Typological patterns of traditional buildings in Lingnan traditional villages: a quantitative analysis based on digital technology

Traditional villages and their buildings are deteriorating rapidly under urbanization, highlighting the need for efficient digital preservation. This study analyzes ten traditional villages in Guangzhou, China, using a previously established dataset derived from UAV oblique photogrammetry, multiscale segmentation, and machine learning. The dataset comprises 3414 traditional buildings, 30 attributes, and 102,420 data entries. Quantitative analyses of spatial distribution, building scale, roof morphology, and plan form establish a typological system comprising five spatial distribution types, three building scale types, three roof slope types, and three plan morphology types. The results reveal dominant regional characteristics, including compact spatial layouts, hierarchical building scales, and consistent morphological patterns. The proposed cross-village typological framework provides quantitative references for heritage assessment, conservation planning, and contextual renewal in Guangzhou and offers a transferable analytical approach for traditional village studies in the broader Lingnan region.