Scanning handwritten paper notes into PDF ➜ OCR delivers nothing

I still function better and do more original work when I do handwritten notes, in my journal or on small paper slips etc. When I scan those and send them to DT, they don’t work with OCR. Or OCR isn’t complaining, but the RTF or PDF or annotation is empty.

So, either something is not working OR in DT the OCR is no good for handwritten notes. Which one is it?

OCR on handwriting is not broadly supported in OCR engines. And the support is usually fairly strict in terms of what letterforms it can detect as characters. ABBYY’s OCR engine has some capability, but YMMV.

Here’s a simple example I just did, noting it isn’t 100%…

Of course, that could also be a best guess since DEVONthinkers isn’t an actual word.

It also detected an extraneous mark as an s.

Was this handwritten on DTTG and then manually OCR on DT?

I’ve tried handwriting on DTTG but didn’t see any text get OCR’ed (so have been using GoodNotes then exporting). I’ve also tried Apple notes which also seems to do a good job of OCR of handwriting (or taking photo of handwriting), but when I “export” this from Apple notes to DTTG the OCR doesn’t seem to be linked to the pdf.

No. It was handwritten with my hand, pen, and paper.

but when I “export” this from Apple notes to DTTG the OCR doesn’t seem to be linked to the pdf.

That’s because there is no true OCR as that process yields a text layer or a text output. Live Text is smoke and mirrors for momentary recognition but it is not applied to the document, hence you shouldn’t expect anything to transfer.

OK. I see. OCR is good for scanning printed text. So I stay with my reMarkable pad, because their handwriting detection is excellent. I just need to send the text files to DT.

And I still got to find a solution to get the paper journals into DT that I wrote before I got my reMarkable pad.

I wish you better luck than I had. COVID lockdowns fortunately provided sufficient time for mass manual transcription of years of journals. However, by way of encouragement, having them in DEVONthink (with its superlative search facilities) is a revelation: good luck, and it’s worth doing!

Stephen

2 Likes

Improved support for handwritten notes is indeed planned for future releases.

2 Likes

I have been expirmenting a bit, and have found that if I do a handwritten note in apple notes and send it to DTTG (which does not have any OCR), that by converting to a PDF and then converting to searchable PDF it actually does a reasonable job of OCR (for my needs).

If I make the pdf directly into DTTG (with for example a lined template), and then attempt to convert to searchable OCR it does not add any OCR layer. It appears to ignore my handwriting completely.

Addendum:
Apple notes pulls over a formatted note, which then converts to PDF which then ocrs my handwriting. I’m assuming there is something about how DTTG May process the Formatted note to PDF that allows it to do OCR as compared to native PDF creation inside DTTG.

then attempt to convert to searchable OCR it does not add any OCR layer. It appears to ignore my handwriting completely.

That’s correct. Handwriting isn’t supported in the OCR in DEVONthink To Go.