Bug 460320

Summary:	Add OCR feature
Product:	[Applications] Spectacle	Reporter:	kono93 <kochnorman>
Component:	General	Assignee:	Boudhayan Gupta <me>
Status:	CONFIRMED ---
Severity:	wishlist	CC:	akid.anis799, andrea.ippo, christian.tallner, geqch0akc, kde, me, nate, ostap.tymchenko
Priority:	NOR
Version:	20.12.3
Target Milestone:	---
Platform:	Debian stable
OS:	Linux
Latest Commit:		Version Fixed In:
Sentry Crash Report:

Description kono93 2022-10-12 16:47:00 UTC

SUMMARY
***
It would be great to be able to OCR screenshots directly, so you can copy text from them or make them searchable via pdfgrep.
This could be done by Tesseract, enabled only if installed, and would kind of easy to add, but very useful.
***

Comment 1 Nate Graham 2023-01-05 18:47:03 UTC

*** Bug 463177 has been marked as a duplicate of this bug. ***

Comment 2 Nate Graham 2023-04-03 23:20:24 UTC

*** Bug 467942 has been marked as a duplicate of this bug. ***

Comment 3 Nate Graham 2023-04-18 19:14:54 UTC

Skanpage is already using Tesseract for OCR, so that could be a place to look for inspiration.

Comment 4 akid.anis799 2023-04-25 04:17:29 UTC

as a workaround one could try to use this command for OCR with spectacle:

For X11:
`spectacle --nonotify --region --background -o /proc/selt/fd/1 |  tesseract stdin stdout | xclip -in -selection clipboard`

For Wayland:
`spectacle --nonotify --region --background -o /proc/selt/fd/1 | tesseract stdin stdout | wl-copy`

Comment 5 Andrea Ippolito 2023-08-07 12:18:29 UTC

Normcap is doing this, although it's a standalone app:
https://github.com/dynobo/normcap/

Maybe worth looking into/getting in touch with the DEV.

May I add, it would be cool if OCR capabilities weren't limited to Spectacle, but somehow baked-in in some frameworks part, and then be usable by any KDE app that can display images by pressing a button and having the detected text appear as overlay (e.g. gwenview showing a photo of a receipt, okular showing a page that was scanned without OCR, etc).

Sounds complex and impacting quite some apps, but would be a wonderful productivity addition (if OCR accuracy is spot-on)

Comment 6 ostap-tymur-tymchenko 2023-08-07 18:29:35 UTC

(In reply to andrea.ippo from comment #5)
> Normcap is doing this, although it's a standalone app:
> https://github.com/dynobo/normcap/
> 
> Maybe worth looking into/getting in touch with the DEV.
> 
> May I add, it would be cool if OCR capabilities weren't limited to
> Spectacle, but somehow baked-in in some frameworks part, and then be usable
> by any KDE app that can display images by pressing a button and having the
> detected text appear as overlay (e.g. gwenview showing a photo of a receipt,
> okular showing a page that was scanned without OCR, etc).
> 
> Sounds complex and impacting quite some apps, but would be a wonderful
> productivity addition (if OCR accuracy is spot-on)

I dont think it would actually be so hard. Tesseract OCR is both very advanced and open source. All KDE would have to do is to have it preinstalled, and then implement it into the apps. obviously implementing it would be a lot of work but having it be in KDE isnt hard.

Comment 7 akid.anis799 2023-11-19 09:53:19 UTC

New workaround for using ocr with spectacle

For X11:
spectacle --nonotify --region --background -o /tmp/screenshot.png && tesseract /tmp/screenshot.png stdout | xclip -in -selection clipboard

For Wayland:
spectacle --nonotify --region --background -o /tmp/screenshot.png && tesseract /tmp/screenshot.png stdout | wl-copy

Comment 8 Noah Davis 2024-04-18 01:04:46 UTC

*** Bug 479412 has been marked as a duplicate of this bug. ***

Comment 9 Ember 2024-08-26 07:49:14 UTC

This may well be out of scope, but i'd also like the OCR result to be saved.. somewhere, preferably in the image, so that i may search for text content of screenshots through dolphin