Cracklab: A high-precision and efficient concrete crack segmentation and quantification network. (December 2022)

• Record Type: Journal Article
• Authors: Yu, Zhenwei; Shen, Yonggang; Sun, Zhilin; Chen, Jiang; Gang, Wu
• Is Part Of: Developments in the built environment. Volume 12(2022)
• Publication Date: 2022-12
• Subjects: Crack segmentation -- Real-time detection -- Width measurement -- Complex background filtering -- Deep learning; Civil engineering -- Periodicals; Sustainable construction -- Periodicals; 624.05
• DOI: 10.1016/j.dibe.2022.100088
• Languages: English
• ISSNs: 2666-1659

Abstract

Abstract: A deep learning model named Cracklab for pixel level segmentation and measurement of concrete cracks is proposed. Cracklab excels at handling cracks at image edges and enhances scale adaptability, and it reduces physical occupation and increases efficiency by pruning. A crack database containing 685 images collected from complex scenes is established, with resolution ranging from 1024 to 2048 pixels. During training, focal loss is used to improve the recognition effect of complex background images. The results show that our method is more effective and accurate in detecting and quantifying cracks. Cracklab performed better in inference and comparative experiments, with mIoU at least 0.114 ahead and detection speeds 3.9 times faster compared to works that did not include pruning and scale adaptation. The proposed improved medial axis transform method has an error of 2.09 pixels at the maximum crack width, which is 19.6% lower than another work using distance transform method. Highlights: A new method is proposed for pixel level crack detection and quantification. Cracklab strengthens the image edge detection effect and scale adaptive ability. Conducting comparative studies to check the performance of Cracklab. The pruned Cracklab-tiny meets real-time requirements with good accuracy. The error of the proposed method in measuring the maximum width is 2.09 pixels.