Semantic matchmaking for Kinect-based posture and gesture recognition

Innovative analysis methods applied to data extracted by o®-the-shelf peripherals can provide
useful results in activity recognition without requiring large computational resources. In this
paper a framework is proposed for automated posture and gesture recognition, exploiting depth
data provided by a commercial tracking device. The detection problem is handled as a semanticbased
resource discovery. A general data model and the corresponding ontology provide the
formal underpinning for posture and gesture annotation via standard Semantic Web languages.
Hence, a logic-based matchmaking, exploiting non-standard inference services, allows to: (i) detect
postures via on-the-°y comparison of the annotations with standard posture descriptions
stored as instances of a proper Knowledge Base; (ii) compare subsequent postures in order to
recognize gestures. The framework has been implemented in a prototypical tool and experimental
tests have been carried out on a reference dataset. Preliminary results indicate the feasibility of the
proposed approach.