Is there a way to prepare ".box" files using your own "traineddata"?

27 views
Skip to first unread message

Rafael Maciel K

unread,
Oct 15, 2015, 3:31:15 PM10/15/15
to tesseract-ocr
I've been training tesseract to recognize well a new font. I already generate a new "traineddata" for this font, and now i want to add more data to my training set.

The problem is that when I generate the ".box" files, looks like its using the "eng.traineddata" to make the recognition. How can I generate them using my own ".traineddata" (since at this point my ".traineddata" is more accurate for this font, it would generate less rework)?
Reply all
Reply to author
Forward
0 new messages