WORKSHOP ON CONCEPTUAL DESCRIPTIONS FROM IMAGES
              Fourth European Conference on Computer Vision

                              Cambridge, UK
                             19th April 1996

An international workshop on `Conceptual Descriptions from Images' will be held
at the University of Cambridge on April 19th, following ECCV'96.  This activity
is linked to the ECVNet working group of the same name ECVNet.

PROGRAMME COMMITTEE

Mike Brady (Oxford, UK)
Hilary Buxton (Sussex, UK)
Tony Cohn (Leeds, UK)
Peter Hall (Wellington, New Zealand)
David Hogg (Leeds, UK)
Jana Kosecka (Pennsylvania, US)
Paul McKevitt (Sheffield, UK)
Amitabha Mukerjee (Kanpur, India)
Hans-Hellmut Nagel (Karlsruhe, Germany)
Bernd Neumann (Hamburg, Germany)
Patrick Olivier (Aberystwyth, UK)


WORKSHOP ISSUES

The workshop aims to stimulate  collaborative research  within the AI subfields
of computer vision,  natural language,  and spatio-temporal reasoning,  working
towards  the goal of deriving meaningful descriptions  of  scene content.  Most
current work in computer  vision  develops  automated techniques  that  extract
information from  images without high-level knowledge of what is being seen  or
consideration of the high-level goals of intelligent visual agents. However, to
deal with real-world visual tasks,  it seems that contextual knowledge  of some
sort is  required.  The  issues  we want to  address  involve  questioning what
representation  and reasoning  is required for such tasks.  In particular,  for
many  visual  application  systems (for example,  advanced visual surveillance,
multimedia and geographic information systems, even automated driving) there is
a  general need  to  generate  conceptual  descriptions  of  objects  and their
behaviour from images. Papers should address this central theme.

Research  topics  that  address the issues of representing visual knowledge and
processing images to obtain conceptual descriptions include:

   o  Context-based vision.
   o  Integration of vision and natural language.
   o  Spatial and temporal reasoning in images.
   o  Representation and control of visual behaviours.

In addition to presenting original research,  participants are asked to address
the following questions:

   o  How does  your  work  draw upon,  differ from,  refine or extend existing
      computer vision, natural language, and AI approaches?
      What are the limitations and assumptions of your approach?
   o  How  should  domain  knowledge  be  represented?  What is your underlying
      knowledge  representation  and reasoning  formalism and what issues  have
      motivated your choice?
   o  What are the open questions?


ATTENDANCE

It  is  intended that between  30 and 50 people will attend  the workshop.  All
workshop participants are expected to register for the main ECCV conference.


SUBMISSION GUIDELINES

Papers must be a maximum of 30 pages.  Title,  abstract, figures and references
must be included within this length limit.  Four copies should be mailed to the
address  below.  Double sided printing is encouraged.  Electronic submission is
also encouraged (self-contained LaTeX).

Hilary Buxton
School of Cognitive and Computing Sciences
University of Sussex
Falmer
Brighton, BN1 9QH, UK
E-mail: hilaryb@cogs.susx.ac.uk
Tel: +44 1273 678569
Fax: +44 1273 671320

For information on ECCV'96 see the conference home page

DEADLINES

Submission deadline: 20th January 1996
Notification of acceptance: 17th February 1996
Camera ready copy due: 9th March 1996


PUBLICATION

Accepted papers will be published in the workshop proceedings  and distributed
to all workshop participants.


Patrick Olivier
Centre for Intelligent Systems
Department of Computer Science       Tel: +44 1970 622447
University of Wales                  Fax: +44 1970 622455
Aberystwyth                       e-mail: plo@aber.ac.uk
Dyfed, SY23 3DB, UK