Try inferencing with a different combo and check tesseract settings. On Fri, 11 Apr 2025 at 12:47 am, RuePat07 <patil.rucha.man...@gmail.com> wrote:
> I was training tesseract to add an unusual font and some additional > characters. However after finetuning on it, it has been hallucinating on > default fonts. Is there a way to preserve the accuracy of the previous > fonts and just include this one new font and the new characters? > When i trained with the new font and added characters, the finetuned model > does good with these symbols and font both. > When i tried inferencing with combining these two traineddata file > "eng_new+eng" or "eng+eng_new" they give poor results , the new added > characters are mostly inferred wrong. > Is there a way to merge these models somehow? > > -- > 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 visit > https://groups.google.com/d/msgid/tesseract-ocr/70004fd7-72a5-4198-aecb-816ec7feda33n%40googlegroups.com > <https://groups.google.com/d/msgid/tesseract-ocr/70004fd7-72a5-4198-aecb-816ec7feda33n%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- 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 visit https://groups.google.com/d/msgid/tesseract-ocr/CA%2B6rfVGXPmWsTNorL9PgTbk1jJ1Uvg98-WU8ferxhVLe69qU-w%40mail.gmail.com.
