Mole Detection App

Description

This a web app that classifies moles based on images. It uses transfered learning on the deep learning model MobileNetV2 and is trained on the this dataset.

The model is trained and then saved so it can be loaded in the Flask application and used for predictions.

Usage

Website: https://mole-doctor.herokuapp.com/ (Might not be running because of Heroku's memory limit).

Installation

Install required packages.

pip install -r project/requirements.txt

This script will build and save the model. This needs to be done at least once before hosting the app locally or in docker.

python model_building.py

Host the app locally, it will be running on localhost

python app.py

Docker

Build the docker image

docker build . -t mole-doctor

Deploy the docker image to a container and run locally. App will be running on localhost

docker run -d -p 5000:5000 mole-doctor

You should see the mole doctor if the application is running correctly