Image-based crack detection methods have been extensively studied due to their cost-effectiveness in terms of data acquisition and processing. However, automated crack detection still remain a challenging task due to complexity of image background and different patterns of cracks. To address these issues, this paper proposes a deep residual network with transfer learning for pixel-level crack detection on road surface images. The network was trained on 71 images of CrackForest dataset and tested on 47 images of it. Experimental results suggest that the deep residual network is superior to the existing algorithms with recall value and precision value of 84.90% and 93.57%, respectively.
|Publication status||Published - 2018|
|Event||35th International Symposium on Automation and Robotics in Construction and International AEC/FM Hackathon: The Future of Building Things, ISARC 2018 - Berlin, Germany|
Duration: 2018 Jul 20 → 2018 Jul 25
|Other||35th International Symposium on Automation and Robotics in Construction and International AEC/FM Hackathon: The Future of Building Things, ISARC 2018|
|Period||18/7/20 → 18/7/25|
Bibliographical noteFunding Information:
This work was supported by a grant (18CTAP-C133290-02) from Infrastructure and transportation technology promotion research Program funded by Ministry of Land, Infrastructure and Transport of Korean government.
All Science Journal Classification (ASJC) codes
- Computer Vision and Pattern Recognition
- Artificial Intelligence
- Building and Construction