Good idea, but train your OCR properly. There's no reason for recognition mistakes if your text is taken from a screenshot (not deformed, not randomised font).
Thanks, I totally agree. The tessnet2 engine I am using didn't seem to have very in depth training. If it was a more serious project I would have looked to use tesseract 3 which seems to have more flexibility for training.
9 comments
[ 3.1 ms ] story [ 30.6 ms ] thread