PAMI TC Meeting
Centre de Congres Acropolis, Nice, France, 14 October 2003
(At ICCV 2003)

Minutes recorded by Daniela Hall, edited by James L. Crowley

Agenda
1. The PAMI TC
2. Minutes from PAMI TC meeting at CVPR 2003
3. Report on ICCV 2003 (B. Triggs)
4. Plans for ICCV 2005 (H. Shum)
5. Proposals for ICCV 2007
6. Other upcoming Conferences
7. Any other business

Participants:
The meeting was attended by about 100 ICCV participants.

1. The PAMI TC
As the meeting came to order, James Crowley presented a brief  review
of the structure and role of the PAMI TC.

2. Minutes from PAMI TC meeting at CVPR 2003
The highlights of the minutes from the previous meeting at CVPR 2003
were presented. Special mention was given of the election of Kim
Boyer as future PAMI TC Chair, replacing Chuck Dyer in January 2004.

3. Report on ICCV 2003 (B. Triggs):

Bill Triggs reported that the conference received about 1100
abstracts resulting in 960 paper submissions. From these 43 oral
presentations and  156 posters were selected for the conference. This
gives an acceptance rate of about 20.7%.  The conference has
recorded 690 paid registrations as of the time of the meeting. In
response to a question, Bill reported that the acceptance rate at
ICCV has increased slightly at each conference.

In response to a question on the budget, Bill reported that the
recent exchange rate variations between the Euro and the Dollar will
decrease the profit margin of the conference. To a further question,
Bill explained that the IEEE charges 14% of the conference income
for treatment of registrations and other services. After expenses for
catering, and other costs, the profit is about 5% of the conference
budget. This profit may be reduced by the recent drop in the value of
the dollar because registration is paid in dollars while expenses for
the conference are in Euros.

Jatendra Malik commented that B. Triggs agreed on 24 hours notice to
switch the venue to Nice, and asked for a round of applause. He also
remarked that H. Shum deserves support for the next ICCV.

4. Plans for ICCV 2005 (H. Shum)

Harry made a brief review of plans for ICCV 2005.  ICCV 2005 will be
in Beijing and may be held in May or October 2005.

5. Proposals for ICCV 2007

A proposal was made by Larry Davis for ICCV 07 to be held in
Rio de Janeiro, Brazil
General Chairs: L. Davis, K. Ikeuchi, P. Bouthemy
Local Chair: S. Goldenstein, L. Velho, D. Geiger
Program Co-Chairs: Metaxas, Vemuri, Shashua, Shum
Workshops Chair: T. Darrell
Short Courses Chair: D. Samaras


It was noted that ICCV has never been in South America. Brazil has
active research community in Computer Vision and in Image  Processing.

There are several possible sites for the venue. A likely choice is
the Hotel Gloria with  600 rooms and a meeting room that can seat
1000. A member of the audience asked if this will be enough seating?
The risk is small that ICCV increases significantly in size. It was
pointed out that the current meeting room seats 750 and that this was
more than adequate for the conference this year.

A discussion was held about the dates. The Conference could be held
between  late August and mid October. In response to a question, a
discussion was held about dates. A show of hands was held, and
roughly  75% of the audience indicated a preference for a preference
for end of august or early September because of conflicts with the
academic year in North America and Europe.
It was remarked that a conference in october would allow better
separation with CVPR. The issue is to be re-discussed  at CVPR in
Washington.

A vote was held, and the TC unanimously accepted the proposal for
ICCV 07 to be held in Rio de Janeiro.

A discussion was then held on the review process. James Crowley
observed that several innovations had been instituted this year at
CVPR.  In particular, authors receive a one-week period to comment
and refute reviews. The program committee then has an opportunity to
revise reviews. Several participants remarked that authors preferred
this process. It was generally found to be much fairer. In
particular,  this procedure offers a solution to the perceived
problem that the most innovative papers generally receive mixed
reviews. With selection rates at around 20%, the result is that only
work on established problems is would receive unanimous approval.
Other members observed that this procedure increases the work load of
reviewers and area chairs, both because they must devote extra time
to the review process, and also that they must be more thorough in
their reviews.

Sample Comments:
(B. Freeman): This variation of the review process may be useful but
increases the work load.
(R. Szeliski): Area chair needs to make sure that reviewers do their work.
(H. Shum): Agrees with Szeliski. We must be careful about how many
reliable reviews can be produced by the community (increasing number
of conferences, increasing number of submissions). Considering the
working time for reviews, the review process is expensive.

B. Triggs made a novel proposal that authors be required to pay for
submission. With such a process, a fund would be available to reward
reviewers.

Although a consensus seemed to emerge in favor of this new procedure,
the TC decided  not to make a formal motion on these issues. It is
recommended that this issue be discussed at the PAMI-TC meeting.

6. Other upcoming Conferences
Overview of upcoming conferences:
CVPR04: Date: June 2004, Paper deadline: 19 November 2003

(Huang): This date has a conflict with ICME. (International
Conference on Multi-Media and Expo).
It was observed that , while ICME is an IEEE conference,  it is not
sponsored by PAMI and draws only a small number of PAMI members.  It
is difficult to avoid such overlaps.

ECCV04: Date: May 2004, Venue: Prague, submissions closed, 9 associated workshops.

ICVS05: Organized by R. Kjeldsen, date March/April 2005 in New York area.

CVPR05: Date: Spring 2005 in San Diego. Paper deadline: November 2004.

(Malik): states that for a good review process the program committee must be large.

The meeting closed with no other issues.