The PASCAL Visual Object Classes Challenge Workshop 2009

3rd October 2009, ICCV 2009, Kyoto, Japan

Introduction

This workshop will present and discuss the results of the PASCAL Visual Object Classes Challenge (VOC2009). The challenge has three main competitions, one testing image classification ("does the image contain an instance of this class?"), one testing object detection ("provide a bounding box for each instance of the class, if any"), and one evaluating object segmentation at the pixel level. In addition there is a single 'taster' competition, evaluating the object layout in more detail ("detect the hands, feet etc for a person").

A new database has been prepared consisting of 20 classes with around 35,000 annotated instances in total. The images are obtained from flickr. The classes include people, cats, dogs, cars, motorbikes, bottles and sofas. The annotation includes a rectangular bounding box and flags to indicate pose and level of difficulty.

Full details of the challenge.

Venue

The workshop was held on 3rd October 2009, in association with ICCV 2009.

Prizes

The following prizes were announced at the workshop:

Classification
Winner: NEC/UIUC
Yihong Gong, Fengjun Lv, Jinjun Wang, Chen Wu, Wei Xu, Jianchao Yang, Kai Yu, Xi Zhou, Thomas Huang
NEC Laboratories America; University of Illinois at Urbana-Champaign
Honourable mentions: UVA/SURREY
Koen van de Sande, Fei Yan, Atif Tahir, Jasper Uijlings, Mark Barnard, Hongping Cai, Theo Gevers, Arnold Smeulders, Krystian Mikolajczyk, Josef Kittler
University of Amsterdam; University of Surrey

CVC
Fahad Shahbaz Khan, Joost van de Weijer, Andrew Bagdanov, Noha Elfiky, David Rojas, Marco Pedersoli, Xavier Boix, Pep Gonfaus, Hany salahEldeen, Robert Benavente, Jordi Gonzalez, Maria Vanrell
Computer Vision Centre Barcelona
Detection
Joint Winners: UoC/TTI Chicago
Pedro Felzenszwalb, Ross Girshick, David McAllester
University of Chicago; Toyota Technological Institute at Chicago

Oxford/MSR India
Andrea Vedaldi, Varun Gulshan, Manik Varma, Andrew Zisserman
University of Oxford; Microsoft Research India
Segmentation
Winner: Bonn
Joao Carreira, Fuxin Li, Cristian Sminchisescu
University of Bonn
Runner-up: CVC
Xavier Boix, Josep Maria Gonfaus, Fahad Kahn, Joost van de Weijer, Andrew Bagdanov, Marco Pedersoli, Jordi Gonzalez, Joan Serrat
Computer Vision Center Barcelona

Programme

09:30 - 09:50 Overview and results of the detection challenge
Mark Everingham
University of Leeds
[PDF]

09:50 - 11:00 Detection and segmentation methods

Multiple kernels for object detection (OXFORD_MKL)
Andrea Vedaldi1, Varun Gulshan1, Manik Varma2, Andrew Zisserman1
1University of Oxford; 2Microsoft Research India
[PDF]

Using poselets for detection and segmentation (BERKELEY_POSELETS)
Lubomir Bourdev1,2, Subhransu Maji1, Jitendra Malik1
1UC Berkeley; 2Adobe Systems, Inc.
[Powerpoint]

Discriminatively trained part based models (UoCTTI_LSVM-MDPM)
Pedro Felzenszwalb1, Ross Girshick1, David McAllester2
1University of Chicago; 2Toyota Technological Institute at Chicago
- to be presented by Deva Ramanan
[PDF]

Part-based layered shape models for segmentation (UCI_LAYEREDSHAPE)
Charless Fowlkes, Sam Hallman, Deva Ramanan, Yi Yang
UC Irvine
[PDF] NB: This is unpublished work. Please contact the authors if you plan to make use of any of the ideas presented.

11:00 - 11:30 Break
11:30 - 11:50 Overview and results of the classification challenge
Mark Everingham
University of Leeds
[PDF]

11:50 - 12:10 Classification methods

Image classification using Gaussian mixture and local coordinate coding (NECUIUC_CLS-DTCT)
Yihong Gong1, Thomas Huang2, Fengjun Lv1, Jingjun Wang1, Chen Wu4, Wei Xu1, Jianchao Yang2, Kai Yu1, Tong Zhang3, Xi Zhou2
1NEC Laboratories America; 2University of Illinois at Urbana-Champaign; 3Rutgers University; 4Stanford University
[PDF] NB: This is unpublished work. Please contact the authors if you plan to make use of any of the ideas presented.

12:10 - 13:50 Lunch
13:50 - 14:10 Overview and results of the segmentation challenge
John Winn1, Mark Everingham2
1Microsoft Research Cambridge; 2University of Leeds
[PDF]

14:10 - 15:10 Segmentation methods

Ranking figure-ground hypotheses for†object segmentation (BONN_SVM-SEGM)
Jo„o Carreira, Fuxin Li, Cristian Sminchisescu
University of Bonn
[PDF] NB: This is unpublished work. Please contact the authors if you plan to make use of any of the ideas presented.

Combining local and global bag-of-word representations for semantic segmentation (CVC_HOCRF)
Xavier Boix, Josep Maria Gonfaus, Fahad Shahbaz Kahn, Joost van de Weijer, Andrew Bagdanov, Marco Pedersoli, Jordi GonzŠlez, Joan Serrat
Computer Vision Center Barcelona
[PDF] NB: This is unpublished work. Please contact the authors if you plan to make use of any of the ideas presented.

Hierarchical CRF for object class segmentation (BROOKESMSRC_AHCRF)
Lubor Ladicky1, Chris Russell1, Pushmeet Kohli2, Philip H.S. Torr1
1Oxford Brookes University; 2Microsoft Research Cambridge
[PDF]

15:10 - 15:30 Conclusions and discussion
[PDF]

Organizers

Support

The preparation and running of this challenge is supported by the EU-funded PASCAL2 Network of Excellence on Pattern Analysis, Statistical Modelling and Computational Learning.