[PYTHON] I made a demo that lets the model learned in the Tensorflow mnist tutorial distinguish the handwritten numbers written on the canvas.

I have a feeling that the title says it I happened to have to make a public presentation on Deep Learning material, I wish I could explain it with some simple demos. I wonder if there is a sample that can be easily moved at hand and can actually make predictions I searched for this and that, but I couldn't find it, so I made it.

What I'm doing is using TensorFlow MNIST Tutorial as it is, saving the trained model, and saving it. All you have to do is skip the number image drawn on the canvas to the Flask app and let it be identified. I hope it helps.

demo: munky69rock/mnist-demo

(Note 1: Since I made it with time-constrained movement first, I have not verified the correctness of the code inside, I'm sorry ...) (Note 2: The browser has been confirmed to work only on Chrome, it does not work on smartphones)

reference

• Deep MNIST for Experts  |  TensorFlow -Use trained model with TensorFlow (application of Deep MNIST for Experts) --walkingmask ’s development log
• Create a Drawing App with HTML5 Canvas and JavaScript { William Malone }