Oz is a system for gesture-based authentication using a webcam.
The camera
directory contains files for the acquisition of images from the
webcam, processing those images, and the machine-learning algorithm files.
The extension
directory contains files for the Oz Chrome extension.
The util
directory contains scripts that allow Oz to text a user for the
password-reset functionality.
To use Oz, run camera/oz.py
in the background and load the extension as
a Chrome plugin. It is also necessary to first train the gesture classifier
using gather/trainer.py. The Chrome plugin makes calls to a running instance of
oz.py
to get current hand gestures.
The system relies on slurpy (a library to allow javascript-python communication), scikit-learn for classification, and opencv for isolating the hand image.
Created for COS436 - Human Computer Interaction