Demo: gesture based interaction with the Hololens 2

Gesture recognition is one of the default interaction modalities in many XR applications, although the gesture types recognized by many applications is typically limited to few static poses. In this demo we show that a recent network-based solution for online, sliding window, gesture classification from hand pose streams (On-Off deep Multi-View Multi-Task) can be used for the simultaneous detection and recognition of heterogeneous gestures, including dynamic coarse and fine grained ones, enabling interaction designers to create novel ways to perform interactive tasks that can be applied to different domains.