Point Cloud Semantic Segmentation of Building Elements of Fatih Mosque, Istanbul

The architectural heritage digital model is important for high-accuracy documentation, archive security, and research opportunities. This study focuses on the autonomous documentation of the digital model of Fatih Mosque’s facade elements. Fatih Mosque literature focuses on its restoration process, historical importance, and architectural values. The literature on the documentation of the mosque with current technological methods is limited. This study applies semantic segmentation on point-cloud data to detect facade elements. Point-cloud data was produced via photogrammetry from the southwest and northwest facades. The data was labeled with masonry wall, main load-bearing wall, column, window, entrance, staircase, arch, and spouts. CANUPO classifier in CloudCompare software is used for semantic segmentation. Changing the classification parameters in CANUPO increased the accuracy rate in predicting facade elements. This study contributes to the literature by providing autonomous documentation of the Fatih Mosque’s facade and a guide for using the CANUPO classifier in digital model production.