Action: A framework for semantic annotation of events in video

Abstract

We propose a model for semantic annotation of events, such as weddings or birthday parties, as depicted in video. Our framework consists of an event taxonomy, implemented as a faceted classification, and an event partonomy, implemented using the ABC ontology proposed by Lagoze and Hunter [1]. Our approach enables the annotation of a low-level physical action depicted in video, such as a kiss, to be linked to its higher-level event context (such as the kiss that signifies the conclusion of a Western wedding ceremony).