2025 Volume 6 Issue 1 Pages 117-136
Cracks in civil engineering infrastructures, such as bridges, tunnels, and retaining walls, are common visual indicators of structural weakening. Damage to these structures can directly or indirectly impact human lives and property. In Japan, many of these infrastructures are ageing, making their maintenance a top priority to ensure safety. Traditional crack inspection methods for concrete structures often involve visual inspection using crack scales or gauges. However, this approach is time-consuming, labour-intensive, costly, and subject to the inspector’s judgement. In recent years, research has focused on leveraging Digital Image Processing techniques to address these challenges. This research aims to automate crack width monitoring by using YOLOv5 (You Only Look Once version 5) and the OpenCV library. Reflective targets are placed on either side of a crack to serve as a reference scale. The eight circles on the targets are automatically detected and measurement of the distance between the pair can be determined. The method was tested over a two-year period on a retaining wall in Tamano, Okayama, Japan, to demonstrate its applicability and feasibility. The results show that the model detects target circles with an accuracy of 95.9% and measures the crack movement. This method, compared to the manual approach currently used by the authors, significantly reduces crack width monitoring time, allowing for more frequent inspections and enhancing overall safety.
