[humaine news] CFP: IEEE Trans PAMI, Special Issue on Real-World Face Recognition (deadline: March 15, 2010)

[Kostas Karpouzis]

IEEE Transactions on Pattern Analysis and Machine Intelligence
Special Issue on Real-World Face Recognition

Guest Editors:
Gang Hua, Nokia Research Center Hollywood
Ming-Hsuan Yang, University of California, Merced
Erik Learned-Miller, University of Massachusetts, Amherst
Yi Ma, Microsoft Research Asia
Matthew Turk, University of California, Santa Barbara
David Kriegman, University of California, San Diego
Thomas Huang, University of Illinois at Urbana-Champaign


Topic Description

Research on face recognition has had a revival in recent years. This
is largely due to the need for robust face recognition technologies
for consumers to tag digital photos and facilitate their organization
and online sharing. Unlike traditional access control scenarios, where
facial images are taken under controlled lighting, pose, and
expression, face recognition in consumer digital imaging suffers from
uncontrolled lighting, large pose variation, a range of facial
expressions, make-up, and severe partial occlusions. Likewise, in many
surveillance scenarios, video may be acquired in uncontrolled
situations, from moving cameras.  These factors have challenged robust
face recognition for decades, but they are still unresolved.

In a traditional face recognition system, a set of labeled gallery
faces is first compiled; a new probe image is then matched against the
gallery face database to be recognized as a known face or rejected.
Depending how much control we have on the gallery and probe faces, we
can roughly categorize face recognition tasks into three scenarios:
face recognition under well-controlled, moderately controlled, or
uncontrolled environments. In a well controlled environment, image
capture is constrained for both the gallery and probe faces. In a
moderately controlled environment, we lose control of either the
gallery faces or the probe faces, but not both. In an uncontrolled
environment, we have control over neither. We can similarly define
these three categories for face recognition tasks such as face
clustering.

Face recognition in well-controlled environments is relatively mature
and has been heavily studied, but face recognition in uncontrolled or
moderately controlled environments is still in its early stages.
While earlier face recognition methods applied pattern recognition and
machine learning techniques for matching in the image space and
subsequent work focused on geometric, lighting and reflectance models
of faces, these techniques alone appear insufficient to solve new set
of challenges.  However, many are under the impression that face
recognition in general is solved, or that the uncontrolled scenarios
are too difficult to solve in practice. Neither seems to be the case.
Practical face tagging systems are emerging based on existing face
recognition technologies. For these, a good user interface (UI) and
user experience (UX) design is essential in order to compensate for
the possible failures from the face recognition algorithm. Moreover,
in many of these applications, there may be additional contextual
information we can leverage to improve face recognition accuracy.
Multidisciplinary approaches may be necessary to deliver real working
systems.

This PAMI special issue is dedicated to addressing the interesting
challenges in this domain and to promote systematic research and
evaluation of promising methods and systems. We encourage the
submission of novel solutions and thorough evaluations of issues in
face recognition in environments that are not highly controlled,
including, but not necessarily limited to:

*Design of robust face similarity features and metrics
*Robust face clustering and sorting algorithms
*Novel user interaction models and face recognition algorithms for face 
tagging
*Novel applications of face recognition on the web
*Novel computational paradigms for face recognition
*Challenges in large scale face recognition tasks, e.g., on the internet
*Face recognition with contextual information
*Face recognition benchmarks and evaluation methodology for moderately
controlled or uncontrolled environments
*Video face recognition


Timelines

*Submissions due:	03/15/2010
*First review results:	05/28/2010
*Revision due:		06/25/2010
*Second review results:	07/23/2010
*Final manuscript due:	08/20/2010
*Publication date:	01/2010


Paper submission and review

Papers should be well aligned with the theme of the special issue and
must be submitted online through the PAMI manuscript central site,
with a note/tag designating the manuscript for this special issue. All
submissions will be peer-reviewed by at least three experts in the
field. Priority will be given to work with high novelty and potential
impact. Particular attention will be given to comparisons with the
state of the art, and to the discussion of statistical significance of
the results.