A small framework taking over the manual training process described in the Tesseract3 Wiki: https://code.google.com/p/tesseract-ocr/wiki/TrainingTesseract3
I am using your system (very nice by the way) to train a new Font for Tesseract, and ran into an issue I thought I would post about. The box files being generated for this particular font appear to have a -1 pixel height error, and a +1 pixel offset error (which I am compensating for with +1, -1 respectively).
It does not appear to me that this is an issue with TesseractTrainer, but rather, I believe this may actually be an issue with PIL/Pillow. Regardless, I wanted to post so that other folks running into this issue have some clue as to how to remedy.
Hi,
I am using your system (very nice by the way) to train a new Font for Tesseract, and ran into an issue I thought I would post about. The box files being generated for this particular font appear to have a -1 pixel height error, and a +1 pixel offset error (which I am compensating for with +1, -1 respectively).
It does not appear to me that this is an issue with TesseractTrainer, but rather, I believe this may actually be an issue with PIL/Pillow. Regardless, I wanted to post so that other folks running into this issue have some clue as to how to remedy.