***************************************************************************
I recommend you look at my other repository containing a much faster OpenCL
implementation of the neural network over this current repository. There
is also a nice demo available in the OpenCL repository!
***************************************************************************

This project is comprised of three parts

The first part is a simple python implementation of a feed forward 
neural network that is trained through backpropagation. 

The second part is a collection of scripts as well as the images in the 
data folder. The scripts preprocess the images and then uses the
neural network to build a digit recognition neural network.

The third part consists of two scripts; one script called ExtractCharacters.py,
takes an image and attempts to extract the individual characters from it, and 
then converts it into the feature vector. The other script takes the feature 
vector, feeds it to a supplied neural network and prints out its guess of the
numerical representation of the digits.


HOW TO TEST THIS:

The script testMultipleNumbers.py reads the image from ./data/number/test.png
and then attempts to classify it with the neural net stored under
BestNet.pkl. To see this,draw your own digits in the ./data/number/test.png file.
Then run testMultipleNumbers.py and press "Enter" to tell the program to
read the file and parse the digits. Each time you press "Enter" the program
will reload the file and attempt to parse the digits.