Sent from my iPhone
> On Aug 7, 2019, at 22:08, Cristobal Jesus Muñoz Solano <cmunoz...@gmail.com> > wrote: > > this image return > L2007190588S37<<<<<<<<<<<<\n77F1912157PER22344783<K<3\n<RODRIGUEZ<<LORENZA<SO > but in https://www.doubango.org/webapps/mrz/ work good. i dont know why :( This is what I explained in my previous post. The issue isn’t the trained model but the preprocessing. The Mrz lines are skewed and the contrast isn’t uniform. Fine tuning the model won’t help. Tesseract uses Otsu binarization, dump the binary result to see how it works. Using doubango’s online demo it works fine because the lines are de-skewed, de-slanted and binarized using TMMS and Wolf. See https://groups.google.com/forum/m/#!topic/doubango-ai/WxMG4F3X5XQ. This said, add few pixels to the image to make sure the first letters are taken. > -- > You received this message because you are subscribed to the Google Groups > "tesseract-ocr" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to tesseract-ocr+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/tesseract-ocr/b182f7cd-cb15-439d-a4a0-105aeedd65bb%40googlegroups.com. > <Selección_001.png> -- You received this message because you are subscribed to the Google Groups "tesseract-ocr" group. To unsubscribe from this group and stop receiving emails from it, send an email to tesseract-ocr+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/A5405E6E-D10D-40C7-A6CA-57D4CE1A275B%40yahoo.fr.
