Hi
Thanks for the links, Im interested in understanding how tesseract works for Arabic language and I need to document this in my thesis, I read most the docs in the links and I still have some questions if you could help me understand those:
* What are the inputs to the LSTM? are they images of words or characters or maybe feature vectors? What is it responsible for?
* In the word recognizer what is the structure of the Adaptive and the static character classifiers? They are Neural nets right? what type of Neural nets?
* The character chopper is explained using English but how does it work for cursive languages like Arabic for example?
Thanks,