Machine Learning

Step 1 - Capture movement and creation of the learning dataset 

Each file in the dataset had information of the kinetic properties on time of one movement detailed for each of the 17 sensors. 

Graph depicting the overall process.

Step 2- Movement anotation - to each file with kinetic information about the captured movement, we had to ascribe the indication of the frames of begining and start of action, and the respective quality of movement.  

Step 3 - Model training. 

Step 4 - Automatic recognition

Experimental tryout of the ML model reacting in real time to a dancer using XSens/Awinda suit.