README_HMM

### Readme for the HMM (Hidden Markov Model) classifier in the gesture recognition application

The HMM is a stochastic Model for sequential Data, which uses markov model to model hidden states.
Each HMM represents a gesture in the application, every state (frame of a gesture) is represented 
by an GMM (Gaussian Mixture Model)

The HMM-Module provides training and classifying given record data. It also can be used to plot the raw, 
preprocessed and sample data from trained GMM. 


### Usage:
Config parameters are explained in config/default.cfg. On first start a config/personal.cfg file is created. Changes can be made here. 
Two already trained networks are provided and specified in "classifier.hmm.config.py" 
You can choose:
0	all gestures provided by the application, trained with all data. gesture 0, 1, 2, 4, 6, 7
1	3 gestures trained with well formed data - classifying work kinda good.
	gesture 0, gesture 1
	gesture 5 changed to "Bottom-To-Top-One-Hand"
	 


some commands for console:
u hmm 				- loads HMM module, with configured gesture set 
c					- start classification
l ds gesture_number - (Overloaded) starts plot class and show raw training data
	right click 	- change mode: raw, preprocessed or gmm samples
	mousewheel		- change data index
	0-7				- change gesture