In the ongoing Hugin Munin project, our goal is to achieve high-quality handwritten text recognition (HTR) for the majority of documents in the National Library of Norway. For this purpose, we present the NorHand dataset, which consists of annotated Norwegian documents. We train and evaluate three state-of-the-art models for handwritten text recognition and provide a qualitative and quantitative evaluation of the results.
Check out the slides presented at DH2024 here.
