CVPR 2009 Program IEEE Computer Society Conference on Pattern Recognition and Pattern Recognition Mon, 22 Feb 2010 15:12:41 +0000 en Longuet-Higgins Prize for Fundamental Contributions in Computer Vision Sat, 20 Jun 2009 11:38:39 +0000 pc-chairs Longuet-Higgins Prize for Fundamental Contributions in Computer Vision is will be awarded at IEEE CVPR for papers that have “withstood the test of time.” Named after H. Christopher Longuet-Higgins, this award aims to recognize papers from CVPR meeting of 10 years back that have had a fundamental impact on the field of Computer Vision. More specifically, “The contribution of Longuet-Higgins to computer vision and pattern recognition was a series of sudden and pinpoint advances resulting from extended scholarship rather than an accretion of archived output.  Thus the Longuet-Higgins prize rewards an individual advance, represented by a single conference paper—underlining the special role of conference publications as the harbingers of innovation in this discipline.” This award was established in 2005.

The award papers are selected by a small committee, drawn from across the field by the conference chairs of a specific conference, and as far as possible excluding those who were authors of papers in the source conference.

Following is the list of winner from CVPR 2005, when this award was first instituted.

Awarded in CVPR 2005 (from CVPR 1995)

  • David Mumford and Jayant Shah (1995), “Boundary detection by minimizing functionals,”  CVPR 1985, pages 22-26.
  • Ted Adelson and John Wang (1995), “Layered representation for motion analysis,” CVPR 1993, pages 361-366.

Awarded in CVPR 2006 (from CVPR 1996)

  • H. Rowley, S. Baluja and T. Kanade (1996), ”Neural Network-Based Face Detection” CVPR 1996
  • Cordelia Schmid and Roger Mohr (1996), “Combining greyvalue invariants with local constraints for object recognition”, CVPR 1996.

Awarded in CVPR 2007 (from CVPR 1997)

  • J. Shi and J. Malik (1997), “Normalized Cuts and Image Segmentation.” CVPR 1997.
  • E. Osuna, R. Freund, and F. Girosi (1997), “Training Support Vector Machines: An Application to Face Detection.” CVPR 1997.

Awarded in CVPR 2008 (from CVPR 1998)

  • H Schneiderman and T Kanade (1998) “Probabilistic modeling of local appearance and spatial relationships for object recognition.” CVPR 1998.
  • C Bregler and J Malik (1998) “Tracking people with twists and exponential maps.” CVPR 1998

Awarded in CVPR 2009 (from CVPR 1999)

  • Jinggang Huang, David Mumford, “Statistics of Natural Images and Models,” cvpr, vol. 1, pp.1541, 1999 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’99) - Volume 1, 1999
  • Chris Stauffer, W.E.L. Grimson, “Adaptive Background Mixture Models for Real-Time Tracking,” cvpr, vol. 2, pp.2246, 1999 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’99) - Volume 2, 1999
Doctoral Spotlight Session 4 Wed, 17 Jun 2009 16:56:11 +0000 cvpr Time: Wednesday, 7/24/09, 10:30-11:00 am
Location: Sparkle East
Session Chair: Bernt Schiele

Title Poster Slot Authors
Noninvasive Volumetric
Imaging of Cardiac Electrophysiology
Wed Linwei Wang, Heye
Zhang, Ken C. L. Wong, Huafeng Liu, and Pengcheng
Discriminative 3D Subvolume
Search for Efficient Action Detection
Wed Junsong Yuan, Zicheng
Liu, and Ying Wu
Interval HSV: Extracting
Ink annotations
Wed John C. Femiani and Anshuman
Fourier Analysis and Gabor
Filtering for Texture Analysis and Local Reconstruction of General Shapes
Wed Fabio Galasso and Joan
Unsupervised Learning
of Hierarchical Spatial Structures in Images
Wed Devi Parikh, C. L. Zitnick,
and Tsuhan Chen
A Robust Approach for
Automatic Registration of Aerial Images with Untextured Aerial LiDAR
Wed Lu Wang and Ulrich Neumann
From Contours to Regions:
An empirical evaluation
Wed Pablo Arbelaez, Michael
Maire, Charless Fowlkes, and Jitendra Malik
Locally Time-Invariant
models of Human Activities using Trajectories on the Grassmannian
Wed Pavan Turaga, Rama Chellappa
Multi-Class Active Learning
for Image Classification
Wed Ajay J. Joshi, Fatih Porikli,
and Nikolaos Papanikolopoulos
Marked Point Processes
for Crowd Counting
Wed Weina Ge and Robert Collins
"A Novel Feature
Descriptor Invariant to Complex Brightness Changes
Wed Feng Tang, Suk-Hwan Lim,
Nelson L. Chang, and Hai Tao
Tubular Anisotropy for
2D Vessel Segmentation
Wed F. Benmansour, L. D. Cohen,
M. W. K. Law, and A. C. S. Chung
Doctoral Spotlight Session 3 Wed, 17 Jun 2009 16:49:32 +0000 cvpr Time: Tuesday, 7/23/09, 10:30-11:00 am
Location: Sparkle East
Session Chair: Stan Sclaroff

Title Poster Slot Authors
Observable Subspaces for
3D Human Motion Recovery
Tues Andrea Fossati, Mathieu
Salzmann, and Pascal Fua
Nonnegative Matrix Factorization
with Earth Mover's Distance Metric
Tues Roman Sandler and Michael
Motion Capture Using Joint
Skeleton Tracking and Surface Estimation
Tues Juergen Gall, Carsten
Stoll, Edilson de Aguiar, Christian Theobalt, Bodo Rosenhahn, Hans-Peter
Distributed Multi-Target
Tracking In A Self-Configuring Camera Network
Tues Cristian Soto, Bi Song,
and Amit Roy-Chowdhury
Learning to describe objects Tues Ali Farhadi, Ian Endres,
Derek Hoiem, and David Forsyth
Building text features
for object image classification
Tues Gang Wang, Derek Hoiem,
and David Forsyth
The Geometry of 2D Image
Tues Lennart Wietzke, Gerald
Sommer, and Oliver Fleischmann
A Perceptually Motivated
Online Benchmark for Image Matting
Tues Christoph Rhemann, Carsten
Rother, Jue Wang, Margrit Gelautz, Pushmeet Kohli, and Pamela Rott
Towards high-resolution
large-scale multi-view stereo
Tues Vu Hoang
Hiep, Renaud Keriven, Patrick Labatut, and Jean-Philippe Pons
Continuous Ratio Optimization
via Convex Relaxation with Applications to Multiview 3d Reconstruction
Tues Kalin Kolev and Daniel
A Similarity Measure Between
Vector Sequences with Application to Handwritten Word Image Retrieval
Tues Jose A. Rodriguez-Serrano,
Florent Perronnin, Josep Llados, and Gemma Sanchez
Cooperative Mapping of
Multiple PTZ Cameras in Automated Surveillance Systems
Tues Chung-Hao
Chen, Yi Yao, Anis Drira, Andreas Koschan, and Mongi Abidi
Predicting High Resolution
Image Edges with a Generic, Adaptive, 3-D Vehicle Model
Tues Matthew J. Leotta and
Joseph L. Mundy
Face Verification and
Identification using Facial Trait Code
Tues Ping-Han Lee, Gee-Sern
Hsu, and Yi-Ping Hung
P-Brush: Continuous Valed
MRFs with Normed Pairwise Distributions for Image Segmentation
Tues Dheeraj Singaraju, Leo
Grady, and René Vidal
Doctoral Spotlight Session 2 Wed, 17 Jun 2009 16:37:36 +0000 cvpr Time: Monday, 7/22/09, 3:10 pm-3:40 pm
Location: Sparkle East
Session Chair: Pedro Felzenszwalb

Title Poster Slot Authors
Pose Estimation for Category
Specific Multiview Object Localization
Mon pm Mustafa Ozuysal, Vincent
Lepetit, and Pascal Fua
Global Connectivity Potentials
for Random Field Models
Mon pm Sebastian Nowozin and
Christoph Lampert
Max-Margin Hidden Conditional
Random Fields for Human Action Recognition
Mon pm Yang Wang and Greg Mori
Unsupervised Learning
for Graph Matching
Mon pm Marius Leordeanu and Martial
Recognizing Linked Events:
Searching the Space of Feasible Explanations
Mon pm Dima Damen and David Hogg
Learning from Ambiguously
Labeled Images
Mon pm Timothee Cour, Benjamin
Sapp, Chris Jordan, Ben Taskar
Fuzzy-Cuts: A Knowledge-Driven
Graph-Based Method for Medical Image Segmentation
Mon pm Deepak Roy
Chittajallu, Gerd Brunner, Uday Kurkure, Raja Yalamanchili, Ioannis
Active Learning for Large
Multi-class Problems
Mon pm Prateek Jain and Ashish
Shape Constrained Figure-Ground
Segmentation and Tracking
Mon pm Zhaozheng Yin and Robert
New Appearance Models
for Natural Image Matting
Mon pm Dheeraj Singaraju, Carsten
Rother and Christoph Rhemann
Multi-Cue Onboard Pedestrian
Mon pm Christian Wojek, Stefan
Walk, and Bernt Schiele
Doctoral Spotlight Session 1 Wed, 17 Jun 2009 00:06:29 +0000 cvpr Time: Monday, 7/22/09, 10:30-11:00 am
Location: Sparkle East
Session Chair: Kristen Grauman

Title Poster Slot Authors
Hardware-Efficient Belief
Mon am Chia-Kai
Liang, Chao-Chung Cheng, Yen-Chieh Lai, Liang-Gee Chen, Homer H. Chen

Support Vector Machines
in Face Recognition with Occlusions
Mon am Hongjun Jia & Aleix
Early Spatiotemporal Grouping
with a Distributed Oriented Energy Representation
Mon am Konstantinos G. Derpanis
and Richard P. Wildes
Understanding Images of
Groups of People
Mon am Andrew Gallagher and Tsuhan
Volterrafaces: Discriminant
Analysis using Volterra Kernels
Mon am Ritwik Kumar, Arunava
Banerjee, Baba C. Vemuri
Efficient Planar Graph
Cuts with Applications in Computer vision
Mon am Frank R. Schmidt, Eno
Toeppe and Daniel Cremers
Digital Face Makeup by
Mon am Dong Guo and Terence Sim
Pedestrian Detection:
A Benchmark
Mon am Piotr Dollár, Christian
Wojek, Bernt Schiele, Pietro Perona
Twitter Updates Sun, 14 Jun 2009 01:58:14 +0000 pc-chairs

Doctoral Spotlight Committee Wed, 10 Jun 2009 01:11:29 +0000 cvpr Christoph Lampert
Margrit Gelautz
Trevor Darrell
Horst Bischof
Michael S. Brown
Larry Davis
Fei-Fei Li
Jean Ponce
Marshall Tappen
Frank Dellaert
Ramesh Raskar
Rob Fergus
Pedro Felzenszwalb
Bill Freeman
Rene Vidal
Baba Vemuri
Amit Roy-Chowdhury
Stefano Soatto
Yann LeCun
Jianbo Shi
David Hogg
Mark Everingham
Pascal Fua
Edmond Boyer
Daniel Cremers
Alyosha Efros
Cristian Sminchisescu
Matthew Turk
Ben Taskar
Serge Belongie
Mubarak Shah
Dorin Comaniciu
Shih-Fu Chang
Todd Zickler
Greg Mori
Jitendra Malik
Tsuhan Chen
Anthony Hoogs
Robert Collins

Doctoral Spotlights Tue, 09 Jun 2009 23:47:20 +0000 cvpr The Doctoral Spotlight will consist of short talks by the student authors, to be given during the main program. The goal is to give more exposure to student authors who are near graduation, and to specifically highlight their work to the rest of the community, including potential employers.

Update: The winners of the Doctoral Spotlight Awards are:

Efficient Planar Graph Cuts with Applications in Computer
Vision, by Frank R. Schmidt, Eno Toppe and Daniel Cremers

Shape Constrained Figure-Ground Segmentation and Tracking,
by Zhaozheng Yin and Robert Collins

Towards High-Resolution Large-Scale Multi-View Stereo, by
Vu Hoang Hiep, Renaud Keriven, Patrick Labatut, and Jean-Philippe Pons

Discriminative Subvolume Search for Efficient Action Detection, by
Junsong Yuan, Zicheng Liu, and Ying Wu

Each talk is 2 minutes long with 5 seconds between each talk. The slides will be on auto-play. Presentations are split among the following 4 sessions:

Session 1

Time: Monday, 7/22/09, 10:30-11:00 am
Location: Sparkle East
Session Chair: Kristen Grauman

Session 2

Time: Monday, 7/22/09, 3:10 pm-3:40 pm
Location: Sparkle East
Session Chair: Pedro Felzenszwalb

Session 3

Time: Tuesday, 7/23/09, 10:30-11:00 am
Location: Sparkle East
Session Chair: Stan Sclaroff

Session 4

Time: Wednesday, 7/24/09, 10:30-11:00 am
Location: Sparkle East
Session Chair: Bernt Schiele

POSTERS Fri, 05 Jun 2009 03:00:44 +0000 pc-chairs Below are the instructions for poster presenters. Please read carefully.

Those presenting posters in the main sessions of the conference will hang their posters on tack boards. EACH POSTER HAS BEEN ASSIGNED TO A SPECIFIC BOARD (i.e., it is not “first come first served” as in previous years). Authors can find their tack board location via labels with a poster ID number. The poster ID number can be found in the pocket guide. Morning poster IDs will contain a green “AM” after the poster number. Afternoon session poster IDs will have a red “PM” after the poster number. Note that each poster will be displayed for the ENTIRE DAY. Presenters can put up their posters any time prior to their session. Presenters are responsible for REMOVING THEIR POSTERS at the end of the day of their session, so that the boards are available for use by others.

Push pins will be available on each tack board. The usable space on these tack boards is approximately 43.5 inches high x 91.25 inches wide. A FedEx printing center in the hotel lobby can print posters in one day.

ORALS Fri, 05 Jun 2009 02:53:53 +0000 pc-chairs


Congratulations on having your paper’s selection for an ORAL presentation at CVPR 2009 in Miami, FL. As you well know, getting an ORAL presentation is a special privilege as you will be presenting to a room full of anywhere from 500-1000 people. It is essential that you take this presentation seriously and prepare your presentation to highlight the novel contribution of your research to the fullest. There are two primary aspects of this ORAL presentation. (1) Appropriate presentation materials to showcase your work and (2) Preparation, and appropriate (and repeated) practice of your presentation , to get you ready for a professional presentation of you research to a large crowd of your peers.  Here we provide some guidelines to help you do both of these.  Please pay attentions to these guidelines and recommendations.

1. Preparing Slides.

A key element of your CVPR 2009 presentation is your electronic slide preparation. The audience will base its evaluation of you and your subject matter partly on the appearance of your images. An attractive, legible, and organized presentation will reflect positively on the content, and therefore on you. Please review these basic guidelines to ensure that your CVPR 2009 presentation is the best it can possibly be:

A. Timing

Each ORAL presentation at CVPR is allocated 20 minutes. 15-16 minutes for the presentation and the rest for question / answers from the audience moderated by the session chair. Plan your talk and the number of slides to allow for a relaxed pace. A usual recommendation is of a slide per minute. If you have a lot more than that (say 30!), you will be rushing it for sure. Practice your talk before the conference. Time yourself. Think of what you would change if the presentation had to be shortened or lengthened. Force yourself to slow down a little. A rushed presentation will create more stress for you and won’t be compelling. Practice, practice, practice.  You will notice that after a few practices you will be better paced and more comfortable with the presentation.

B. Legibility (Text/Fonts/Style)

The ORAL sessions rooms at CVPR 2009 are large and can accommodate up-to 900+ people in conference/classroom style seating. Your presentation must be legible from the back row, with decent lighting. If you can stand two meters away from your computer’s monitor and easily read your slides, your text is large enough. To achieve this, limit each slide to eight lines of text or less and limit each line of text to 30 characters or less. Use a bold typeface, no smaller than 28 points, with generous line spacing. Also, remember that slides with material (text/figures) all the way to the bottom of the slide are hard to read from the back rows. Leave the bottom inch of the slide content free.

Use key words, so that your slides will be quick and easy to read. You want the audience to hear your presentation, while the slides accentuate the points to remember. Use standard fonts. That way your presentation will be truly portable. Incorporate only the essential parts of a diagram and simplify whenever possible. While it is tempting to include detail for the sake of accuracy, too much will make the slide difficult to read and become a distraction. Break up complex diagrams into sections if you can, so that each section can be made larger and therefore more legible. If you use plots, make sure axis are marked and also legible.

C. Capitalization

Avoid the use of ALL CAPITAL letters. Words written in ALL CAPS are harder to read and take up more space on the screen. Use bold face and italics for emphasis, or use a bright color such as yellow text when normal body text is white. Underlined text is not recommended as it is hard to see.

D. Color and Contrast

Make good use of color and contrast. Dark backgrounds tend to be easier to view, especially with light text and graphics. These days with brighter projectors, light background slides work fine too, If you use a light background, use black or very dark text and graphics. Do not use yellow of lighter colors for text.

Maintain consistency throughout your slides. Using the same background color, text size, text color, and uniform fonts throughout all the slides makes it easier for the audience to follow the flow of your ideas.

E. Templates

Following are templates in Powerpoint (ppt, pptx, pot, potx) and in Keynote (key, kth).  Feel free to use and modify.

Guidelines about Speaking/Presenting

Being able to stand in front of an audience an convey your message, and in this case, your research methodology, results and findings is no trivial matter. We expect all of you to take this task seriously and come prepared for a high-quality oral presentation, representative of your research, that resulted in an ORAL presentation at CVPR. Following are a few sites we have gathered to help you understand, appreciate, and more importantly learn from, to aid you in giving the best ORAL presentation possible.

Again, above are simply recommendation and guidelines. We have no doubt that each and every presenter at CVPR will do their best to give an excellent presentation. Hopefully, the above guidelines will help you with that.  At the risk of repeating ourselves, we encourage all presenters to prepare good materials, with great visual quality and prepare/practice your talk in front of a live audience.
Chairs, CVPR 2009.