Crack-ConvT Net: A Convolutional Transformer Network for Crack Segmentation in Underwater Dams
Abstract Crack detection is a critical approach to ensuring the structural health of dams. However, challenges like uneven underwater lighting, sediment interference, and complex backgrounds often hinder traditional detection methods, leading to feature loss and false detections. To address these is...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-06-01
|
| Series: | Complex & Intelligent Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s40747-025-01957-y |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849738232603869184 |
|---|---|
| author | Pengfei Shi Hongzhu Chen Zaiming Geng Xinnan Fan Yuanxue Xin |
| author_facet | Pengfei Shi Hongzhu Chen Zaiming Geng Xinnan Fan Yuanxue Xin |
| author_sort | Pengfei Shi |
| collection | DOAJ |
| description | Abstract Crack detection is a critical approach to ensuring the structural health of dams. However, challenges like uneven underwater lighting, sediment interference, and complex backgrounds often hinder traditional detection methods, leading to feature loss and false detections. To address these issues, this paper proposes Crack-ConvT Net, a U-Shape architecture that integrates Convolutional Neural Networks (CNNs) and Transformers for underwater dam crack segmentation. Firstly, a Global Information Aggregation Block is introduced to enhance the model’s ability to capture the spatial distribution of cracks by leveraging multiscale pooling and channel expansion strategies, which improve global context awareness while preserving high-resolution details. Secondly, to address the limitations of traditional skip connections in crack segmentation under complex environments, an Adaptive Feature Fusion Module is designed to optimize the interaction and integration of multi-level features. Finally, a Deep Prediction Head is developed, incorporating cascaded $$3\times 3$$ 3 × 3 convolutions and Leaky ReLU activation functions to enhance the network’s capacity for modeling intricate crack features. Experimental results demonstrate that the proposed framework significantly outperforms existing methods in underwater dam crack segmentation, effectively improving segmentation accuracy under noise interference. |
| format | Article |
| id | doaj-art-aec9dfc1160d4c44b2d1aab934832c32 |
| institution | DOAJ |
| issn | 2199-4536 2198-6053 |
| language | English |
| publishDate | 2025-06-01 |
| publisher | Springer |
| record_format | Article |
| series | Complex & Intelligent Systems |
| spelling | doaj-art-aec9dfc1160d4c44b2d1aab934832c322025-08-20T03:06:40ZengSpringerComplex & Intelligent Systems2199-45362198-60532025-06-0111811410.1007/s40747-025-01957-yCrack-ConvT Net: A Convolutional Transformer Network for Crack Segmentation in Underwater DamsPengfei Shi0Hongzhu Chen1Zaiming Geng2Xinnan Fan3Yuanxue Xin4College of Artificial Intelligence and Automation, Hohai UniversityCollege of Information Science and Engineering, Hohai UniversityChina Yangtze River Electric Power Co., LtdCollege of Information Science and Engineering, Hohai UniversityCollege of Information Science and Engineering, Hohai UniversityAbstract Crack detection is a critical approach to ensuring the structural health of dams. However, challenges like uneven underwater lighting, sediment interference, and complex backgrounds often hinder traditional detection methods, leading to feature loss and false detections. To address these issues, this paper proposes Crack-ConvT Net, a U-Shape architecture that integrates Convolutional Neural Networks (CNNs) and Transformers for underwater dam crack segmentation. Firstly, a Global Information Aggregation Block is introduced to enhance the model’s ability to capture the spatial distribution of cracks by leveraging multiscale pooling and channel expansion strategies, which improve global context awareness while preserving high-resolution details. Secondly, to address the limitations of traditional skip connections in crack segmentation under complex environments, an Adaptive Feature Fusion Module is designed to optimize the interaction and integration of multi-level features. Finally, a Deep Prediction Head is developed, incorporating cascaded $$3\times 3$$ 3 × 3 convolutions and Leaky ReLU activation functions to enhance the network’s capacity for modeling intricate crack features. Experimental results demonstrate that the proposed framework significantly outperforms existing methods in underwater dam crack segmentation, effectively improving segmentation accuracy under noise interference.https://doi.org/10.1007/s40747-025-01957-yUnderwater damCrack segmentationSkip connectionFeature extractionInformation aggregation |
| spellingShingle | Pengfei Shi Hongzhu Chen Zaiming Geng Xinnan Fan Yuanxue Xin Crack-ConvT Net: A Convolutional Transformer Network for Crack Segmentation in Underwater Dams Complex & Intelligent Systems Underwater dam Crack segmentation Skip connection Feature extraction Information aggregation |
| title | Crack-ConvT Net: A Convolutional Transformer Network for Crack Segmentation in Underwater Dams |
| title_full | Crack-ConvT Net: A Convolutional Transformer Network for Crack Segmentation in Underwater Dams |
| title_fullStr | Crack-ConvT Net: A Convolutional Transformer Network for Crack Segmentation in Underwater Dams |
| title_full_unstemmed | Crack-ConvT Net: A Convolutional Transformer Network for Crack Segmentation in Underwater Dams |
| title_short | Crack-ConvT Net: A Convolutional Transformer Network for Crack Segmentation in Underwater Dams |
| title_sort | crack convt net a convolutional transformer network for crack segmentation in underwater dams |
| topic | Underwater dam Crack segmentation Skip connection Feature extraction Information aggregation |
| url | https://doi.org/10.1007/s40747-025-01957-y |
| work_keys_str_mv | AT pengfeishi crackconvtnetaconvolutionaltransformernetworkforcracksegmentationinunderwaterdams AT hongzhuchen crackconvtnetaconvolutionaltransformernetworkforcracksegmentationinunderwaterdams AT zaiminggeng crackconvtnetaconvolutionaltransformernetworkforcracksegmentationinunderwaterdams AT xinnanfan crackconvtnetaconvolutionaltransformernetworkforcracksegmentationinunderwaterdams AT yuanxuexin crackconvtnetaconvolutionaltransformernetworkforcracksegmentationinunderwaterdams |