Abstract
Resource management has been a long standing issue in capacity-constrained organizations. Using video technologies may be a means to an end for this problem by monitoring and analyzing patterns of usage. However, due to wide field of views of surveillance cameras, several resource stations may be viewed on the same camera or different cameras. Therefore, setting a static threshold is not a good solution as several stations require different threshold on different camera views due to resource usage variability. In this paper, we introduce a new approach to dynamic notification generation when a situation of interest is discovered for video analytics. We applied motif detection to tracking of resource usage using information from observed video stream. In this approach, resource usage and replenishment pattern are automatically used to make dynamic alert notification threshold decisions by altering and reacting to their dynamically changing environment. We show that with the increasing usage of resources due to people inflow, the motif network leads to optimal threshold decision than a randomized network or naïve approach. This work is an extension to the trolley tracking system we developed and implemented in Hong Kong International Airport (HKIA) for resource monitoring.