Defect inspection of existing buildings is receiving increasing attention for digitalization transfer in the construction industry. The development of drone technology and artificial intelligence has provided powerful tools for defect inspection of buildings. However, integrating defect inspection information detected from UAV images into semantically rich building information modeling (BIM) is still challenging work due to the low defect detection accuracy and the coordinate difference between UAV images and BIM models. In this paper, a deep learning-based method coupled with transfer learning is used to detect defects accurately; and a texture mapping-based defect parameter extraction method is proposed to achieve the mapping from the image U-V coordinate system to the BIM project coordinate system. The defects are projected onto the surface of the BIM model to enrich a surface defect-extended BIM (SDE-BIM). The proposed method was validated in a defect information modeling experiment involving the No. 36 teaching building of Nantong University. The results demonstrate that the methods are widely applicable to various building inspection tasks.

Ancient Chinese architecture, with its typical symmetrical structures, curved roofs, and upturned eaves presenting a unique architectural aesthetic, is a treasure of Chinese culture. Recently, unmanned aerial vehicle oblique photogrammetry and laser scanning technology have greatly facilitated the realistic replication of ancient buildings and have become crucial data sources for the HBIM of ancient buildings. However, parameter extraction and geometric model representation are more difficult because of the curved surfaces and upturned eaves of traditional Chinese roofs. As symmetrical features are typical of ancient Chinese architecture, the parameter quantity and modelling difficulty of the model representation can be effectively reduced by recognizing the symmetrical structure of traditional Chinese roofs and using “mirror replication” to quickly generate the other half of the model. Accurate symmetry detection and highly efficient parameter extraction are crucial for the HBIM of traditional Chinese roofs. Therefore, in this study, a deep learning network, namely, TCRSym-Net, is proposed to identify the symmetry from point clouds of traditional Chinese roofs. Each roof point cloud is then relocated and reoriented to obtain longitudinal and cross sections, and parametric modelling scripts are coded in Dynamo to model traditional Chinese roofs via curve lofting and solid Boolean operations. The experimental results reveal that the symmetry detection network is effective for symmetry detection, and five different types of traditional Chinese roofs are successfully recreated, which confirms the dependability of the method.