Ajak Final Project

This is our final project at Makers Academy, presented at Demo Day on Friday 24 May 2019.

Visit our web app on Heroku!

Getting started | Project description | Technologies | Manifesto | Learning documentation | Authors | Acknowledgements

Getting started

git clone https://github.com/amyj0rdan/ajak-final-project
pip3 install -r requirements.txt # to install python dependencies
npm install # to install node dependencies

To download data

Download Crown, Camera and Rabbit from Google QuickDraw Dataset in numpy-bitmap format and save to /data folder in the project under crowns.npy, cameras.npy and rabbits.npy.

To train model

python3 model_config/train.py

When prompted by running the above command, save the model as cameras_rabbits_crowns_model. Move the saved model to the /models folder.

To see model predict random image from test data

python3 model_config/predict_on_command_line.py

To run tests

The pytest framework is used for unit testing.

To run tests:

pytest

To run test coverage:

pytest --cov=lib

To run linter

pylint [options] module_or_package

For example:

pylint lib

Project description

Our project is pictionary played against a model trained to recognise three drawings:

• crown
• camera
• rabbit

A user draws on a canvas against a timer. The model then predicts which of the above the user has drawn and the prediction is displayed to the user.

Technologies

Backend

• Python3
• Flask

Testing

• Pytest
• Splinter

Machine learning libraries and data

• Scikit-learn
• Keras
• TensorFlow
• Google QuickDraw Dataset

Frontend

• JavaScript
• jQuery
• Fabric JS

Deployment

• Travis CI
• Heroku

Manifesto

Our project manifesto has individual and team project goals, and our ways of working.

Our Trello board.

Our presentation from Demo Day at Makers Academy on Friday 24 May 2019.

Learning documentation

See our wiki.

Authors

Alex Chen, Amy Jordan, James Palmer, Kim Diep.

Acknowledgements

Makers Academy