Alexei (Alyosha) Efros
Adjunct Associate Professor
The Robotics Institute
and Computer Science
Department
School of Computer Science
Carnegie Mellon University
Pittsburgh, PA 15213, USA
Office: 225 Smith Hall
Phone: +1 (412) 268-1234
Email:
efr...@cs.cmu.edu
After a wonderful decade at CMU, I have moved the the EECS
Department at UC Berkeley in Fall 2013. Please visit my
new web page.
I am originally from
St.
Petersburg, Russia, but, due to some mixup at
the travel agency, I now live in
Pittsburgh, USA.
I got my PhD from
UC
Berkeley in 2003 under the supervision of Jitendra Malik. I then
spent a year as
a fine fellow in
lovely Oxford, working
with Andrew Zisserman and
the Visual Geometry
Group.
Since 2004, I am faculty at CMU with a joint appointment in
the
Robotics Institute
and CS Department (and a courtesy
appointment in
Machine Learning Department).
I am
a member of the CMU Computer Vision
Group
and the
CMU Graphics
Lab, and also have close collaboration with the
WILLOW Lab
at INRIA/ENS in Paris.
Here is a slightly more official bio.
Research
Antonio Torralba and I
are guest-editing an IJCV special issue
on
Big Visual Data. Submission Deadline: extended to May 31.
My research is in the area of computer vision and computer graphics,
especially at the intersection of the two. I am particularly
interested in using data-driven techniques to
tackle problems which
are very hard to model parametrically but where large quantities of
visual data, now called "Big Data", are readily available (see my recent
talk on "What Makes Big Visual Data Hard?"). The ultimate
goal is to use the
ever-growing amount of stored visual information (digital photo
albums, webcams, movies, etc.) to learn, understand, and resynthesize
the visual world around us.
In very broad strokes, here are the main current themes of my
group's research:
-
Qualitative Reasoning for Image Understanding: The ability to
see and understand the three-dimensional world behind a
two-dimensional image goes to the very heart of the computer vision
problem. The overall objective of this research effort is, given a
single image, to automatically produce a coherent
interpretation of the depicted scene. On one level, such
interpretation should be able to qualitativly capture scene properties
such as geometric layout, occlusion relationships, camera viewpoint,
scale, illumination, geographic properties, etc. But more than that,
the goal is to capture the overall qualitative sense of the
scene. For an overview, see my recent keynote
lecture at BMVC'09.
-
Building the "Visual Memex": What could you do with a billion
images? Can this vast heap of data be organized without loosing its
representation power? Taking inspiration from Vannevar Bush's
memex, our
aim is to utilize the huge amount of existing visual data to discover
links and connections between all visual elements, i.e. a giant
multi-graph. Once constructed, the visual memex can be used to
"explain" novel visual signal not in terms of hard categories ("car",
"chair", "city"), but rather in terms of what has been seen before.
Rather than asking "What is this?", we'd like to ask "What is
this like?". For an overview, see my recent
presentation and representative publications:
Malisiewicz'08,
Malisiewicz'09,
Hays'08,
Russell'09,
Malisiewicz'11,
Shrivastava'11.
-
Understanding (and Faking) Visual Realism: Why is it that most
computer-generated imagery doesn't look very realistic? What is it
that the Renaissance artists knew that we don't? Which bits of the
visual experience is it important to "get right", and which could be
safely faked without anyone noticing? The ultimate goal is to make
synthesized images appear as real and convincing as regular
photographs. Representative publications:
Lalonde'07,
Lalonde'07b,
Hays'07,
Lalonde'09.
Postdocs
Graduate Students
Former PhD Students
- Dr. Santosh Kumar
Divvala (Robotics, PhD 2012, now postdoc at University of Washington),
co-advised with Martial Hebert
- Dr. Tomasz Malisiewicz
(Robotics, PhD 2011, now postdoc at MIT)
- Dr. Jean-François
Lalonde (Robotics, PhD 2011, now postdoc at Disney Research),
co-advised with Srinivas Narasimhan
-
Dr. James Hays
(CSD, PhD 2009, now assistant professor at Brown)
-
Dr. Derek Hoiem
(Robotics, PhD 2007, now assistant professor at UIUC), co-advised with Martial Hebert
Former Postdocs
Teaching
Selected Projects
List of all my publications is also available on
Google
Scholar.