FUBI, otherwise known as the Full Body Interaction Framework was designed as a framework for recognizing full body gestures and postures. This is done in real time from the data of an OpenNI-applicable depth sensor, especially the Microsoft Kinect sensor.
The Framework therefore distinguishes between four gesture categories:
· Static postures: Configuration of several joints, no movement.
· Gestures with linear movement: Linear movement of several joints with specific direction and speed.
· Combination of postures and linear movement: Combination of 1 and 2 with specific time constraints.
· Complex gestures: Detailed observation of one (or more) joints over a certain amount of time.
What's New in This Release:
· Complete redesign of the .NET CS GUI, many new options, switched to .NET 4.0
· Refactored FubiUser class and added FubiXMLParser class
· Now optionally using 1€ filter for easy and efficent joint data filtering, see Casiez et al. in CHI 2012 proceedings. Filtered data usabable in all recognizers and for the tracking data rendering. The two parameters (filter strength for still joints, and how much it is reduced according to the movement) also adjustable in the FUBI GUI.
· Fix for the face joint references and their confidence values
· Some more fixes for platform/compiler independence (thanks to Christian Frisson and Fabien Grisard)
· Fix for specific combinations of basic relations/directions in XML
· New options for combination recognizers: onFail for going back one state instead of a complete restart
· Fixed Kinect SDK IR stream integration
· Better OpenCV integration with automatic version check
· Defines all have a "FUBI" prefix to avoid conflicts
· Added linear acceleration recognizers (currently ...