CSE 590V: Computer Vision Reading Group

CSE590V is a Seminar/Reading Group focused on recent trends in computer vision. We will cover papers from recent computer vision conferences (CVPR, ICCV, ECCV). The seminar is open to everyone. We especially encourage first year graduate students who may be considering computer vision or related areas as potential research avenues.

The seminar meets every Wednesday from 2:30PM to 3:30PM in CSE 403.

The class mailing list is cse590v(at)cs.washington.edu.

Those of you who are taking the course or are interested in keeping up with the topics being discussed in the seminar, please subscribe to the mailing list at https://mailman.cs.washington.edu/mailman/listinfo/cse590v

Requirements:

Prepare a 45 min. presentation that will explain the topic and the methods used.

The students that will present on Wednesday should meet with the organizers on Monday 4PM for a dry run of the presentation and questions (email us for the location).

Attend all meetings and participate in the discussion.

Organizers

  1. Changchang Wu, ccwu(at)cs.washington.edu
  2. Ira Kemelmacher-Shlizerman, kemelmi(at)cs.washington.edu
  3. Lynn Yang, yang(at)cs.washington.edu

Schedule

Papers

1) Image distortion  estimation
    Yuandong Tian and Srinivasa G. Narasimhan
    A Globally Optimal Data-Driven Approach for Image Distortion Estimation, CVPR 2010
    Project webpage: http://www.cs.cmu.edu/~ILIM/projects/IM/globalopt/research_globalopt.html

2) Text detection in natural scenes

    B.Epshtein, E. Ofek, Y. Wexler,
    Detecting Text in Natural Scenes with Stroke Width Transform, CVPR 2010,

    Kai Wang, Serge Belongie,
    Word Spotting in the Wild (project), ECCVV 2010

3) Face tracking

    Jason M. Saragih, Simon Lucey, Jeffrey F. Cohn
    Face Alignment through Subspace Constrained Mean-Shifts, ICCV 2009
    http://www.ri.cmu.edu/publication_view.html?pub_id=6417&menu_code=0307

4) Blocks World
    Abhinav Gupta, Alexei A. Efros and Martial Hebert,
    Blocks World Revisited: Image Understanding Using Qualitative Geometry and Mechanics,  ECCV 2010
    [BEST PAPER RUNNER UP]

5) Low rank matrices

    RASL: Robust Alignment by Sparse and Low-rank Decomposition for Linearly Correlated Images,
    Yigang Peng, Arvind Ganesh, John Wright, Wenli Xu, and Yi Ma. Submitted to IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), July 2010.
    [Project website with sample code and data] [Oral at CVPR 2010]

    Robust Principal Component Analysis: Exact Recovery of Corrupted Low-Rank Matrices via Convex Optimization,
    John Wright, Arvind Ganesh, Shankar Rao, Yigang Peng, and Yi Ma. In Proceedings of Neural Information Processing Systems (NIPS), December 2009.

    [BEST PAPER AWARD] Efficient Computation of Robust Low-Rank Matrix Approximations in the Presence of Missing Data using the $L_1$ Norm (PDF)
    Anders Eriksson, Anton van den Hengel

    Tutorial in CVPR 2010: http://vision.jhu.edu/gpca/cvpr10-tutorial-multi-subspace.htm

6) Facial expression

    F. Zhou, F. De la Torre and J. F. Cohn 
    Unsupervised Discovery of Facial Events   
    IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2010 (oral).

7) Object detection with Pictorial structures:

   
P. Felzenszwalb, R. Girshick, D. McAllester, D. Ramanan
    Object Detection with Discriminatively Trained Part Based Models
    IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 32, No. 9, September 2010

    P. Felzenszwalb, R. Girshick, D. McAllester
    Cascade Object Detection with Deformable Part Models
    IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2010,  (oral cvpr)

    "Efficient Matching of Pictorial Structures"  CVPR 2000
    Pedro F. Felzenszwalb and Daniel P. Huttenlocher
    (got the LH prize in CVPR 2010)

8) Human pose estimation

    Cascaded Models for Articulated Pose Estimation, (oral ECCV)
    Authors: Benjamin Sapp (University of Pennsylvania), Alexander Toshev (University of Pennsylvania), Ben Taskar (University of Pennsylvania)

    Combining Discriminative and Generative Methods for 3D Deformable Surface and Articulated Pose Reconstruction
    Authors:  Mathieu Salzmann (EECS & ICSI, UC Berkeley) , Raquel Urtasun (Toyota Technological Institute, Chicago)

    Lubomir Bourdev, Subhransu Maji, Thomas Brox, Jitendra Malik,Detecting People Using Mutually Consistent Poselet Activations, ECCV 2010

    Monocular 3D Pose Estimation and Tracking by Detection
    Authors:  Mykhaylo Andriluka (TU Darmstadt) , Stefan Roth (TU Darmstadt) , Bernt Schiele 

    Contour People: A Parameterized Model of 2D Articulated Human Shape
    Authors:  Oren Freifeld (Brown University) , Alex Weiss (Brown University) , Silvia Zuffi () , Michael Black (Brown University )

9) Recognition by context:

    Object-Graphs for Context-Aware Category Discovery
    Yong Jae Lee, Kristen Grauman

    Exploiting Hierarchical Context on a Large Database of Object Categories
    Myung Jin Choi, Joseph Lim, Antonio Torralba

    The Chains Model for Detecting Parts by Their Context (PDF, supplementary material)
    Leonid Karlinsky, Michael Dinerstein, Daniel Harari, Shimon Ullman

    Modeling Mutual Context of Object and Human Pose in Human-Object Interaction Activities  
    Bangpeng Yao, Li Fei-Fei, [BEST PAPERS HONORABLE MENTION AWARD]

10) Mathematical classifier design:

    On the design of robust classifiers for computer vision
    Hamed Masnadi-Shirazi, Nuno Vasconcelos, Vijay Mahadevan

    On the Design of Loss Functions for Classification: theory, robustness to outliers, and SavageBoost.
    Hamed Masnadi-Shirazi and Nuno Vasconcelos.
    Proceedings of Neural Information Processing Systems (NIPS), Vancouver, Canada, Dec 2008.

11)  Depth from Diffusion

    "Depth from Diffusion,"  Changyin Zhou, Oliver Cossairt, Shree Nayar,
    IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Jun, 2010.
    http://www1.cs.columbia.edu/CAVE/projects/depth_from_diffusion/

12)  Photometric Stereo for Dynamic Surface Orientations

    Hyeongwoo Kim, Bennett Wilburn, and Moshe Ben-Ezra, Photometric Stereo for Dynamic Surface Orientations, ECCV 2010
    http://research.microsoft.com/apps/pubs/default.aspx?id=138645

13) Reconstruction

    Hybrid Multi-view Reconstruction by Jump-Diffusion
    Florent Lafarge, Renaud Keriven, Mathieu Brédif, Vu Hoang Hiep

    Building Reconstruction using Manhattan-World Grammars
    Carlos A. Vanegas, Daniel G. Aliaga, Bedřich Beneš

14) 5D Motion Subspaces for Planar Motions
    Roland Angst (ETH Zurich), Marc Pollefeys (ETH Zurich)PDF

15) Face Recognition Based on Image Sets
    Hakan Cevikalp (Eskisehir Osmangazi University) , Bill Triggs (CNRS)PDF