A demo of a Convolution Neural Network trained on MNIST data. The CNN is exposed as an HTTP endpoint using Flask. The application is deployed as a Docker container on an EC2.
- static/main.js
- This file creates the canvas, implements drawing functionality, and sends base64 encoded image of the canvas to the server.
- cnn.py
- This file creates the CNN Pytorch model.
- server.py
- This file is the entire Flask server which handles '/' and '/classify' routes.
