Supporting Database: Defect Dataset, Annotations, and Model Files for Surface Defect Identification on Traditional Chaoshan Dwelling Exterior Walls

This dataset, "Supporting Database," constitutes the technical core of a larger research data collection on surface defect identification in traditional Chaoshan dwellings. It contains all the structured data and model-related files necessary to reproduce the deep learning experiments described in the associated study.

The dataset is organized into the following subfolders:

1.Defect Dataset – The curated image dataset of exterior wall surface defects, covering six defect categories (including mottled discoloration, surface layer detachment, microbial colonization, and fine cracks), used for model training and evaluation.

2.Annotation and Training Data – The annotation files, class labels, and training configurations associated with the defect images, including data splits for training, validation, and testing.

3.YOLO Model – The YOLOv12 model configuration files, trained model weights, and related parameter settings used in the experiments.

Together, these materials enable full reproduction of the automated defect detection pipeline reported in the associated research. This database is provided solely for the purpose of verifying the authenticity of the associated research and supporting peer review; no other uses are permitted.