When is it beneficial to train Tesseract?

Question

I'm using Tesseract in my project to convert images that I've scanned from french newspapers. I want to know if I need to train Tesseract in order to recognize the french fonts and the specification of this language such as "caret", "circumflex accent" , etc

Answer 1

If the standard issue fra data language does not produce acceptable results, then you may consider training. Tesseract recognizes diacritical characters very well.