Efficient, Low-cost Bridge Cracking Detection and Quantification Using Deep-learning and UAV Images

A large number of bridges in the State of Louisiana and the United States are working under serious degradation conditions where cracks on bridges threaten the structural integrity and public security. To ensure the structural integrity and public security, it is required that bridges in the US be inspected and rated every two years. Currently, this biannual assessment is largely implemented using manual visual inspection methods, which is slow and costly. In addition, it is challenging for workers to detect cracks in regions that are hard to reach, e.g., top part of bridge tower, cables, mid-span of the bridge girders and decks. It is possible that there will be cracks undetected during inspection, which might cause bridge to collapse when the undetected damage on load-carrying members is beyond the critical level. A well-known catastrophic example is the I-35 Bridge that collapsed in Minneapolis during the summer of 2007. As unmanned aerial vehicles (UAVs) become more and more popular, researchers started to resort to images and videos from places which are hard to reach. Especially for bridges, UAVs can quickly fly to the desired locations to take images and videos. Hence, it is promising to integrate the deep learning method with UAV images to develop an automatic crack damage identification method. This research will develop an efficient low-cost deep learning-based framework to detect and quantify cracks on bridges using computer vision-based technique and deep learning. The Convolutional Neural Networks (CNN) deep learning method which is powerful in extracting and learning image features will be used to identify cracks from images. Specific research activities include: (1) collection of a large volume of images from the Internet with subsequent categorization into five classes (intact surfaces, cracks, multiple joints and edges, single joint or edge, etc.); (2) collection of images of target structures using a UAV camera; (3) development of a deep CNN model using collected images and their augmentation; and (4) identification of cracks using the learned deep learning model. The research outcomes of this project will enable automatic crack damage detection and quantification of bridge key components in a cost effective manner. The methodology to be developed is expected to facilitate crack damage identification for other transportation infrastructures, e.g. pavement and traffic sign structures. Essentially, this research is an inter-disciplinary subject that involves expertise from structural engineering, image processing, deep learning and statistical analysis. In addition, undergraduate and graduate students will be directly engaged in the project to develop workforce for the transportation community. On top of that, research data from the project will be incorporated into a relevant course that the PI is teaching. In the later phase of the project, the new methodology to be developed will be standardized and applied for crack damage detection of real bridges in the State of Louisiana and across the country through collaboration with engineers/researchers from local transportation industry and national Department of Transportation.

• Record URL:
  https://transet.lsu.edu/research-in-progress/
• Supplemental Notes:
  • 20STLSU12

Language

• English

Project

• Status: Completed
• Funding: $ 90165
• Contract Numbers:

  69A3551747106

• Sponsor Organizations:

  Office of the Assistant Secretary for Research and Technology

  University Transportation Centers Program
  Department of Transportation
  Washington, DC  United States  20590
• Managing Organizations:

  Transportation Consortium of South-Central States (Tran-SET)

  Louisiana State University
  Baton Rouge, LA  United States  70803
• Project Managers:

  Mousa, Momen

• Performing Organizations:

  Louisiana State University

  3660G Patrick F. Taylor Hall
  Civil and Environmental Engineering
  Baton Rouge, LA  United States  70803
• Principal Investigators:

  Sun, Chao

• Start Date: 20200801
• Expected Completion Date: 20220201
• Actual Completion Date: 0
• USDOT Program: University Transportation Centers Program

Subject/Index Terms

• TRT Terms: Bridges; Cracking; Detection and identification technologies; Drones; Image processing; Machine learning; Neural networks
• Geographic Terms: Louisiana
• Subject Areas: Bridges and other structures; Highways; Maintenance and Preservation;

Filing Info

• Accession Number: 01757500
• Record Type: Research project
• Source Agency: Transportation Consortium of South-Central States (Tran-SET)
• Contract Numbers: 69A3551747106
• Files: UTC, RIP
• Created Date: Nov 10 2020 3:50PM