OCR Text Recognition troubleshooting

Common issues

• Confidence below 50%: the scan is too low quality. Rescan at 300+ DPI with good lighting.
• Wrong characters: check the language setting. English model will misread accented characters in French/German/etc.
• Processing stalls: reduce max pages. Each page renders at 300 DPI which uses significant memory.
• First run is slow: it's downloading the ~15MB Tesseract language data. This only happens once per language.

Recovery steps

1. Retry with a smaller sample file.
2. Refresh and run the tool again.
3. Use an alternative workflow from /tools if needed.
4. Check /status for current incidents.

What this does not protect

• Troubleshooting guidance does not guarantee recovery for damaged files.
• It does not bypass document owner restrictions when cryptography is enforced.