Tessaract OCR uses neural networks as the default learning mechanism

Question

Sorry, this is probably a dumb question. but I'm pretty new to computer training and Tessaract OCR. I heard that Tessaract OCR can be trained.

What I need to know is does Tessaract OCR use neural networks as the default learning mechanism, or should we program it explicitly to use neural networks?

Sorry if I think incorrectly about this concept of learning. but what I need to know for sure is Tessaract already using NN, or if not, can I use NN with Tessaract OCR to improve recognition accuracy?

If you can offer me some good resources / way to refer / try and get started, this will be a big help too.

what I now know about the basic concept of machine learning to learn and to perform the basic OCR image operation in OCR Tessaract.

Answer 1

It seems that Tessaract uses the Adaptive Classifier by default. Check this out for a good read:

https://github.com/tesseract-ocr/docs/blob/master/tesseracticdar2007.pdf

There seems to be an option called “Cube Mode” in which it switches to using NN for the training system instead of the adaptive classifier ( https://code.google.com/p/tesseract-ocr-extradocs/wiki/Cube ). Additional information about adaptive classifiers:

http://www.cs.indiana.edu/~rawlins/website/adaptivity/information-helper.html

In addition, the training classifier system is very closely related:

http://en.wikipedia.org/wiki/Learning_classifier_system

In addition, your terminology for “learning” is very close. Learning is how you teach an image recognition system or a learning system, what answers it should give to specific input sets. He then uses affinities when he discovers unknown data to classify new data. Machine learning is one of the coolest fields, in my opinion (probably a biased opinion, but whatever!), Keep learning! You are a meta-student: learn how to teach a machine to learn! Cool stuff!