Machine Vision algorithms on cadaster plans

Cadaster plans are cornerstones for reconstructing dense representations of the history of the city. However, as some of these handwritten documents are more than 200 years old, the establishment of processing pipeline for interpreting them remains extremely challenging. The talk will present the implementation of an automated process capable of segmenting and interpreting Napoleonic Cadaster Maps of the Veneto Region dating from the beginning of the 19th century. Using a deep learning approach, the system extracts the geometry of each of the drawn parcels, reads and interprets the handwritten labels. This efficient and automated process opens new perspectives to reconstitute the past.