On Thursday, December 7, 2017 at 11:55:53 PM UTC-5, shree wrote:
>
> Please check the last section on
> https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract-4.00
>
Thank you for this tip. I'm getting farther than before. I thought
--trainedata was my final traineddata output file.
I now made the final eng.trainedata 'lstmtraining --stop_training ...." as
follows
$tesstrain_dir/lstmtraining \
--stop_training \
--continue_from $train_output_dir/pluschars_checkpoint \
--traineddata $train_output_dir/eng/eng.traineddata \
--U $train_output_dir/eng/eng.unicharset \ # not sure if this is
necessary; doesn't make a difference
--model_output $final_trained_data_file
And I get a $final_trained_data_file that I can use to replace
/usr/local/share/tessdata/eng.traineddata and it doesn't fail on init3()
any more. But it doesn't recognize any of the new chars either.
However, in running
/usr/local/bin/tesseract-training/lstmeval \
--model ./trained_plus_chars/pluschars_checkpoint \
--traineddata ./trained_plus_chars/eng/eng.traineddata \
--eval_listfile ./trained_plus_chars/eng.training_files.txt
it DID recognize the new chars most of the time. So I think there may
still be something something wrong with the construction of the --model_output
$final_trained_data_file.
My entire training sequence bash script is here:
*https://pastebin.com/gNLvXkiM*
Can you tell if there is anything obviously wrong?
Thanks
--
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 [email protected].
To post to this group, send email to [email protected].
Visit this group at https://groups.google.com/group/tesseract-ocr.
To view this discussion on the web visit
https://groups.google.com/d/msgid/tesseract-ocr/10194cda-9e8d-494c-ae4a-157e3d25f913%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
