Vectorisation of the 1890 London OS – the different colours represent the different particle orientation. (Rémi Petitpierre, Paul Guhennec)
The team develops new methods for analyzing and interpreting large-scale historical cartography. Our approaches include computer vision, machine learning and large language models to extract information from historical cartography and interpret it. We have activated two research paths, one more specific to the analysis of cadastral mapping over time.
The other for the visual analysis of historical cartography.
CADASTRA is an initiative aimed at automatically vectorizing cadastral mapping. Thanks to our training base, we can achieve optimal results.
CARTONOMICS is an initiative that aims to analyze a corpus of over 200,000 historical maps compiled over time by comparing similarities in graphic and textual features.
We organize several training workshops.
