Research on Wall Crack Recognition Algorithm Based on YOLOv8

In order to overcome the low efficiency of digital image processing, this paper introduces DNN to detect wall cracks. By constructing the training data set of deep learning model, the method of test image acquisition and preprocessing is studied, and the test image library of crack detection model is established. The crack open data set is used to improve the crack training. The deep learning target detection algorithm was studied, and the wall crack recognition and positioning model based on YOLOv8 algorithm was constructed. The width, depth parameters and boundary frame loss function of YOLOv8 algorithm were optimized and adjusted by using the crack recognition data set. It effectively reduces the risk and cost of detection work, and has a strong engineering application prospect and social value.