Digit Recognizer

A beginner's take on deployment of an ML model for digit recognition trained on the MNIST dataset. Built to be hosted locally only, and to learn more about deployment of ML APIs.

Client

To start the client locally, use a Live Server development server to host the client HTML page.

Server

1. Create a conda environment and install the required dependencies with conda.

• conda env create --name fastapi-ml --file environment.yml

2. In the server directory, run the following command to start a uvicorn ASGI development server.

• uvicorn main:app --reload

Possible Future Features

1. Using a CNN to better learn the patterns in the image.
2. Online Learning by confirmation of the correct result from the user.
3. Understand other possible preprocessing techniques (because of the downsampling, canvas properties) that might help.