concrete Crack Images for Classification (CCIC), which is a public dataset collected using smartphones, and the Crack Image Database (CIDB), which was automatically collected using vehicle-mounted charge-coupled device cameras, with identification accuracy reaching 91.4% and 86.4%, respectively. The proposed model has a higher recognition accuracy and recall rate than Faster RCNN and different depth models, and can extract more complete and accurate crack features in CIDB. We also analyzed translation processing, fuzzy, scaling, and distorted images. The proposed model shows a strong robustness and stability, and can automatically identify image cracks of different forms. It has broad application prospects in practical engineering problems.

Deep learning-based planar crack damage evaluation using convolutional neural networks

Ultrasonic Array Tomography-Oriented Subsurface Crack Recognition and Cross-Section Image Reconstruction of Reinforced Concrete Structure Using Deep Neural Networks

Structural Crack Detection Using Deep Learning–based Fully Convolutional Networks

A Deep Learning-Based Approach for Crack Damage Detection Using Strain Field

Structural Damage-Causing Concrete Cracking Detection Based on a Deep-Learning Method

Investigations on Plate Crack Damage Detection Using Convolutional Neural Networks

A Deep Learning Method for Pavement Crack Identification Based on Limited Field Images

Application of Deep Convolution Neural Network in Crack Identification

Deep neural networks for crack detection inside structures

Crack Length Measurement Using Convolutional Neural Networks and Image Processing

An Intelligent Crack Damage Assessment Method by Integrating Information and Physics

Enhancing Structural Crack Detection through a Multiscale Multilevel Mask Deep Convolutional Neural Network and Line Similarity Index

Automatic Crack Detection and Calculation Based on Deep Learning and Digital Image Processing

Intelligent Detection Model Based on a Fully Convolutional Neural Network for Pavement Cracks

Computational deep learning algorithm (vision/frequency response)-based damage detection in engineering structure

A CNN-Based Crack Analysis Model for Metal Specimens

Deep learning-based three-dimensional crack damage detection method using point clouds without color information

DeepCrack: Learning Hierarchical Convolutional Features for Crack Detection

Complex Crack Segmentation and Quantitative Evaluation of Engineering Materials Based on Deep Learning Methods

Deep Convolutional Neural Network Probability Imaging for Plate Structural Health Monitoring Using Guided Waves