Kalmar Regiment Transcription Project (HTR/OCR)

This site documents an ongoing effort to convert Swedish National Archives (Riksarkivet) images related to the Kalmar Regiment into searchable text. The workflow uses an HTRflow pipeline together with the trocr-base-handwritten-hist-swe-2 handwriting model, followed by quality checks and manual spot review.

What you’ll find here

• Transcriptions exported from the OCR/HTR pipeline, organized by source image batch.
• Quality control notes highlighting empty or unusually short outputs that likely need reprocessing or human checking.
• Context pages that explain Swedish Army organization around 1790–1814 (indelningsverket, värvade units, militia/landvärn), to help interpret the documents.

Current processing status

Initial runs produced text outputs for four image batches (A0028396_all to A0028399_all): 1,757 text files in total. A small number of outputs were flagged as empty (10) or very short (6), often corresponding to cover sheets, stamps, or pages where segmentation likely missed the writing.

How to use this site

• Start with Browse to explore items and transcriptions.
• Check Quality notes when you see a transcription that looks incomplete.
• Use the context pages if you need help understanding unit terms and organization.

Scope & limitations

These texts are machine-generated and may contain errors (especially in names, diacritics, and older handwriting styles). Where possible, questionable pages are flagged for follow-up review.