How to copy recognized text from an image ?

[godot42 2]

So, evernote recognizes the text, in a photo I've put in evernote: I can search for text in the image.
Now I would like to cop the "image" as ASCII text - but I didn't find a way to do this.

What is the trick ? (After all, evernote recognizes the text...)

[Mike P 2,627]

2 hours ago, godot42 said:

What is the trick ?

There isn't one. If you want the text out of an image you will need to use proper OCR software. The way EN recognises text is very clever. You used to be able to see this in the downloaded enex file but since V10 the data is stored elsewhere. For each word in the image EN may well assign more than one recognised word. e.g for the word cat it might record cat and cot because they look similar. This will lead to some false postives when searching but massively reduces the possibility of a false negative.

[godot42 2]

1 hour ago, Mike P said:

There isn't one. If you want the text out of an image you will need to use proper OCR software. The way EN recognises text is very clever. You used to be able to see this in the downloaded enex file but since V10 the data is stored elsewhere. For each word in the image EN may well assign more than one recognised word. e.g for the word cat it might record cat and cot because they look similar. This will lead to some false postives when searching but massively reduces the possibility of a false negative.

Thanks !
At least that means, I don't need to look further...
(Google Lens did the trick - but I try to avoid google and had hoped I have another solution...)

[PinkElephant 8,077]

On a Mac (newer model) it would be no problem: You can pick any text from pictures using the build in tools of MacOS. And there are free apps that do the job as well.

On Windows there are probably tools that do the same.

[gazumped 11,652]

14 minutes ago, PinkElephant said:

On Windows there are probably tools that do the same.

Hmmn.  None that I'm aware of.  I use Google Lens (or Docs) too.  The only other option I can think of would be to create a searchable PDF from an image which (I think) then has selectable text - but it depends on the image content.

[PinkElephant 8,077]

Wow - I am feeling my age, being away from Windows that long ?!

On the Mac everything is executed locally, on device, not feeding any information to the ever growing data hoarded in some data centers. A free Mac app that does OCR would for example be OwlOCR. They even do PDFs, and install into the right click menu. Select document, right click, select OCR, and it starts crunching characters out of pixels.

It runs on newer Intel Macs as well, but all this excels on the Apple Silicon Macs. The neural engine KI area is larger than the main CPU surface, and just races through jobs like that. You open a picture in the Fotos app, and the text can be extracted in practically the very same moment. But the Mac runs the pictures down in the background as well, making them searchable for text. Had some WOW moments recently.

What makes EN still stand out: Searchability of all sort of documents, supported by tags, and adding relevant information to them.

[gazumped 11,652]

Just took some pics on my Android and looked on the laptop at Google Photos (where they wind up) - saw this:

...and when I checked it had ALL the content of that picture,  including some text you don't see here that was vertically placed in a black-on-black logo!  That link just calls Lens to deliver the content,  but it was scary good and very fast...

[PinkElephant 8,077]

This is how it looks on iOS 16. It runs on device and works even for pictures not synced into iCloud. When the symbol at the tip of the green arrow shows, text was detected and can be extracted: