They will use this opportunity to train and run, HTR/OCR pipelines consisting of segmentation models, text-recognition models and post-correction models.
"This will enable us to process large amounts of scanned images, index them with metadata, and make them searchable. It will improve access to our archives and streamline our internal case handling” says Erik Lenas, data scientist at Riksarkivet.
He adds, “access to Vega, and the ability to train models specialized on archival texts, will further our larger goal of adapting AI-technology for use in archives. Thereby opening up new possibilities for academic researchers, genealogists and others to explore and use our archival holdings."
ENCCS continues to offer HPC support to industry and the public sector by providing training on popular techniques and languages, helping with the application process and also hands-on HPC/AI programming support. For more information please visit https://enccs.se/industry-public-administration/.