Funding & Support
Open Source Software
GMU C4I Center Seminar
Co-sponsored by AIT Department
Context Adaptive Event Ontology for
Personal Photo Annotation
Tuesday, May 8, 2012 at 2:00 PM
Nguyen Engineering building, Room 4705
People love to take pictures to represent varied subjects and actions. This gives them the ability
to capture what makes them who they are. The advances in technology have led to the explosion of user
generated content, and to the problem of better photo mining and retrieval efficacy. Dealing with this
problem requires annotating photos with universal tags that are interpretable by other people. In this
paper, we propose a better quality tagging machinery for personal pictures that are taken with modern
cameras. Our approach annotates photos based on personal actions that are the abstractions that come
from events. We use the model of a specific event that is adapted to a particular personal context using
all the information and then used for assigning appropriate tags to photos. The model includes the structural
aspect of the event because for events, we have sub-events. Ontological event models are used to represent
event-subevent-object-place relationships. We present our experiments using 1889 photos in 18 albums.
SETAREH RAFATIRAD is currently a PhD Candidate at the Department of Computer Science at the University of
California, Irvine. Her current research is on event-based personal photo annotation, where event ontology
and a combination of information sources are used in the solution strategy. Setareh received her Masters
degree from UC-Irvine in 2009. During her M.Sc. studies she worked on modeling of events, event relations,
composite event operators using event structure, and applying events in the context of semantic computing.
Setareh was a recipient of the two-years UC-Irvine School of Information and Computer Science Dean's fellowship.
She is a member of the IEEE and the ACM.