Automated Detection for Concrete Surface Cracks Based on Deeplabv3+ BDF
Concrete cracks have always been the focus of research because of the serious damage they cause to structures. With the updating of hardware and algorithms, the detection of concrete structure surface cracks based on computer vision has received extensive attention. This paper proposes an improved a...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-01-01
|
| Series: | Buildings |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-5309/13/1/118 |
| _version_ | 1797445057449558016 |
|---|---|
| author | Yonggang Shen Zhenwei Yu Chunsheng Li Chao Zhao Zhilin Sun |
| author_facet | Yonggang Shen Zhenwei Yu Chunsheng Li Chao Zhao Zhilin Sun |
| author_sort | Yonggang Shen |
| collection | DOAJ |
| description | Concrete cracks have always been the focus of research because of the serious damage they cause to structures. With the updating of hardware and algorithms, the detection of concrete structure surface cracks based on computer vision has received extensive attention. This paper proposes an improved algorithm based on the open-source model Deeplabv3+ and names it Deeplabv3+ BDF according to the optimization strategy used. Deeplabv3+ BDF first replaces the original backbone Xception with MobileNetv2 and further replaces all standard convolutions with depthwise separable convolutions (DSC) to achieve a light weight. The feature map of a shallow convolution layer is additionally fused to improve the detail segmentation effect. A new strategy is proposed, which is different from the two-stage training. The model training is carried out in the order of transfer learning, coarse-annotation training and fine-annotation training. The comparative test results show that Deeplabv3+ BDF showed good performance in the validation set and achieved the highest mIoU and detection efficiency, reaching real-time and accurate detection. |
| first_indexed | 2024-03-09T13:20:17Z |
| format | Article |
| id | doaj.art-47db6cf7a9d94cda92d3ff85940b1a11 |
| institution | Directory Open Access Journal |
| issn | 2075-5309 |
| language | English |
| last_indexed | 2024-03-09T13:20:17Z |
| publishDate | 2023-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Buildings |
| spelling | doaj.art-47db6cf7a9d94cda92d3ff85940b1a112023-11-30T21:30:08ZengMDPI AGBuildings2075-53092023-01-0113111810.3390/buildings13010118Automated Detection for Concrete Surface Cracks Based on Deeplabv3+ BDFYonggang Shen0Zhenwei Yu1Chunsheng Li2Chao Zhao3Zhilin Sun4College of Civil Engineering and Architecture, Zhejiang University, Hangzhou 310058, ChinaCollege of Civil Engineering and Architecture, Zhejiang University, Hangzhou 310058, ChinaZhejiang Communications Construction Group Co., Ltd., Hangzhou 310051, ChinaZhejiang Communications Construction Group Co., Ltd., Hangzhou 310051, ChinaCollege of Civil Engineering and Architecture, Zhejiang University, Hangzhou 310058, ChinaConcrete cracks have always been the focus of research because of the serious damage they cause to structures. With the updating of hardware and algorithms, the detection of concrete structure surface cracks based on computer vision has received extensive attention. This paper proposes an improved algorithm based on the open-source model Deeplabv3+ and names it Deeplabv3+ BDF according to the optimization strategy used. Deeplabv3+ BDF first replaces the original backbone Xception with MobileNetv2 and further replaces all standard convolutions with depthwise separable convolutions (DSC) to achieve a light weight. The feature map of a shallow convolution layer is additionally fused to improve the detail segmentation effect. A new strategy is proposed, which is different from the two-stage training. The model training is carried out in the order of transfer learning, coarse-annotation training and fine-annotation training. The comparative test results show that Deeplabv3+ BDF showed good performance in the validation set and achieved the highest mIoU and detection efficiency, reaching real-time and accurate detection.https://www.mdpi.com/2075-5309/13/1/118damage detectionnon-destructive evaluationdeep learningconcrete structurecrack segmentation |
| spellingShingle | Yonggang Shen Zhenwei Yu Chunsheng Li Chao Zhao Zhilin Sun Automated Detection for Concrete Surface Cracks Based on Deeplabv3+ BDF Buildings damage detection non-destructive evaluation deep learning concrete structure crack segmentation |
| title | Automated Detection for Concrete Surface Cracks Based on Deeplabv3+ BDF |
| title_full | Automated Detection for Concrete Surface Cracks Based on Deeplabv3+ BDF |
| title_fullStr | Automated Detection for Concrete Surface Cracks Based on Deeplabv3+ BDF |
| title_full_unstemmed | Automated Detection for Concrete Surface Cracks Based on Deeplabv3+ BDF |
| title_short | Automated Detection for Concrete Surface Cracks Based on Deeplabv3+ BDF |
| title_sort | automated detection for concrete surface cracks based on deeplabv3 bdf |
| topic | damage detection non-destructive evaluation deep learning concrete structure crack segmentation |
| url | https://www.mdpi.com/2075-5309/13/1/118 |
| work_keys_str_mv | AT yonggangshen automateddetectionforconcretesurfacecracksbasedondeeplabv3bdf AT zhenweiyu automateddetectionforconcretesurfacecracksbasedondeeplabv3bdf AT chunshengli automateddetectionforconcretesurfacecracksbasedondeeplabv3bdf AT chaozhao automateddetectionforconcretesurfacecracksbasedondeeplabv3bdf AT zhilinsun automateddetectionforconcretesurfacecracksbasedondeeplabv3bdf |