Slip and fall event detection using Bayesian Belief Network

This paper proposes a method to detect slip-only events and fall events based on the motion activity measure and human silhouette shape variations. Here, we also apply the Bayesian Belief Network (BBN) to model the causality of the events before and after the fall and slip-only events. The motion measure is obtained by analyzing the energy of the motion active (MA) area in the integrated spatiotemporal energy (ISTE) map. Unlike the motion history image (MHI), the ISTE map can be applied to detect fall and slip-only events. The contributions of this study are: (a) proposing the ISTE map; (b) detecting the fall parallel to the optical axis; (c) application to non-fixed frame rate video; (d) identifying the slip-only event; and (e) using BBN to model the causality of the slip or fall events with other events. Early identification of a slip-only event can help prevent falls and injuries. In the experiments, we demonstrate that our method is effective in detecting both fall and slip-only events.