MNIST_DrawAndClassify_01

KIVY-Keras interactive MNIST classifier

This UI allows testing of three pre-trained CNN that classify 28x28 grey scale hand-drawn images of the numbers 0-9 from the classic MNIST dataset.

Draw a number - then classify and analyse it.

Switch between CNN using the Network buttons.

Chose a human classification to archive the image with the MC (CNN) and HC (Human) classifications. Over time this should build a folder of labelled images.

Also works on touch screen eg MS Surface Hub.