SEMINARIOS DE INVESTIGACIÓN EN INGENIERÍA INFORMÁTICA Y DE
TELECOMUNICACIÓN 2007-2008
Actividad de Formación Continua del Programa Oficial de Posgrado en Ingeniería
Informática y de Telecomunicación
Escuela Politécnica Superior, Universidad Autónoma de
Madrid
jueves, 24 de Abril de 2008, 12:00
Salón de Grados, Escuela Politécnica Superior,
Universidad Autónoma de Madrid
Contextual Content Analysis for Semantic
Knowledge Extraction
Noel O'Connor
Centre for Digital Video Processing, Adaptive Information Cluster, Dublin City
University, Irlanda
Abstract
Extracting semantic real world information from content is a key enabler for
many different digital media applications. The key challenge to be addressed is
the so called "Semantic Gap" problem. This has been a key driver for research
efforts in a number of traditionally disparate research communities. This can be
seen in the recent work of researchers targeting audio transcription, aspects of
computer vision, multi-modal indexing and retrieval and the general Semantic Web
effort, for example. Whilst this has stimulated much useful research, and led to
many publications and PhDs, increasingly we can see that the Semantic Gap
problem is intractable except in very constrained applications. This is because
what we perceive from what we see and hear, depends on factors such as who we
are, where we are and when and why we're looking or listening in the first
place. None of this can be measured easily (if at all) directly from a single
content source. In this presentation, I will argue that useful advances can be
made by considering either multiple similar data sources (e.g. visible and
thermal visual modalities) or non-content contextual data sources (e.g. from
simple time, date and location to more complex channels such as physiological
data from wearable technology). Both approaches are now possible in practical
applications because of the increasing availability / decreasing cost of a
variety of sensor modalities. I will show some examples of initial research in
my lab along these lines in the framework of applications such as personal photo
management, event detection in visual diaries, and activity analysis for sports
applications.
Noel O'Connor
Noel E. O’Connor graduated from
Dublin City University (DCU) with a B.Eng. in Electronic Engineering (1992)
and a PhD (1998), after working for 2 years as a research assistant for Teltec
Ireland at DCU. He is currently a Senior Lecturer in the
School of Electronic
Engineering and a PI in the
Centre for Digital Video
Processing (CDVP), a 55-person research centre, which he co-founded with
Alan Smeaton.
Since 1999 he has published over 130 peer-reviewed publications, 11 standards
submissions, filed 5 patents and spun off a campus company, Aliope Ltd, with
others from CDVP. He is the Irish representative to the world-wide
ISO/IEC Moving
Pictures Expert Group (MPEG) standards body. He acted as integrator for the
software implementation of MPEG-4 that now forms an informative annex to the
standard and was part of a small team whose technology proposal to MPEG-7 was
selected as the starting point for the standard. He has acted as PC Chair for 3
international conferences (CIVR
2004, SAMT 2006 &
WIAMIS 2007) and
regularly reviews for a number of respected journals and acts as a PC member for
many international conferences (8 in 2006). He was an expert evaluator for EU
FP6 proposals as well as acting as a reviewer for specific projects for both the
EU and Enterprise Ireland. He has guest edited 5 special issues of different
journals including
Multimedia Tools and Applications (Springer), the
Journal of Web Semantics (Elsevier),
Signal Processing: Image Communication (Elsevier), the
Journal of
Embedded Systems (Eurasip) and the
Journal of Image
and Video Processing (Eurasip). He is a member of the IEEE, Engineers
Ireland and the Institution of Engineering and Technology.