What is the best ocr for linux?

Question

linux

Need ocr for the most accurate character recognition in photos for linux, preferably with the ability to work from the command line.
So far I have settled on tesseract, but the error is quite large.
Finereader on windows recognizes well, but on linux the paid version is limited.

Answer 1

Try to run through scantailor before recognition.

Answer 2

tesseract is like Linux from scratch. Somehow, not without suffering, they managed to recognize the pre-reform text very well with all these i and b - well, in the sense of "yats". Well, very good. He needs to be fed dictionaries and teaching him to do. I don’t remember the Nuans, although I remember I wanted to describe its setting somewhere on Habré. Well, it's been two years already... (
Also, Abby seems to have released her Finereader for Linux and even, as it were, "free". But in fact (two years ago, why she switched to tesseract) a lot of words from this company are not enough free .

Answer 3

tesseract without options. Everything else is worse and slower. But, personally, I use FineReader 12 in a virtual machine, because it recognizes stress and pre-revolutionary texts.
As far as I remember reading, Tesseract developers from Google are now working on its 3rd version. But for now, they will make Russian recognition modules a neural network. language will pass more than one year.
By the way, all sorts of "online recognition services" are just tesseract web interfaces, less often commercial FR, or some simple recognition tool.