OCR text in DIP

Main Page > Development > Development documentation > OCR test in DIP

Requirements

See related issues:

• #6257

Add OCR text files to DIP and AIP

• Add open-source OCR tool to Archivematica
  • Tesseract is the most actively developed; other open-source OCR is either moribund (Cuneiform, OCRopus) or released only sporadically without major improvements (OCRad, gOCR).
  • The actively-developed options are:
    • Tesseract - Very actively developed. Best accuracy of all the open-source options, good speed.
    • OCRad - Moderate-to-poor accuracy, excellent speed.
    • gOCR - Poor accuracy, slow speed.
  • Tesseract appears to be the best solution in most cases. If speed was paramount and mediocre or poor accuracy would be acceptable, there might be an argument to use OCRad.
• Add micro-service to OCR files in DIP (post-normalization)
  • Micro-service: Transcription
  • User choice Yes/No (default NO)
• Add FPR purpose - Transcription - OCR first and only tool in that section
• Add OCR files to DIP in "[DIP]/OCRfiles" directory
• Run OCR on originals or preservation copies??
• Add OCR files to AIP in "AIP/data/objects/metadata/OCRfiles"
  • METS file PREMIS event Transcription (add to PREMIS events) PREMIS_metadata:_events#Transcription
  • METS file : use text/ocr fileGrp in METS#.3CfileSec.3E
• Add configuration setting to administrative tab of the dashboard to pre-select OCR options