Is it possible to get percentage of accuracy of recognized text?
I need to recognize multiple languages (2 languages) and tesseract doesn't know exactly what language is when I put parametar -l lang1+lang2.
What I want to do is to scan with both languages separately, but I would need some percentage of accuracy to determine probability of language.