100 likes | 238 Views
VL 2011 , 1 5 .09.2011, Brighton, UK. Spatio-Temporal Visual Ontolog y. Dr Joanna I sabelle Olszewska. VL 2011 , 1 5 .09.2011, Brighton, UK. Contents. Ontology Definition Spatio-Temporal Visual Ontology (STVO) Motivation Concepts Relations
E N D
VL 2011, 15.09.2011, Brighton, UK. Spatio-Temporal Visual Ontology Dr Joanna Isabelle Olszewska
VL 2011, 15.09.2011, Brighton, UK. Contents • Ontology Definition • Spatio-Temporal Visual Ontology (STVO) • Motivation • Concepts • Relations • Application: STVO-Coupled Active Contours • Experiments • Evaluation • Conclusions
VL 2011, 15.09.2011, Brighton, UK. Ontology Definitions • In Philosophy (IVth century BC) • « study of being » • In AI (1980’s) • specific vocabulary used to describe a certain reality • + a set of explicit assumptions regarding the meaning of the vocabulary • In Computer Science (1993) • « specification of a conceptualization »
VL 2011, 15.09.2011, Brighton, UK. Spatio-Temporal Visual Ontology • Motivation • mapping of low-level descriptor values to higher-level semantics • bridging the gap between visual features and semantic knowledge
VL 2011, 15.09.2011, Brighton, UK. Spatio-Temporal Visual Ontology • Concepts
VL 2011, 15.09.2011, Brighton, UK. Spatio-Temporal Visual Ontology • Relations • Spatial • Topological relations: Region Connection Calculus (RCC-8) relations • Directional relative postions: o’clock intra and inter-object relations • Relative distances (close/far relations) • Temporal • Visual RCC-8 topological relations O’clock inter-object spatial relations O’clock intra-object spatial relations
VL 2011, 15.09.2011, Brighton, UK. STVO-Coupled Active Contours • Experiments
VL 2011, 15.09.2011, Brighton, UK. STVO-Coupled Active Contours • Evaluation • Precision of detection • Precision of answers = # correct contours / # targets = # correct answers / # queries
VL 2011, 15.09.2011, Brighton, UK. Conclusions • Spatio-Temporal Visual Ontology (STVO) • Granular conceptualization and a hierarchical representation of the visual observation domain • Semantical characterization of each visual scene and its components (e.g. objects of interest) • New sematically meanignful inter- and intra-object spatio-temporal relations • Support of the reasoning about photomoetric, geometrical and spatio-temporal relations between and within the multiple observed objects or parts of tem for the effective interpretation of visual information. • Application to Dynamic Visual Scene Analysis
VL 2011, 15.09.2011, Brighton, UK. Spatio-Temporal Visual Ontology ThankYou Contact Details Dr Ir Joanna Olszewska j.olszewska@hud.ac.uk School of Computing and Engineering University of Huddersfield, Queensgate Huddersfield, HD1 3DH, United Kingdom