I've noticed that, although what it does capture tends to be more accurate, the Monterey version doesn't like short numbers. Presumably it doesn't recognise them as characters because there's an insufficient pattern.
20
39
49
303
73
488
83
65
359
104
15
355
Notice that 7, 15, 8 and 1 are all missing.
There's one more thing you can do to speed it up a tiny bit.
Instead of writing the screen capture to a file and running the shortcuts on it, write the screen capture to the clipboard, then run the shortcuts on that.
My shortcuts is just 3 lines long:
Get clipboard
Extract text from Clipboard
Copy Text from Image to clipboard
Thanks. That's what I'm doing now actually. I can't trust it for everything though, as it often doesn't recognise single or double digit numbers.
I run into the same problem with capital "I" if it's a skinny non-serif font, even when it's in the middle of a sentence.
I've found that sometimes, it might help to capture some "noise"...like if the numbers are surrounded by a box or something like that, capturing the box or even part of the box seems to help the OCR understand that single characters (especially "ones" and "I's") aren't just debris. YMMV of course.
I've just seen the below by @gglick - it works without having to make a temporary file first and seems to be faster.
