Decay detection in historic buildings through image-based deep learning

Nowadays, built heritage condition assessment is realized through on-site or photo-aided visual inspections, reporting pathologies manually on drawings, photographs, notes. The knowledge of the state conservation goes subjective and time cost consuming procedures. This even relevant for a historic building characterized by geometrical morphological complexity huge extension, at risk collapse. In this context, advancements in field Computer Vision Artificial Intelligence provide an opportunity to address these criticalities. proposed methodology based Mask R-CNN model, detection decay morphologies heritages, and, particularly buildings. experimentation has been carried out validated highly heterogeneous dataset images buildings, representative regional Architectural Heritage, such as: castles, monasteries, noble rural outcomes highlighted significance remote, non-invasive inspection technique, support technicians preliminary conservation, most all, mapping some particular classes alterations (moist area, biological colonization).