Abstracts – Browse Results
Click on the titles below to expand the information about each abstract.
Viewing 6 results ...
Chen, C, Zhang, Y, Xiao, B, Cheng, M, Zhang, J and Li, H (2024) Deep learning-based image steganography for visual data cybersecurity in construction management. Journal of Construction Engineering and Management, 150(10).
Liu, H, Zhang, F, Ma, R, Wang, L, Chen, Z, Zhang, Q and Guo, L (2024) Intelligent noncontact structural displacement detection method based on computer vision and deep learning. Journal of Construction Engineering and Management, 150(10).
- Type: Journal Article
- Keywords: computer vision; deep learning; displacement detection; Euclidean distance method based on inverse perspective mapping
- ISBN/ISSN: 0733-9364
- URL: http://doi.org/10.1061/JCEMD4.COENG-14893
- Abstract:
Accurate identification of structural displacements is important for structural state assessment and performance evaluation. This paper proposes a real-time structural displacement detection model based on computer vision and deep learning. The model consists of three stages: identification, tracking, and displacement resolution. First, the displacement target is identified and tracked by the improved YOLO v7 algorithm and the improved DeepSORT algorithm. Then, the Euclidean distance method based on inverse perspective mapping (IPM-ED) is proposed for the analytical conversion of the displacement. Next, the accuracy and effectiveness of this displacement detection model are evaluated through four groups of bamboo axial compression tests. A comparative analysis is conducted between the IPM-ED displacement analysis method and the commonly used ED displacement analysis method. Finally, the robustness of this method is tested by using a cable breakage test of a cable dome structure as an application case. The research results demonstrate that the maximum average error of the four groups of bamboo displacement tests is only 3.10 mm, and the maximum relative error of peak displacement is only 6.54 mm. The RMSE basically stays around 3.5 mm. The maximum displacement error in the application case is only 4.91 mm, with a maximum MAPE of 4.94%. In addition, the error percentage under the IPM-ED algorithm is basically within 5%, while the error percentage of the ED algorithm is more than 10%. The method in this paper achieves efficient and intelligent identification of structural displacements in a non-contact manner. The proposed method is suitable for environments where the contact displacement sensor is easily affected by vibration, the site is complex and requires additional displacement sensor fixing equipment, the displacement sensor with super-high structure is unsafe to deploy, and the contact displacement sensor in narrow space is inconvenient to deploy, so it has broad application prospects.
Nelson, T N T, Poleacovschi, C, Appelgate, M, Drake, R, Swalwell, K, Jackson, C, Svec, J and Cetin, K (2024) Making social justice central to construction engineering: Testing interventions for educating reflexive engineers. Journal of Construction Engineering and Management, 150(10).
Sun, Y, Gheisari, M and Jeelani, I (2024) Robosite: An educational virtual site visit featuring the safe integration of four-legged robots in construction. Journal of Construction Engineering and Management, 150(10).
Xue, H, Zhang, S, Chen, J, Cong, W, Wu, G and Zhao, X (2024) Effects of organizational elements on emerging information and construction management technology implementation in building professionals: Moderating role of top management support. Journal of Construction Engineering and Management, 150(10).
Yan, L, Pan, Y and Chen, Y (2024) Understanding the double-edged sword effect of contract flexibility on contractor's opportunistic behavior in construction project: Moderating role of BIM application degree. Journal of Construction Engineering and Management, 150(10).