Sistema di riconoscimento semantico delle azioni mediante matchmaking semantico

The gesture recognition framework proposed here is based on key posture detection. The retrieval process occurs in three steps: (i) posture description, which provides posture annotations; (ii) posture detection, which sequentially identi¯es a few reference postures, named key postures; (iii) gesture identi¯cation, which labels recognized gestures as sequences of key postures. Data capture is provided by the popular Kinect platform.