Activities

Report on the Second International Workshop on Parts and Attributes, held in conjunction with ECCV 2012 and benefiting from DAGM support.

Christoph Lampert, Rogerio Feris (co-organizers)

The workshop took place in Florence, Italy, on Saturday, October 13th, 2012, the second day of workshops at ECCV 2012. It consists of four sessions, each being opened by an invited talk from one of our four distinguished invited speakers:

● Irving Biederman, USC
● David Forsyth, UIUC
● Kristen Grauman, U Texas
● Deva Ramanan, UC Irvine

During each of the first three sessions, the invited talk was followed by two contributed talks ones, which were selected by the organizers based on double-blind reviews performed by the program committee. In the last workshop session, the invited talk was followed by a panel discussion about the role of parts, attributes and global representations. The panel consisted of the following renowned experts:

● Alyosha Efros, CMU
● Vittorio Ferrari, U Edinburgh
● Yann LeCun, NYU
● Jitendra Malik, UC Berkeley

It was moderated by David Forsyth.
Overall, the workshop was a tremendous success. On average there were more than 50 people present in the audience, despite the fact that the workshop took place on the last day of the conference and many conference attendees had already left Florence. In fact, we were told by several participants that they intentionally stayed longer in order to be able to visit the workshop. At some times the number grew beyond 100, such that the room assigned by the ECCV organizers did not provide enough seats. From the discussions we had after the workshop our impression was that the audience was very pleased with the selection of speakers and found the panel discussion informative as well as inspiring. We as organizers were also pleased with the quality of contributed talks, all of which contained innovative and relevant work and were presented in a professional way. We would like to thank the DAGM again for its generous sponsoring, which allowed us to invite Irving Biederman as keynote speaker, who otherwise would not have attended ECCV 2012. We acknowledged the sponsoring during the official workshop opening, and it is also highlighted on the workshop’s website,http://ist.ac.at/~chl/PnA2012/, where the full workshop program as well as pictures taken during the event can be found.

Report on the First International Workshop on Parts and Attributes, held in conjunction with ECCV 2010 and benefiting from DAGM support.

Christoph Lampert, Rogerio Feris, Tiberio Caetano and David Forsyth (co-chairs)

The workshop took place on Friday, September 10th, 2010, the first workshop day at ECCV 2010. It consisted of four invited talk by very distinguished speakers:

● Derek Hoiem, University of Illinois at Urbana-Champaign
● Jitendra Malik, University of California, Berkeley
● Pietro Perona, California Institute of Technology
● Bernt Schiele, Technical University Darmstadt / Max-Planck Institute for Informatics, Saarbrücken

as well as six contributed talks that were selected by the organizers based on double-blind reviews performed by the program committee. The last workshop session consisted of a panel discussion with the following renowned experts as participants:

● Derek Hoiem, University of Illinois at Urbana-Champaign (replacing the deceased Vittorio Ferrari)
● Fei-Fei Li, Stanford University
● Jitendra Malik, University of California, Berkeley
● Jiri Matas, Czech Technical University, Prague

The panel was moderated by Christoph Lampert. Immediately after introducing the topic the panel began a lively discussion about the role and relevance of parts and attributes in Computer Vision. Several comments from the audience were discussed and questions -as far as possible-answered. The workshop closed at 6 p.m., but it was obvious that even in leaving, many participants were involved in intense scientific discussions themselves. Overall, the workshop was a tremendous success. It had consistently more than 50 participants. At some times the number grew beyond 100, such that the room assigned by the ECCV organizers did not provide enough seats. We heard from several participants that the invited talks as well as the panel discussion impressed and inspired them to look deeper into the topic of part-based and attribute-based models for object recognition themselves. In addition, we as organizers were very pleased with the quality of contributed talks, all of which contained innovative and relevant work and were presented in a professional way. The full workshop program as well as pictures taken during the event can be found at http://rogerioferis.com/PartsAndAttributes/.

Report on the Third International Shape Workshop on Shape Perception in Human and Computer Vision, held in conjunction with ECCV and benefiting from DAGM support.

Sven Dickinson and Zygmunt Pizlo (co-chairs)

This was the third in a series of workshops that alternate between premiere computer vision and human vision conferences.  The format of the workshop was, as has always been the case, 12 very distinguished invited speakers, 6 from human vision and 6 from computer vision, to reflect on their experience in shape perception, outline the problems and challenges that face the community, and attempt to bring the two communities closer together.

The workshop was a tremendous success. Prior to the workshop, we were told by the ECCV organizers that our workshop had the second largest subscription (second to the VOC challenge workshop) of all the ECCV workshops.  This was reflected in the attendance during the workshop, which was strong, and never dipped below 50 persons throughout the day. The talks were given by: (see http://viper.psych.purdue.edu/workshops/iwsphcv2010/)

James Elder, York University
Doug DeCarlo, Rutgers University
Donald D. Hoffman, University of California, Irvine
Lena Gorelick, University of Western Ontario
Phil Kellman, University of California, Los Angeles
Ian Jermyn, Ariana group, INRIA
Zoe Kourtzi, University of Birmingham
Bernt Schiele, MPI Informatics, Saarbrücken
Bosco S. Tjan, University of Southern California
Anuj Srivastava, Florida State University
Christopher Tyler, Smith-Kettlewell Institute
Chris Taylor, University of Manchester

and covered a wide array of topics (abstracts listed on website).

The workshop culminated in a rousing, well-attended panel discussion that continued to 6:30 pm.  Issues such as 3-D vs 2-D shape models, bottom-up vs. top-down shape recovery, the role of perceptual grouping, evaluation methods for shape recovery research, future challenges, etc., engaged both audience and panel members.

Upon reflection, the organizers were extremely pleased, especially with the interest shown to the perspectives offered by the human vision speakers -- something students in computer vision are rarely exposed to. There's a wealth of research experience, and data coming from the human vision community that should not be ignored by the computer vision community.  In turn, the human vision speakers were very impressed with the progress made by our community.  We believe that the workshops served to help bring the two communities closer together.  We will continue to do this at the next installment, to be held at a human vision conference.