Crack Detection and Quantification on Concrete Surface of Sewage Treatment Plant Based on Improved Deep Learning Method

Cracks can seriously affect the safety of concrete structures in sewage treatment plants, while existing manual inspection procedures are time-consuming, labor-intensive, and prone to errors and omissions. In order to achieve timely warning and management, this study integrates the YOLO-V5 algorithm with data enhancement technology for concrete crack detection and at the same time combines the Otsu image processing method to quantify and analyze the cracks. The proposed crack detection method is validated based on a wastewater treatment plant construction project. The results show that the method can effectively detect the cracks with an average accuracy of 88.6% under unfavorable conditions. The quantification errors of crack length and width are small. The method contributes to the efficient detection and analysis of safety-critical cracks in sewage treatment plant projects.