Seedproject information
Project manager
Thomas Holgersson
Other project members
Ahmad M. Kamal & Daniel Ocic Ihrmark, Linnaeus University
Participating organizations
Linnaeus University
Financier
Linnaeus University Centre for Data Intensive Sciences and Applications
Timetable
1 Aug 2025–31 Jan 2026
Subject
Digital Humanities (Department of Cultural Sciences, Faculty of Arts and Humanities)
More about the seedproject
This project explores the application of machine learning tools for the enrichment of historic maps. More specifically, the seed project tests the feasibility of currently available AI models for handwritten text recognition (HTR) on cartographic material. This testing will be done using the maps of Danish cartographer, Johannes Mejer (1606-1674). The objective of the seed project is to exploratory test and evaluate different approaches to enrich historical maps.
Mejer’s collection presents a challenge due to being handwritten text and – being on maps – difficult to segment properly given the abundance of non-textual features. To date, while several openly available HTR tools are available, there are relatively few examples of HTR being applied to historic maps (compared to optical character recognition or OCR applied to printed maps with typeface text).
The long-term objective is to develop data intensive methods for interpreting the contents of old handwritten maps with particular focus on Johannes Mejer’s works. This essentially comprises two parts: first, discovering AI tools for transcribing and indexing words and symbols on the maps; second, applying these tools on a large number of Mejer’s maps to create searchable list of transcribed contents. Ultimately, the goal is to make the difficult writings on the maps readily available to researchers in various fields (such as geography, linguistics, and history).
