HACKER Q&A
📣 beatthatflight

Scanning of Old Handwritten Diaries


OCR tends to work reasonably well on text, even block hand writing.

However, a relative has just scanned in an ancestor's diary from the Boer War era in South Africa. It'd be fascinating to read.

However, it's in beautiful but (to me) barely legible cursive.

An example page: https://imgur.com/a/SJpyj94

Are there any tools out there that might actually have a chance with handwriting like this?

I tried a sample app I've worked on with OpenCV. It recognised there was "text" but that was about it.

Google translate camera app gave me nothing.

Word Lens (Same base code) as well.

Is there one out there or a technique or suggestion anyone might have?


  👤 brudgers Accepted Answer ✓
Just transcribe it. A few pages a day and it will done before too long. That's what the Smithsonian does, so maybe it's even best practice.

https://transcription.si.edu/


👤 cerberusss
How many pages is the diary? I wonder if it isn't much simpler to brute-force it via Amazon Mechanical Turk and similar platforms.

👤 rubidium
If you find anything post it here. Would also help with a bunch of deed records from 100 years ago.

👤 pratikshadake
Going forward, you may wanna user reMarkable.