Crack Width Estimation for Efficient Scanning of Large Concrete Structures

As Unmanned Aerial Vehicles (UAVs) become increasingly accessible, they pave the way for monitoring and inspecting civil infrastructure in complex and dynamic settings. This study proposes an autonomous crack width estimation method for large concrete structures using segmentation by convolutional neural network and post-image processing. The primary goals include achieving a precise crack width estimation. The suggested solution is incorporated and tested using our collected image databases, and the findings demonstrate that the proposed algorithm enables accurate width estimation. However, it should be noted that certain unexpected inefficiencies and inaccuracies have been identified in a few specific scenarios.