How to train tesseract for your font?

Question

Python

Good afternoon. I'm trying to train tesseract v5 for my font, because standard even after image processing show an incorrect result. Can someone tell me how it's done or throw a manual, for several days I've been looking everywhere to no avail.

Answer 1

tesseract 4.1.1, pytesseract 0.3.0 eats everything (well, almost)
Code:

def ocr(self):
        global dtout
        pytesseract.pytesseract.tesseract_cmd = '/usr/bin/tesseract'
        dtout = pytesseract.image_to_string(Image.open(fname), 'rus+eng+chi_tra+jpn')
        self.ui.textOut.setText(dtout)

Show the picture you are going to recognize.

Answer 2

For special tasks, the Gamera framework may be suitable . You can even recognize hieroglyph scripts and various historical inscriptions. But this is more for researchers.
Learning takes place interactively.

https://gamera.informatik.hsnr.de/docs/gamera-docs...