I have tested this add-on using English, Japanese and Spanish text, and its yield has been surprisingly successful. Not perfect, but good enough.
I tried it on complex and difficult to read text (from blurry or pixelated photos) and the results, although not completely accurate, were sufficiently close, and in some cases 100% perfect after a few retries.
Copyfish cannot read vertical text at all (such as Chinese or Japanese) but horizontally formatted text is generally not a problem.
There are some quirks with it and the interface can be fiddly. A few times I found myself having to reload the whole page from the browser in order to make the add-on retry the OCR work since the buttons just stopped responding, but it was never more than a small time-consuming problem.
All-in-all this add-on may very well save you a lot of time typing from an image, specially if you don't have the means or ability to type from a language that uses a different alphabet than your own, be that Latin, Cyrillic or Hiragana.
:)