GeoAI-driven building facade classification for urban pattern analysis: A case study in Murcia, Spain

The rapid expansion of cities presents new challenges for urban planning and housing policy, making the adoption of advanced technologies essential for supporting evidence-based decision-making. This study introduces an integrated GeoAI (Geospatial Artificial Intelligence) methodology for analyzing urban patterns in Murcia, Spain, combining cadastral records and facade images obtained from the General Directorate of Cadastre platform, Geographic Information Systems (GIS) and deep learning models. Using clustering and a pre-trained EfficientNetB0 architecture, we developed an automated classification model that identifies two main groups of dwellings, reflecting clear spatial contrasts in building typologies, heights, and construction quality. The results, integrated into a GIS, provide detailed territorial insights that align with housing cost and construction category variables. Beyond technical innovation, this approach offers practical applications for municipal authorities, enabling more efficient built-space management, improved zoning, and enhanced strategies for housing policy, urban growth control, and climate resilience. The model represents a scalable tool for large-scale urban analysis, adaptable to other cities facing rapid transformation. • Deep learning model for automated classification of facades images in Murcia, Spain. • Two facade categories identified, linked to construction category and housing cost patterns. • Spatial analysis reveals urban structure differences across central and peripheral areas. • Automated AI-GIS method offers scalable, efficient alternative to manual surveys. • Integrates AI and geospatial analysis to support data-driven urban planning and policy-making processes. • Lays groundwork for future research on climate adaptation in urban residential areas.