OCR

[jer32]

Native OCR support would definitely be important to have.

[msteffens]

Native OCR support would definitely be important to have.

Hi, many thanks for the request!

I've added this as a feature request: Provide OCR capabilities when importing a PDF

I agree that native OCR capabilities would be indeed very useful. However, depending on the license and/or pricing of the OCR library or service, this may be tricky. It's definitely something that won't come soon.

For users who also use DEVONthink Pro, this app supports OCR and could be used as a workaround.

Matthias

[jer32]

What if the PDF has no text layer?
How would you highlight a section?
Could you still create an excerpt of the text?

[msteffens]

What if the PDF has no text layer?
How would you highlight a section?
Could you still create an excerpt of the text?

If the PDF has no text layer, then one wouldn't be able to select and highlight (or copy etc) some text from it. Any app w/o built-in OCR capabilities has the same issue. There are tools that can run OCR on a PDF (such as DEVONthink), and these could be used as a workaround to preprocess the PDF.

Built-in OCR capabilities are certainly highly desirable, but it's not a small feature that can be implemented quickly.

[Theo]

Why not use the native framework offered by Apple? The Vision framework offers multilingual character recognition. It is efficient but if it is not perfect (it has trouble with multicolumn texts), but is fully integrated into the system, evolves regularly and avoids cost problems and licenses for external solutions, OpenSource or not.
In addition, it's On-Device OCR engine, respectful of privacy that does not use online servers.

Apple OCR is used by more and more OCR applications like Prizmo for example…

https://developer.apple.com/documentati ... _in_images

What do you think ?

[msteffens]

Why not use the native framework offered by Apple? The Vision framework offers multilingual character recognition. It is efficient but if it is not perfect (it has trouble with multicolumn texts), but is fully integrated into the system, evolves regularly and avoids cost problems and licenses for external solutions,

Hi Theo,

many thanks for pointing out the text-recognition features of Apple's Vision framework! Use of a built-in framework would indeed be ideal and I should indeed look into this.

That said, in the case of my app, the typical use case would be to add a text layer over the (scanned) text of an imported PDF. After the OCR process, users should be able to search, select & copy the text of the PDF. So, ideally the framework itself would understand how to deal with PDFs. Also, multi-column PDFs are very common in the academic world, so it would be good if these could be dealt with properly.

Anyways, I appreciate the pointer, thanks again!
Matthias