Research Information System
XXXII - International Seminar on Urban Form Urban Morphology in the Age of Artificial Intelligence, Turin, Italy, 17 - 20 July 2025, vol.1, pp.103, (Full Text)
Morphological analysis using AI is an
emerging but rapidly evolving field in urban
studies, whereas analyzing urban tissues to
differentiate the characteristics of various
cities has long been a core approach in
urban morphology. The relationship between
form and function remains central to urban
morphology discussions; however, the
quantitative and dynamic application of AI
across different urban patterns remains
unexplored. In this context, this study explores
the influence of morphological metrics of the
building fabric on land use patterns across
different urban contexts. It aims to investigate
the relationship between land use and its most
relevant morphological elements by analyzing
3D spatial data using an AI- driven model.
The proposed method extracts 3D data
from OpenStreetMap and processes it with
classification methods using Scikit Learn. It
leverages land use types using morphological
features by using the building as a unit of
analysis including frontage ratio, height/
width ratio, built-up coverage, compactness,
accessible number of neighboring buildings,
closeness, and betweenness centralities to
predict land use types. The model will be
developed by using 10 km radius catchment
areas from the central business districts of
Berlin, Istanbul, Amsterdam, Rome, Barcelona,
London, Paris, and Moscow. It is trained using
morphological parameters to achieve high-
accuracy land use classification which is later
tested across multiple cities to identify how
effective the typological patterns of the building
fabric condition land use patterns. The AI model
not only predicts land use but also uncovers
typological patterns at an urban scale, offering a
comparative and data-driven approach to urban
morphology. By evaluating the accuracy of AI-
generated predictions, this study contributes
to the advancement of computational methods
in spatial analysis. Ultimately, this research
enhances our understanding of what makes an
urban tissue unique or similar across different
cities, bridging AI-driven analysis with the long-
standing theories of urban morphology.