eduzhai > Applied Sciences > Engineering >

Automatic Crack Detection on Road Pavements Using Encoder Decoder Architecture

  • king
  • (0) Download
  • 20210505
  • Save

... pages left unread,continue reading

Document pages: 18 pages

Abstract: Inspired by the development of deep learning in computer vision and objectdetection, the proposed algorithm considers an encoder-decoder architecturewith hierarchical feature learning and dilated convolution, namedU-Hierarchical Dilated Network (U-HDN), to perform crack detection in anend-to-end method. Crack characteristics with multiple context information areautomatically able to learn and perform end-to-end crack detection. Then, amulti-dilation module embedded in an encoder-decoder architecture is proposed.The crack features of multiple context sizes can be integrated into themulti-dilation module by dilation convolution with different dilatation rates,which can obtain much more cracks information. Finally, the hierarchicalfeature learning module is designed to obtain a multi-scale features from thehigh to low-level convolutional layers, which are integrated to predictpixel-wise crack detection. Some experiments on public crack databases using118 images were performed and the results were compared with those obtainedwith other methods on the same images. The results show that the proposed U-HDNmethod achieves high performance because it can extract and fuse differentcontext sizes and different levels of feature maps than other algorithms.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×