[this might be a repost; the first attept didn't show up] I'm using the C API of tesseract 4.0 on OS X, and I tried to add some more characters. (4.0 seems much better than 3.x, I should add - thanks to everyone who made this possible!)
I used this manual section: https://github.com/tesseract-ocr/tesseract/wiki/TrainingTesseract-4.00#fine-tuning-for--a-few-characters as a guide to construct the following script: https://pastebin.com/4n2mRSpq Before running, I modified langdata/eng/eng.training_text with the extra chars, maybe 15 instances of each, as instructed. I'm using only a subset of the original training fonts, but I figure it is OK, since I'm adding only a few distinctive chars. The NN optimizer lstmtraining ran, and gave a bunch of checkpoints, and a final file $train_output_dir/eng/eng.trainedata But this eng.traineddata was 5MB when the original one was 15.4MB. And when I tried to copy it over the pre-loaded 'best' eng.traineddata and run tesseract it failed in TessBaseAPIinit3 with error=-1. Does anyone know why 1) my eng.trainedata is so much smaller and 2) why it fails to even load in API init()? Thanks for any tips! -- 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 post to this group, send email to tesseract-ocr@googlegroups.com. 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/50c6b233-602e-4479-a518-3bfd6baa10c9%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
