Machine Learning predicitons in Web app

Working on approach to use Tensorflow.js library for real time predictions in an web app.

MNIST is a popular dataset of images of handwritten digits link

There are various models to predict Handwritten Digits. This project uses a Convolution 2D model to predict the digits. The dataset has various images to train the model. If this model is deployed on a web application then the trained ML model can be used to predict digits written on browser.

Key Features about the project -

* Uses Conv2D model to predict handwritten digits
* Deployed ML model to Node.js WebApp
* Integrated model predictions in JavaScript code
* Developed CI pipeline for Master branch

Getting Started -

To get started with the project follow the below metioned steps -

* This project requires Node.js.
* In the terminal of project folder, type npm install to install the dependencies.
* Once the dependencies are installed, type npm start to run the project locally.
* You can see the app on localhost:4545/ in your browser.

Editing the ML model

* To edit the ML model you will need pip to install the Python packages. 
* Once pip is installed in your computer, install the libraries.
* Now the Jupyter Notebook can be edited.