Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Deep learning‐based automatic classification of three‐level surface information in bridge inspection
AbstractBridge inspection ensures that in‐service bridges are managed and maintained in conformity. To enhance the accuracy and efficiency of bridge inspection, an automatic hierarchical model is proposed, which enables the classification and correlation of bridge surface images at three levels, namely, at the structure, component, and defect type level. Thus, the impact of both the defect types and the affected components on bridge safety can be simultaneously considered. The proposed model uses a group of sub‐models instead of the common flat network to realize the multiple tasks, which is advantageous in accuracy, training simplicity, and scalability. The classification accuracy of the hierarchical model in three levels has reached 96%, 92%, and 81%. Results demonstrate the effectiveness of the proposed method in the classification of multi‐scale targets. This study may provide a new strategy for developing a systematic and easily adaptable detection framework for practical bridge engineering.
Deep learning‐based automatic classification of three‐level surface information in bridge inspection
AbstractBridge inspection ensures that in‐service bridges are managed and maintained in conformity. To enhance the accuracy and efficiency of bridge inspection, an automatic hierarchical model is proposed, which enables the classification and correlation of bridge surface images at three levels, namely, at the structure, component, and defect type level. Thus, the impact of both the defect types and the affected components on bridge safety can be simultaneously considered. The proposed model uses a group of sub‐models instead of the common flat network to realize the multiple tasks, which is advantageous in accuracy, training simplicity, and scalability. The classification accuracy of the hierarchical model in three levels has reached 96%, 92%, and 81%. Results demonstrate the effectiveness of the proposed method in the classification of multi‐scale targets. This study may provide a new strategy for developing a systematic and easily adaptable detection framework for practical bridge engineering.
Deep learning‐based automatic classification of three‐level surface information in bridge inspection
Computer aided Civil Eng
Zhang, He (Autor:in) / Shen, Zhijing (Autor:in) / Lin, Zhenhang (Autor:in) / Quan, Liwei (Autor:in) / Sun, Liangfeng (Autor:in)
Computer-Aided Civil and Infrastructure Engineering ; 39 ; 1431-1451
01.05.2024
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Deep Learning-Based Crack Detection and Classification for Concrete Structures Inspection
| Springer Verlag | 2023