Abstract
YTU-CrackIS is a dataset of high-resolution crack images captured across multiple locations of St. Theodore church in Cappadocia using both UAV and handheld cameras. The original images were divided into 640×640 pixel patches, yielding 1,904 training images and 469 validation images. Each patch is annotated in polygon format with cracks labelled as a single class, with annotations automatically adjusted to align with the patch boundaries. The dataset covers heterogeneous crack types with varying widths, lengths, and scales, reflecting the real-world complexity of damage found on historic masonry surfaces. We also provide weights for the YOLOv8x model trained on this dataset. Detailed instructions and code for using the weights are available at: https://github.com/TolgaBkm/CH_Crack_Detection The creation of this dataset was funded by the Scientific and Technological Research Council of Türkiye (TUBITAK) 1001 program under Project no. 122Y017.