Handwritten digits recognition

This a Flask web app to predict handwritten digits using a deep convolutional neural network built with Keras and trained on the MNIST dataset.

Training the model

To re-train a different model or see the existing model, check the training notebook

Running the app locally

Prerequisites

• Docker
• Docker Compose

Installation

1. Clone the repo:

git clone https://github.com/ihebu/digit-recognizer.git

2. Navigate to the project directory:

cd digit-recognizer

3. Build the docker images

docker-compose build

4. Start running the containers

docker-compose up

Usage

to use the app, simply go to http://localhost