I used jTessBoxEditor to create tessdata for a text and a font file.
It works pretty well.
But as I need to OCR photos of printed labels, the font is very similar to the one used in the labels, but I guess it would be better to directly train tesseract with the label letters themselves.
It's actually only digits, so I only need 10 symbols, I can cut them and prepare a TIF training file by hand, but jTessBoxEditor doesn't seem to accept that, it only generates a TIFF from text and a font.
Is this possible to do with tesseract?
or jTessBoxEditor? (I know this is not part of tesseract, but maybe some of you use it?)
Thanks!