FROM MAPPING TRENDS TO BUILDING BENCHMARKS
This is our latest review paper that has been published in Spatial Information Research (Springer) journal.
Title:
Using Structural Topic Modelling to Identify Fragmentation and Integration in AI-Driven Townscape Assessment
As Artificial Intelligence continues to reshape urban analytics, we wanted to answer a simple but important question:
Where is the field of AI-driven townscape assessment heading?
To answer this, we systematically reviewed research published between 2007 and 2025 and applied Structural Topic Modelling (STM) to uncover how the field has evolved over time.
Our findings revealed that while areas such as image-based urban modelling, LiDAR, point cloud processing, vegetation mapping, and urban clustering have matured significantly, several important aspects of townscape assessment, including architectural style, skyline, topography, and water-related attributes, remain underexplored.
One of the key outcomes of this study is the introduction of TownscapeBench v0.1, an open reporting framework designed to improve the reproducibility and benchmarking of AI-enabled townscape assessment. We hope this framework will serve as a useful reference for researchers working at the intersection of Artificial Intelligence, Machine Learning, Computer Vision, GIS, and Urban Analytics.
Research is not only about developing new algorithms. Sometimes, it is equally important to step back, understand where a field currently stands, identify existing gaps, and provide a roadmap for future research.
My sincere appreciation goes to my team for the excellent collaboration. I look forward to seeing how this work contributes to the next generation of AI-driven urban and geospatial research.
📄 Read the paper here:
#ArtificialIntelligence #MachineLearning #Townscape #UrbanAnalytics #GIS
