Draw a number from 0 - 9 and click Predict. The trained neural network will try to predict the number.

This is a learning project to demonstrate the working of Neural Networks. The Neural Network is trained on MNIST Handwritten digits. It contains 70,000 images with 28x28 pixel each. 60,000 images are used for training the network and rest 10,000 samples are used for testing. The neural network architecture has 1 input layer with 784 input nodes, 2 hidden layers each with 16 hidden nodes, and 1 output layer with 10 nodes.