[4eyes] Quick CVPR debrief

[Matthew Turk]

Very quick, high-level overview of CVPR last week:

 

The overall message from the conference is that computer vision is hot. We
hit a new attendance record with over 2000 attendees (in Columbus, Ohio!),
and also a new record with $181,000 in corporate sponsorship, which is much,
much higher than 2-3 years ago (and 50% higher than last year). There were
30 exhibitors (Google, Qualcomm, Xerox, Amazon, NVIDIA, Metaio, and lots of
small companies) and 26 technical demonstrations. There were two excellent
keynote talks (CVPR usually doesn’t have keynotes) – by Doris Tsao, a
neurophysiologist from Cal Tech, and Stephane Mallat, a mathematician from
École Normale Supérieure, Paris (and famous for his work on wavelets). This
year’s conference was four days for the main conference (a departure from
the normal three days) and two days for 19 workshops and 15 tutorials.

 

It was a busy week.

 

It is really amazing, for those of us who remember the “old days” in the
field, how much industry interest there is now in CV. Everyone is trying to
hire vision people – especially Amazon, but also Apple, Facebook, etc., and
many small companies.

 

Anything that had “deep learning” in its title (including a tutorial and a
workshop) drew a big audience. Many are not convinced that it’s the solution
to everything; time will tell whether it’s a fad or something substantial.

 

Johnny Lee from Google gave an interesting talk at an embedded vision
workshop. I already sent out a video he showed, but here it is again:
https://www.youtube.com/watch?v=44vppay5UDc. To me, this seems like the most
important vision-related project around.

 

Amazon is still not talking about their secret “shiny new project” involving
computer vision, but Gerard Medioni from USC is now there on sabbatical (and
maybe for good), and they are being very aggressive about hiring. I thought
that maybe the new Fire phone was the result of the secret project, but
supposedly it’s something much more interesting.

 

Next year’s CVPR is in Boston - http://www.pamitc.org/cvpr15/. 

 

Papers from the main conference and workshops are available at
http://www.cv-foundation.org/openaccess. There are some video spotlights
(1-minute video overviews) of papers at
http://techtalks.tv/cvpr-2014-viddeo-highlights/. Videos of the oral
presentations will be posted, but I don’t think they’re available yet.

 

PAMI TC meeting voting results:
http://www.pamitc.org/cvpr14/files/cvpr2014-vote-results.pdf

 

Awards:

Best Paper Award: "What Camera Motion Reveals About Shape with Unknown BRDF"

Manmohan Chandraker

 

Best Paper Honorable Mention: "3D Shape and Indirect Appearance by
Structured Light Transport"

Matthew O'Toole, John Mather, Kyros Kutulakos

 

Best Student Paper Award: "Partial Optimality by Pruning for MAP-inference
with General Graphical Models"

Paul Swoboda, Bogdan Savchynskyy, Joerg Kappes, Christoph Schnörr

 

2014 Longuet-Higgins Prize: "A performance evaluation of local descriptors"

Krystian Mikolajczyk and Cordelia Schmid

 

2014 PAMI Young Researcher Awards: Derek Hoiem and Jamie Shotton

 

Pictures: http://www.pamitc.org/cvpr14/awards.php

 

Best Demo Award: “Real-Time Video Magnification”

Neal Wadhwa, Michael Rubinstein, Frédo Durand, William T. Freeman

 

Best Demo Honorable Mention: “Learning to be a Depth Camera”

Sean Ryan Fanello, Cem Keskin, Shahram Izadi, Pushmeet Kohli, David Kim,
David Sweeney, Antonio Criminisi, Jamie Shotton, Sing Bing Kang, Tim Paek 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cs.ucsb.edu/pipermail/ilab-users/attachments/20140702/b2aad579/attachment.html>