International Journal of Computer Vision
Special Issue Call for Papers
Structured Prediction and Inference

Editors:
Matthew B. Blaschko. University of Oxford, blaschko@robots.ox.ac.uk
Christoph H. Lampert, Max Planck Institute for Biological Cybernetics, chl@tuebingen.mpg.de

Background:
Many computer visions problems can be formulated naturally as
prediction tasks of structured objects. Image segmentation, stereo
reconstruction, human pose estimation and natural scene analysis are
all examples of such problems, in which the quantity one tries to
predict consists of multiple interdependent parts. The structured
output learning paradigm offers a natural framework for such tasks,
and recently introduced methods for end-to-end discriminative training
of conditional random fields (CRFs) and structured support vector
machines (S-SVMs) for image classification and interpretation show
that computer vision is not just a consumer of existing machine
learning developments in this area, but one of the driving forces
behind their development. The complexity of structured prediction
models makes the problem of inference in these models an integral part
of their analysis. While the machine learning literature has largely
focused on message passing, computer vision research has introduced
novel applications of branch-and-bound and graph cuts as inference
algorithms.  Articles addressing these issues are particularly
encouraged for submission to the special issue.


Topics:
Original papers are being solicited that have as topic one or more aspects
of the structured prediction framework in a computer vision setting, that is
they address the problem of prediction from an input space, such as
images or video, to a structured and interdependent output space.
Submissions can be theoretic or applied contributions as well as position
papers. Topics of interest include, but are not limited to:
* Training for structured output learning
   - Probabilistic vs. max-margin training
   - Generative vs. discriminative training
   - Semi-supervised or unsupervised learning
   - Dealing with label noise
* Inference methods for structured output learning
   - Exact vs. approximate inference techniques
   - Pixel, voxel, and superpixel random field optimization
   - Priors and higher order clique optimization
   - Approaches that scale to large amounts of training and test data
* Computer vision applications of structured output learning
   - Segmentation
   - Stereo reconstruction
   - Relationship between scene components
   - Hierarchical models

Paper Submission:
Authors are encouraged to submit high quality, original work that has
neither appeared in, nor is under consideration by, other
journals. All open submissions will be peer reviewed subject to the
standards of the journal. Manuscripts based on previously published
conference papers must be extended substantially.

Manuscripts should be submitted to: http://VISI.edmgr.com. This online
system offers easy and straightforward log-in and submission
procedures, and supports a wide range of submission file formats.

Paper submission deadline: February 26, 2010
Estimated Online Publication: Fall, 2010