This repository is example of serving Keras model.
Trained Keras model's predict result will be returned via web api.
Web api is served by Flask.
app.py is a Flask app.
Api will predict a number from image which is included in request and return the predict result.
Also app.py includes web app which has canvas, and you can write a number 0 to 9.
static and templates directories are for web app.
train.py is a training keras model script.
Training task is MNIST.
After training, parameter's are saved to model directory.
load.py is for loading trained model.
It will load model from model directory.