Basic HTML Version
Legal imaging was also a successful research
area last year. A EURECOM student was even
honored in Australia for her work dedicated to
two subjects that are usually studied separately:
Video sourcing identification and image tamper-
ing detection”.
Biometrics is still spearheading your
group’s research. What have been
your key accomplishments in this
area in 2011?
J-L. D. :
“Biometrics is entering a new phase.
When it comes to security, people with mali-
cious intent are bursting with imagination to
dupe our systems. The purpose of our work is to
anticipate their tricks, for instance, recognizing
a person wearing a mask, in our applications.
These new challenges lead to second generation
applications that must be efficient in spite of
information scarcity from the captured images.
Our models must work in uncontrolled situa-
tions and environments: a failing lighting sys-
tem, non-cooperative persons, etc.
In the Tabula-Rasa European project, we
have been working on several of these issues.
A dozen prestigious partners are cooperating
in the project which benefits from EURECOM’s
expertise in 3D facial recognition. Our work in
biometrics has also led to novel approaches,
such as facial aesthetics, a subjective concept
that our models are attempting to formalize with
criteria that have proven to be relevant. One of
our PhD students has been prized for her work
in the subject at a Barcelona conference. The
subject could lead to many applications in the
area of fashion or ID photos. We’re planning on
starting a large-scale project in this area in 2012”.
What about videosurveillance?
Haven’t parts of your research been
followed worldwide?
J-L. D. :
“Yes, that’s true. We continued work-
ing on prevention with systems capable of rec-
ognizing hooligans in a crowd, for instance, in
spite of scarves hiding the face. This deals with
‘stolen’biometrics which must be based on cri-
teria other than that of the face: Size, clothing
color, even weight. This last point was much
debated. We showed that it was technically
possible to assess a person’s weight with a 5%
error margin simply based on his/her image.
Obviously, the applications of this tool aremany.
It could even be used by astronauts who need
to check their weight regularly while in weight-
lessness”.
MULTIMEDIA INFORMATIONPROCESSING SEGMENTATIONANDCLASSIFICATION
MULTIMEDIACONTENTFILTERING SPEECHPROCESSING MULTIMEDIA IMAGING
BIOMETRY WATERMARKING SEMENTICWEB
multimedia
communications
nationalprojects
ACAV
BIO-RAFALE
DATALIFT
VIDEO-ID
FAR3D
M4SVR
OpenSEMANR
RPM2
SUMACC
VIDEOSENSE
europeanprojects
ACTIBIO
ALIAS
LinkedTV
OpenSEM
VIDEOSENSE(E)
TabulaRasa
current projects
21
2011annual report