This project is part of this report that can be found on my website. It involves using HOG Features for feature extraction and using this as the input to a 3 layer neural network.
The pipeline for hand tracking and classification is shown below.
Object localization is performed using background subtraction and Color Segmentation. The extracted subset of the image using these two methods is combined together using a weighting function and the result is a binary output with an example shown below.
This project is no longer maintained. If you would like more information to get it up and working, contact me at omar(dot)abid4(at)gmail(dot)com