GYU-DET

GYU-DET has collected a large number of bridge surface defects, obtaining a total of 11,123 high-resolution images, including six types of defects: cracks, spalling, water seepage, honeycomb surfaces, exposed reinforcement, and holes. It covers various lighting and environmental conditions, comprehensively and realistically reflecting the diversity and complexity of bridge defects.The dataset is divided into training, validation, and testing subsets in an 8:1:1 ratio, and each is stored in a separate compressed file (train.zip, valid.zip, and test.zip). Within each archive, there are two folders: images, containing the original high-resolution images, and labels, containing TXT annotation files that specify the location and category of each defect instance. Additionally, a classes.txt file lists all defect categories in a fixed order, serving as a reference for mapping class indices in the annotations to their corresponding labels.