Training Acropus 0.60

[JR]

Help in training ocropus.

I'm trying to improve my OCRing of some text that is not coming out well...

I've followed the guide of uw3-500 to train ocropus on my font set (english courier) as it was making way too many errors. Meaning, I collected a few pages of my text, created ??????.bin.png one liners, with .gt.txt text files..

It ran ok, and created me .cmodel .h5 and .tsplit files... What do I do with theses? Do I splitly copy them into /usr/local/share/ocropus ???

Sorry for my ingorance...

[Tom]

If you're getting too many errors on Courier, maybe your resolution is too low or too high?  Can you post a picture (best... add it to the Flickr pool)?

Nevertheless, back to your original question, once you have the .cmodel file, you just use it as an argument to ocropus-lattices, as in the sample script:

ocropus-lattices -m your.cmodel textline.png

ocropus-ngraphs textline.png

Note that if you want to be able to recognized mixed font text, you need to train on a variety of fonts in addition to the one you want to improve on.

Tom