Skip to content

Latest commit

 

History

History
27 lines (22 loc) · 907 Bytes

README.md

File metadata and controls

27 lines (22 loc) · 907 Bytes

This program recognises numbers drawn on a canvas.

It uses the MNIST dataset and is one of the projects that was taught in the University of Toronto Mississauga's Machine Learning Workshop in the Hack Lab.

SPECIAL THANKS TO

Muratovm & cheesywow
They gave me the inspiration and provided Painter.py.

SETTING UP

  • Install Anaconda:
  • Install Keras:
    • sudo ipython -m pip install keras
  • Test that Keras is installed:
    • ipython -c "import keras; print keras.__version__"
  • Install h5py using:
    • sudo ipython -m pip install h5py

USING IT

  • Train the CNN:
    • run:mnist_cnn.py
  • Open Painter to test CNN:
    • run: Painter.py

RESOURCES USED