Have at https://github.com/tesseract-ocr/tesseract/issues/2342 and search
for "tesseract OCR dot matrix", there are several suggestions on how to
improve OCR results e.g.
https://jeffreymorgan.io/articles/improve-dot-matrix-ocr-performance-tutorial/
PS: it does not make sense to post custom traineddata - there could be
also a problem with how it was created/trained.
Zdenko
ne 2. 7. 2023 o 11:27 Simon Plackett <plackett...@gmail.com> napísal(a):
> Hi,
> I have a display I would like to OCR - example image is at the bottom,
> I have found a font that matches as far as I can tell exactly the number
> format
> ('5x7-dot-matrix') I have created 40k files similar to pil_image_10445.png
> with their corresponding .gt.txt files and created a new traineddata file.
>
> My character set is limited to 0-9 and .
>
> I have tried using random sets of characters, and a more structured set
> nnnnn.nnn, the results from all of the traineddata files is poor.
>
> I have also tried turning the image to grayscale, cropping, enhancing the
> contrast etc to no avail. I am lucky to get 1 digit recognised.
>
> Bizarrely I get the same output no matter which input image file I use!!
> (Using the attached traineddata file and the attached image I get 6.) -
> which is the same output for all the files I have tried.
>
> I have stuck to either psm 7 or 13 as the others largely don't give any
> output
>
> I would like some advice about whether continuing to increase the training
> data set will help, or any hints about trying to get better OCR success for
> these digits.
>
> I am using tesseract 5.3.0 leptonica 1.83.0 on a debian 11 machine.
> I built tesstrain as per the github instructions.
>
> I am using ./tesseract -l dot_gas_int --psm 7
> ~/tesstrain/data/dot_gas_int-ground-truth/pil_image_10445.png stdout
>
> Apologies if I am doing something dumb, this is new to me and I am having
> a go :-)
>
> Thanks
> Simon
>
>
> --
> 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/fa2232b3-852b-421e-939e-177971178faen%40googlegroups.com
> <https://groups.google.com/d/msgid/tesseract-ocr/fa2232b3-852b-421e-939e-177971178faen%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 on the web visit
https://groups.google.com/d/msgid/tesseract-ocr/CAJbzG8x3FBHzfm1sXm-aipqYMQMbjQmPEZpEH3SmKRE6PYmynA%40mail.gmail.com.
