Semi-automatic extraction of georeferenced building data from historic Sanborn Fire Atlas maps

The Need: Urban historians, city planners, and preservationists require efficient tools to extract building-level information from historic Sanborn Fire Insurance maps, essential for understanding urban development over time. However, manual extraction is time-consuming and impractical due to the sheer volume of maps and lack of suitable computational methods.

The Technology: Our solution presents a scalable and semi-automated framework specifically designed to extract urban information from Sanborn maps and reconstruct historic urban neighborhoods cartographically. This framework operates in three key steps: georeferencing the maps to a geographic coordinate system, employing machine learning techniques to automatically identify building footprints and properties, and creating 3D visualizations of the historic neighborhoods.

Commercial Applications:

  • Urban historical research and analysis
  • City planning and redevelopment projects
  • Preservation efforts and heritage site management


  • Efficiency: Significantly reduces time and resources required for manual extraction and analysis of building-level information from Sanborn maps.
  • Accuracy: Utilizes machine learning methods for precise identification of building footprints and properties, ensuring high accuracy of extracted data.
  • Visualization: Facilitates the creation of immersive 3D visualizations, aiding in the interpretation and communication of historical urban landscapes.

Loading icon