Heritage Facade Damage Detection (HFDD) - Saint-Jean-du-Baly Church

Description: The Heritage Facade Damage Detection (HFDD) dataset is a publicly available image dataset compiled for the purpose of training and evaluating deep learning-based object detection models for automated surface deterioration assessment in heritage stone structures. The dataset was collected from the Saint-Jean-du-Baly Church, a historic stone building located in Lannion, Brittany, France, and comprises 2,183 facade images annotated across four damage categories. Classes: Normal — undamaged surface regions Color change — stone discoloration due to weathering or biological activity White spots — efflorescence deposits on stone surfaces Plants — biological colonization including moss, lichen, and vegetation Dataset Structure: The dataset is organized in YOLO format and split into three subsets: Train: 1,526 images Validation: 305 images Test: 352 images