IEEE Workshop on

Perceptual Organization in Computer Vision

POCV '04

In association with the IEEE Conference on Computer Vision and Pattern Recognition

Washington, DC, USA

June 28, 2004

Contacts:
hoogs@crd.ge.com
medioni@iris.usc.edu

 

NEW: The program is available here.

A workshop overview is available here.

  Workshop Information

Perceptual Organization (PO) is a rapidly growing area of computer vision research.  As the link between low-level segmentation and high-level algorithms, PO has the potential to enable figure-ground separation, object recognition, scene reconstruction, change detection, spatio-temporal grouping, and many other areas.  Recent progress in PO has encouraged more participation from experts in related areas such as texture, motion analysis and recognition.

This workshop will bring together researchers in PO and related areas to share novel ideas, future directions of PO, and current work. Following up on the successful POCV workshop at ICCV 1999, which produced a book describing current and future directions of PO, the theme for 2004 will be "the next generation of PO."  Participants will present their ideas and opinions on where the field is going, what people should be working on, what the main challenges are, and what results we can expect in the next few years.  One important theme is the role of perceptual organization and segmentation in patch-based object recognition methods, which have shown remarkable progress in the past few years without the use of traditional perceptual organization.

To encourage this, submitted papers do not need to describe fully developed algorithms, methods, or results as would normally be required for acceptance at a conference.  Supporting results would be useful, of course; however, authors are urged to take intellectual risks and argue for ideas that do not yet have experimental support.

The workshop format will support the speculative theme by allowing significant time for discussion and debate, including panel discussions.  Ideally, the workshop will provide a forum for people to collectively think about and discuss the next big leaps in the field. Invited speakers will provide perspectives from fields related to PO, such as psychology and cognitive science.

Papers were solicited in all areas of perceptual organization, including but not limited to:

·        image segmentation

·        feature grouping

·        texture segmentation

·        spatio-temporal/motion segmentation

·        figure-ground discrimination

·        unification of segmentation, detection and recognition

·        PO for object recognition

·        contour completion

·        psychophysically-motivated methods

·        medical image analysis

·        learning in PO

The proceedings will be published electronically by IEEE in the online IEEE publications database.

Paper Submission

The paper submission deadline is now after CVPR decisions.  Since the workshop will emphasize papers of a different nature than those in CVPR, accepted CVPR authors are encouraged to submit a related, but more far-reaching, visionary paper.  This policy is intended to encourage participation from authors whose papers are accepted in CVPR as well as those that are not.

Reviewing will be blind circular.  By submitting a paper, each author agrees to review at least 2 other submissions.  This procedure has produced quality, useful reviews in previous POCV workshops.  As reviewers, authors must comply with the reviewing schedule below.  In addition, each paper will be reviewed by the program committee.  Given the exploratory nature of the workshop, reviews will emphasize the novelty of the ideas and clarity of presentation, and are not expected to be comprehensive (or time-consuming).

 

Organization

General Chairs

Anthony Hoogs, GE Research

Gerard Medioni, University of Southern California

Program Committee

Kim Boyer, Ohio State

James Elder, York University

David Jacobs, University of Maryland

Ben Kimia, Brown University

Michael Lindenbaum, Technion IIT

B.S. Manjunath, UC Santa Barbara

David Martin, Boston College

Peter Meer, Rutgers University

Sudeep Sarkar, University of South Florida

Eric Saund, Palo Alto Research Center

Jeffrey Mark Siskind, Purdue University

Shimon Ullman, Weizmann Institute

S.C. Zhu, UCLA

http://www.research.ge.com/vision/pocv2004.html