Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Deep learning-based crack segmentation for civil infrastructure: data types, architectures, and benchmarked performance
Abstract This paper reviews recent developments in deep learning-based crack segmentation methods and investigates their performance under the impact from different image types. Publicly available datasets and commonly adopted performance evaluation metrics are also summarized. Moreover, an image dataset, namely the Fused Image dataset for convolutional neural Network based crack Detection (FIND), was released to the public for deep learning analysis. The FIND dataset consists of four different image types including raw intensity image, raw range (i.e., elevation) image, filtered range image, and fused image by combining the raw intensity and raw range image. To validate and demonstrate the performance boost through data fusion, a benchmark study is performed to compare the performance of nine (9) established convolutional neural network architectures trained and tested on the FIND dataset; furthermore, through the cross comparison, the optimal architectures and image types can be determined, offering insights to future studies and applications.
Highlights A review in deep learning-based crack segmentation is conducted via benchmark study. A large-scale image dataset (FIND) is used for the benchmark study and released to the public. The use of the fused raw image leads to the highest crack segmentation performance. UNet-FCN and CrackFusionNet demonstrate better accuracy and efficiency for crack segmentation. Precision-Recall Curves provide a more sensitive measure of segmentation performance for imbalanced datasets.
Deep learning-based crack segmentation for civil infrastructure: data types, architectures, and benchmarked performance
Abstract This paper reviews recent developments in deep learning-based crack segmentation methods and investigates their performance under the impact from different image types. Publicly available datasets and commonly adopted performance evaluation metrics are also summarized. Moreover, an image dataset, namely the Fused Image dataset for convolutional neural Network based crack Detection (FIND), was released to the public for deep learning analysis. The FIND dataset consists of four different image types including raw intensity image, raw range (i.e., elevation) image, filtered range image, and fused image by combining the raw intensity and raw range image. To validate and demonstrate the performance boost through data fusion, a benchmark study is performed to compare the performance of nine (9) established convolutional neural network architectures trained and tested on the FIND dataset; furthermore, through the cross comparison, the optimal architectures and image types can be determined, offering insights to future studies and applications.
Highlights A review in deep learning-based crack segmentation is conducted via benchmark study. A large-scale image dataset (FIND) is used for the benchmark study and released to the public. The use of the fused raw image leads to the highest crack segmentation performance. UNet-FCN and CrackFusionNet demonstrate better accuracy and efficiency for crack segmentation. Precision-Recall Curves provide a more sensitive measure of segmentation performance for imbalanced datasets.
Deep learning-based crack segmentation for civil infrastructure: data types, architectures, and benchmarked performance
Zhou, Shanglian (Autor:in) / Canchila, Carlos (Autor:in) / Song, Wei (Autor:in)
18.11.2022
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Civil infrastructure defect assessment using pixel-wise segmentation based on deep learning
| BASE | 2023
| Springer Verlag | 2022
| Elsevier | 2025