| Crack detection on concrete bridges is a critical task to ensure bridge safety.Detecting the surface cracks on bridges could monitor the whole condition of bridges timely and effectively.Then a repair performed in time will significantly increase the bridge longevity.So the research of automatic detecting t echnology possesses economic and practical values on bridge crack detection.In practice,many cracks on concrete bridges show low contrast,bad continuity and blurry edges,which brings challenges to the automatic crack detection based on traditional algorithms.On the other hand,because a large number of images are acquired by bridge inspection robots,it is of great significance to propose an image-based automatic algorithm applied in practice for fast crack detection.So it is an important maintenance research to improve the detection accuracy of blurred cracks and realize the fast crack detection in the field of crack identification and detection.In this paper,to improve the detection accuracy of blurred cracks,we propose the HDCB-Net — a deep learning-based network with the hybrid dilated convolutional block(HDCB)for pixel-level crack detection.Specifically,HDCB is employed to expand the receptive field of the convolution kernel without increasing the computational complexity and to avoid the gridding effect generated by the dilated convolution.Finally,to achieve a reasonable efficiency/accuracy trade-off,the HDCB-Net only contains a few downsampling stages,which can avoid the loss of blurred crack pixels due to excessive downsampling.Meanwhile,in order to realize fast crack detection in a massive number of images with the high resolution(5120×5120 pixels)and make our algorithm could be applied to real-time crack detection,we propose a fast detecting algorithm based on the coarse and fine combined strategy.Specifically,at the first stage,the YOLOv4(You Only Look Once v4)is employed to filter out many images without cracks and generate coarse region proposals.At the second stage,to achieve refined damage analysis,the HDCB-Net is used to detect fine pixel-level cracks from the coarse region proposals.To achieve a reasonable efficiency/ accuracy trade-off,images with the high resolution are input into networks by the overlapping sliding window at the whole stage.We test more than 100,000 bridge images.The experimental results demonstrate that the proposed HDCB-Net is genetic and able to improve the detection accuracy of blurred cracks,and our two-stage algorithm is efficient for fast crack detection.The whole detection process takes only 0.64 seconds to handle a single image with5120×5120 pixels. |
