A Semantic Classification Approach for the Aachen Cathedral

A novel semantic classification approach, integrating 3D survey data and machine learning, demonstrates significant potential for distinguishing materials and construction techniques in historic masonry, particularly at UNESCO World Heritage sites like Aachen Cathedral. The study, focusing on the Westwerk area, combines terrestrial laser scanning (TLS) and photogrammetric point clouds with supervised algorithms (Random Forest) and 2D image segmentation via META’s Segment Anything Model (SAM). Findings reveal that 3D geometry-based classification excels in identifying features with strong morphological differentiation, while 2D approaches are more effective for visually subtle or geometrically similar elements. Annotated 2D masks projected onto the 3D model enhance classification reliability, contributing to scalable, semi-automatic documentation and conservation planning for complex heritage sites.

Attenni, M., Barni, R., Bianchini, C., & Griffo, M. (2025). A Semantic Classification Approach for the Aachen Cathedral. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XLVIII-M-9-2025, 71–78. https://doi.org/10.5194/isprs-archives-xlviii-m-9-2025-71-2025