Trajectory Annotation and Retrieval Based on Semantics

Abstract

Recently, the rapid increase of the available amount of multimedia information has revealed an urgent need for developing intelligent methods for understanding and managing information. There are many features such as color, shape, texture and motion for semantic information in video data. Especially, the most important semantic information is based on a trajectory which is the significant factor for event representation of objects in video. In this paper, we focus on semantic representation using topological and directional relations between non-moving and moving objects. In the experiment part, we compared retrieval results using TSR(Tangent Space Representation) with those using rules represented by the proposed model. We extend queries and motion verbs in a specific domain (not general verbs) and apply the proposed method to an automatic annotation or narration system.