A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Investigation of pavement crack detection based on deep learning method using weakly supervised instance segmentation framework
Abstract This paper presents efficient and cost-effective methods to identify pavement crack distress and thereby substantially increase pavement strength. Detecting the origin of this distress is the key to restoring pavement performance. To do that, a deep learning method is used to detect cracks based on the weakly supervised instance segmentation (WSIS) framework. A bounding box-level crack image data is preprocessed. Pseudo labels are generated by a region growing algorithm and a GrabCut algorithm. Another important contribution is a new dynamically balanced binary cross-entropy loss function. Results show that the WSIS framework reduces manual marking and has a high recognition accuracy of crack distress.
Highlights Improved segmentation framework model developed to detect pavement cracks. Bounding box labeled dataset is used to replace pixel-level labeled dataset. The method proposed in the paper significantly reduces the work of data annotation. The dynamically balanced binary cross-entropy loss function solves the imbalance of positive and negative samples. Proposed method reaches 94.6% accuracy of supervised learning.
Investigation of pavement crack detection based on deep learning method using weakly supervised instance segmentation framework
Abstract This paper presents efficient and cost-effective methods to identify pavement crack distress and thereby substantially increase pavement strength. Detecting the origin of this distress is the key to restoring pavement performance. To do that, a deep learning method is used to detect cracks based on the weakly supervised instance segmentation (WSIS) framework. A bounding box-level crack image data is preprocessed. Pseudo labels are generated by a region growing algorithm and a GrabCut algorithm. Another important contribution is a new dynamically balanced binary cross-entropy loss function. Results show that the WSIS framework reduces manual marking and has a high recognition accuracy of crack distress.
Highlights Improved segmentation framework model developed to detect pavement cracks. Bounding box labeled dataset is used to replace pixel-level labeled dataset. The method proposed in the paper significantly reduces the work of data annotation. The dynamically balanced binary cross-entropy loss function solves the imbalance of positive and negative samples. Proposed method reaches 94.6% accuracy of supervised learning.
Investigation of pavement crack detection based on deep learning method using weakly supervised instance segmentation framework
Zhang, Hancheng (author) / Qian, Zhendong (author) / Tan, Yunfeng (author) / Xie, Yuxin (author) / Li, Miaocheng (author)
2022-09-08
Article (Journal)
Electronic Resource
English
Multi-region Segmentation Pavement Crack Detection Method Based on Deep Learning
| Springer Verlag | 2025
Multi-region Segmentation Pavement Crack Detection Method Based on Deep Learning
| Springer Verlag | 2025