If that is the original quality of image used for ocr, the quality is
really poor. Also Tesseract OCR is not designed to recognize handwritten
text. Improving quality of image should work at least for the printed part
of text.
Hope it helps,
Martin
Dne úterý 14. března 2017 8:26:52 UTC+1
Hi,
I was trying to train tesseract 4 to recognize trademark symbol ™. I was
following examples on wiki:
https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract-4.00---Replacing-Top-Layer-Example
https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract-4.00---Replace-Top-Layer
2 matches
Mail list logo