A video monitoring system using ontology-driven identification of threats

In this paper, we present a video monitoring system that leverages image recognition and ontological reasoning about threats. In the solution, an image processing subsystem uses video recording of a monitored area and recognizes known concepts in scenes. Then, a reasoning subsystem uses an ontological description of security conditions and information from image recognition to check if a violation of a condition has occurred. If a threat is detected, the system sends an appropriate notification depending on configuration. A proto-typical implementation of the solution is presented together with exemplary monitoring results.