OCR with the caffe deep learning framework
DEPRECATED -> Moved to https://github.com/pannous/tensorflow-ocr
use script to generate more training data, or get more training data from here: http://ufldl.stanford.edu/housenumbers/
Also see https://github.com/mateogianolio/mlp-character-recognition for simple js example and https://github.com/tmbdev/ocropy for a very comprehensive phython toolchain.