Announcement and Call for papers for:
                                Workshop 
              Performance Characteristics of Vision Algorithms


Organizer:          Patrick Stelmaszyk, ITMI, Grenoble
Programme Chairs:   Wolfgang Foerstner, Bonn and
                    Henrik I. Christensen, Aalborg
Place:              Cambridge, UK
Time:               19-20 April 1996.
                    Friday/Saturday (morning) after the ECCV '96


Program Committee:

  Kevin Bowyer, USF         Hans Burkhardt, Hamburg      Adrian Clark, Essex.
  Patrick Courtney, ITMI    Al Hanson, UMASS             Bob Haralick, UW
  Josef Kittler, UOS        Chris McGlone, CMU           Peter Meer, Rutgers
  Keith Price, USC          George Vosselman, Delft    

The workshop is motivated by the strong belief that the lack of performance 
characterization of vision algorithms is responsible for the hesitation of 
industry to use Computer Vision as one of its tools.  Reasons for this 
situation are manifold: the lack of commonly accepted criteria for 
evaluation, the lack of a methodology for testing, the lack of translating 
the experience in testing of other engineering areas to Computer Vision and 
possibly also the non-acceptance of empirical or theoretical comparisons of 
vision algorithms, including their replication, as original research.

The goal of the workshop is to investigate the state of the art in 
characterizing the performance of vision algorithms in all stages of the 
development and use, such as the design of algorithms with a pre-specified 
performance, the testing of algorithms with respect to give specifications 
and the self-diagnosis of vision algorithms during their use in an 
automated process.

We encourage authors to submit papers on the following topics:

*    Theory and strategies for performance analysis of vision algorithms
*    Linking analysis of vision experiments to the theory underlying the 
     algorithms
*    Characterization of the limitations of vision algorithms and/or
     the class of image data for which a vision algorithm is suited or
     not suited
*    Demonstrations of the usefulness of performance characterization
     and/or the limitations of statistical testing in Computer Vision
*    Modularization of vision tasks and the characterization of networks of
     vision algorithms 

We will not consider plain comparisons of algorithms without explicit 
relation to the underlying theory, plain statistical papers without 
explicit relation to performance characterization or papers on the analysis 
of the complexity of algorithms without containing a clear cost benefit 
analysis.

We intend to select approx.  4 papers for long (45 min.) and approx.  8 
papers for short (30 min.) oral presentations forming the basis for a 
common discussion.  An additional 12 - 16 papers will be selected for 
poster presentation possibly including software demos.

Timetable:
-----------
15. September 1995    Call for papers
 1. December 1995     Full papers 
15. January 1996      Reviews 
 1. February 1996     Notification of authors 
15. March 1996        Camera ready manuscripts are due
 1. April 1996        Distribution of papers to participants (ftp) 
19. April 1996        Workshop  (1.5 days)
 31. May 1996         Distribution of Proceedings to participants

Submit Papers before 1. December 1995 to:

Dr. Henrik I. Christensen
Laboratory of Image Analysis
Aalborg University
Fr. Bajers Vej 7, Bldg. D1
DK-9220 Aalborg OE, Denmark
E-mail: hic@vision.auc.dk