Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Deep Convolutional Neural Network for Segmentation and Classification of Structural Multi-branch Cracks
https://orcid.org/http://orcid.org/0000-0001-9627-4935
https://orcid.org/http://orcid.org/0000-0003-3927-8598
https://orcid.org/http://orcid.org/0000-0003-4230-7580
https://orcid.org/http://orcid.org/0000-0002-8480-0595
Structural Health Monitoring (SHM) has been a significant research topic to help with damage detection in civil structures and to stop further deterioration. Traditional methods of SHM are time consuming and cost ineffective. In addition, civil structures such as dams and high raised buildings are burdensome and risky to inspect manually, especially after a natural disaster. Crack signals the beginning of failure for any structure. Most of the existing methods largely deal with only the detection of cracks. Proposed work concentrates on segmentation, classification, and subsequent detection of cracks based on pattern i.e., Linear vs branching, apart from the single and multiple cracks. The image dataset was obtained from real-time visual inspections. This study is significant because a branching crack shows greater structural stress than a linear crack. Furthermore, results quantify the damage in the image using instance segmentation techniques. Experimental analysis achieves classification and quantification of the data with good accuracy.
Deep Convolutional Neural Network for Segmentation and Classification of Structural Multi-branch Cracks
https://orcid.org/http://orcid.org/0000-0001-9627-4935
https://orcid.org/http://orcid.org/0000-0003-3927-8598
https://orcid.org/http://orcid.org/0000-0003-4230-7580
https://orcid.org/http://orcid.org/0000-0002-8480-0595
Structural Health Monitoring (SHM) has been a significant research topic to help with damage detection in civil structures and to stop further deterioration. Traditional methods of SHM are time consuming and cost ineffective. In addition, civil structures such as dams and high raised buildings are burdensome and risky to inspect manually, especially after a natural disaster. Crack signals the beginning of failure for any structure. Most of the existing methods largely deal with only the detection of cracks. Proposed work concentrates on segmentation, classification, and subsequent detection of cracks based on pattern i.e., Linear vs branching, apart from the single and multiple cracks. The image dataset was obtained from real-time visual inspections. This study is significant because a branching crack shows greater structural stress than a linear crack. Furthermore, results quantify the damage in the image using instance segmentation techniques. Experimental analysis achieves classification and quantification of the data with good accuracy.
Deep Convolutional Neural Network for Segmentation and Classification of Structural Multi-branch Cracks
https://orcid.org/http://orcid.org/0000-0001-9627-4935
https://orcid.org/http://orcid.org/0000-0003-3927-8598
https://orcid.org/http://orcid.org/0000-0003-4230-7580
https://orcid.org/http://orcid.org/0000-0002-8480-0595
Structural Health Monitoring (SHM) has been a significant research topic to help with damage detection in civil structures and to stop further deterioration. Traditional methods of SHM are time consuming and cost ineffective. In addition, civil structures such as dams and high raised buildings are burdensome and risky to inspect manually, especially after a natural disaster. Crack signals the beginning of failure for any structure. Most of the existing methods largely deal with only the detection of cracks. Proposed work concentrates on segmentation, classification, and subsequent detection of cracks based on pattern i.e., Linear vs branching, apart from the single and multiple cracks. The image dataset was obtained from real-time visual inspections. This study is significant because a branching crack shows greater structural stress than a linear crack. Furthermore, results quantify the damage in the image using instance segmentation techniques. Experimental analysis achieves classification and quantification of the data with good accuracy.
Deep Convolutional Neural Network for Segmentation and Classification of Structural Multi-branch Cracks
Lecture Notes in Civil Engineering
Rizzo, Piervincenzo (Herausgeber:in) / Milazzo, Alberto (Herausgeber:in) / Kandula, Himavanth (Autor:in) / Koduri, Hrushith Ram (Autor:in) / Kalapatapu, Prafulla (Autor:in) / Pasupuleti, Venkata Dilip Kumar (Autor:in)
European Workshop on Structural Health Monitoring ; 2022 ; Palermo, Italy
16.06.2022
9 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Deep convolutional neural networks for semantic segmentation of cracks
| Wiley | 2022
Segmentation of micro-cracks in fractured coal based on convolutional neural network
| DOAJ | 2022
Gemstone Classification Using Deep Convolutional Neural Network
| Springer Verlag | 2024
Gemstone Classification Using Deep Convolutional Neural Network
| Springer Verlag | 2024