I also want to know that how we can train on real images which are not single lines?
On Thursday, October 8, 2020 at 1:37:02 PM UTC+5:30 smn...@gmail.com wrote: > Hello, > > I would like to train *Tesseract 4* to recognize certain > scripts/languages based on real images rather than synthetic ones. Here are > my questions: > > 1. Is there a tool, preferably cross-platform (Windows/Linux) GUI, that > assists in creating .box file based on scanned images? How to get > coordinates of textlines? etc... > > 2. Is there a youtube/video tutorial describing .tiff/.box files > preparation based on real scans? > > 3. What provides better recognition - training on real images or training > on synthetic images? > > 4. How many textlines of real scans do I need to get proper recognition? > > Thank you very much! > ST > -- 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/7ca8dded-7323-4690-b7b4-30f1a7fa605cn%40googlegroups.com.
