Article
Computer Science, Artificial Intelligence
ZhiLong Qi, Donghai Liu, Jinyue Zhang, Junjie Chen
Summary: This paper proposes a three-step method for automatic detection of concrete micro-cracks in underwater structures. The method includes image preprocessing, crack recognition and localization using a convolutional neural network, and precise crack segmentation. The proposed method can effectively detect and localize cracks in underwater optical images, even under low illumination, low signal-to-noise ratio, and low contrast conditions.
MACHINE VISION AND APPLICATIONS
(2022)
Article
Engineering, Electrical & Electronic
Zhang Zhenhai, Ji Kun, Dang Jianwu
Summary: An automatic tunnel crack detection system based on machine vision is presented to effectively detect the surface cracks of subway tunnels. This system can extract complete cracks in complex tunnel environments and meet actual engineering requirements.
JOURNAL OF ELECTRONIC IMAGING
(2021)
Article
Chemistry, Multidisciplinary
Xiaohu Zhang, Haifeng Huang
Summary: Given the significance of convolutional neural network-based attention models in road maintenance, a PSNet method with a Parallel Convolution Module (PCM) and Self-Gated Attention Block (SGAB) was proposed. Experimental results demonstrated competitive segmentation performance for crack detection, showing significant improvements over traditional attention models.
APPLIED SCIENCES-BASEL
(2023)
Article
Construction & Building Technology
Aohui Ouyang, Vanessa Di Murro, Martin Cull, Roddy Cunningham, John Andrew Osborne, Zili Li
Summary: This study presents a remote and automated system for crack monitoring in concrete tunnel linings using robot-mounted imaging technology. The system collects crack images remotely and stitches them together to create a panorama image of the tunnel surface. Transfer learning is employed to optimize the state-of-the-art semantic segmentation model for automatic crack detection. Field trials conducted in tunnels at CERN demonstrate the feasibility and effectiveness of the proposed crack monitoring system.
TUNNELLING AND UNDERGROUND SPACE TECHNOLOGY
(2023)
Article
Construction & Building Technology
Thai Son Tran, Son Dong Nguyen, Hyun Jong Lee, Van Phuc Tran
Summary: Detecting and measuring cracks on a bridge deck is crucial for preventing further damage and ensuring safety. This study proposes a novel deep learning approach for detecting and segmenting cracks on the bridge deck.
CONSTRUCTION AND BUILDING MATERIALS
(2023)
Article
Chemistry, Multidisciplinary
Xiaohu Zhang, Haifeng Huang
Summary: Crack detection is crucial for concrete surface maintenance. Existing deep-learning-based methods have limitations, such as feature information loss for small cracks, lack of global information restoration, edge feature information loss, and interference from stains. To address these issues, we propose a new approach that utilizes a pyramid hierarchical convolution module (PHCM) to extract crack features of different sizes, a mixed global attention module (MGAM) to fuse global information, an edge feature extractor module (EFEM) to learn edge features, and a supplementary attention module (SAM) to handle stain interference. Our proposed PHCNet achieves higher accuracy than traditional convolutional models on multiple datasets.
APPLIED SCIENCES-BASEL
(2023)
Article
Chemistry, Analytical
Xiaofeng Liu, Zenglin Hong, Wei Shi, Xiaodan Guo
Summary: With the increase in urban rail transit construction, crack detection in subway tunnels has become crucial for prolonging their service life and reducing accidents. This paper proposes a new method for crack identification and feature detection using image processing technology and deep learning. Experimental results show that the proposed deep convolutional network algorithm is more accurate and suitable for crack detection in subway tunnels.
Article
Construction & Building Technology
Wenlong Ye, Shijie Deng, Juanjuan Ren, Xueshan Xu, Kaiyao Zhang, Wei Du
Summary: In this study, a fast and effective detection method for concrete cracks on slab tracks using dilated convolution and the watershed algorithm was proposed. The results showed that the STCNet I network had advantages in terms of calculation speed and model robustness, and demonstrated strong generalization ability for concrete cracks.
CONSTRUCTION AND BUILDING MATERIALS
(2022)
Article
Construction & Building Technology
Qimin Gong, Liqiang Zhu, Yaodong Wang, Zujun Yu
Summary: This paper presents an on-board image acquisition system and a robust method for crack detection. The method includes crack enhancing, multi-stage fusion filtering, and improved seed growth algorithm. The proposed method outperforms existing methods for crack detection.
STRUCTURAL CONTROL & HEALTH MONITORING
(2021)
Article
Engineering, Multidisciplinary
Pengfei Shi, Fengting Zhu, Yuanxue Xin, Shen Shao
Summary: In this paper, an automatic pavement crack detection method called U(2)CrackNet is proposed. The method uses an encoding and decoding architecture with a two-level nested U-structure. The experimental results show that U(2)CrackNet can obtain clearer and more continuous cracks.
STRUCTURAL HEALTH MONITORING-AN INTERNATIONAL JOURNAL
(2023)
Article
Construction & Building Technology
Qinghua Han, Xuan Liu, Jie Xu
Summary: This paper proposes an image-based detection and location method for cracks on the surface using unmanned aerial vehicle, which can detect and avoid potential safety accidents caused by cracks. The method consists of steps such as super-pixel segmentation, damage area detection, pixel-level identification, and crack location.
JOURNAL OF BUILDING ENGINEERING
(2022)
Article
Construction & Building Technology
Yulong Chen, Zilong Zhu, Zhijie Lin, Youmei Zhou
Summary: Cracks in building facades, if not corrected immediately, can worsen over time. Automated inspection routines using deep learning technology have been proven effective in civil infrastructures, but limited research exists in the built environment sector. This study focuses on the effectiveness of transfer learning for image classification. Results show that transfer learning achieves better performance compared to the CNN method with the same amount of data input.
Article
Chemistry, Analytical
Jie Wu, Xiaoqian Zhang
Summary: This paper proposes a tunnel crack detection method based on improved Retinex and deep learning. The image enhancement algorithm is used to improve the contrast information, and an improved VGG19 network model is constructed to achieve efficient segmentation of tunnel crack images. The length and width information of the cracks are obtained using a fast parallel-thinning method. Experimental results show that this method has a shorter detection time and higher detection accuracy.
Article
Engineering, Electrical & Electronic
Lixin He, Wangwei Liu, Yiming Li, Handong Wang, Shenjie Cao, Chengying Zhou
Summary: In this work, we propose a new method for crack image detection and segmentation that addresses the issues of poor crack structure detection in complex background conditions and loss of details in segmentation. The method consists of two phases: the coding phase, which uses the channel attention mechanism and crack characteristics to enhance feature extraction, and the decoding phase, which uses the spatial attention mechanism to capture crack edge information and achieve accurate crack positioning through image information fusion. Experimental results show that our method outperforms existing methods in terms of crack segmentation accuracy, with mean intersection over the union ratios of 87.2% and 83.9% on public and self-built datasets, respectively.
Article
Environmental Sciences
Lin Tian, Qingquan Li, Li He, Dejin Zhang, Filiberto Chiabrando
Summary: This study introduces two innovative methods for vision-based tunnel inspection vehicles. The image-range stitching method is used to map sequence images onto a tunnel layout map, solving the mapping equations. The bidirectional heuristic search approach addresses issues with uncertainty and multiplicity in deep learning.
