Face Recognition Research, the Top 20
Institutions
A featured institution selection from
Essential Science IndicatorsSM
This month, ScienceWatch.com presents a
listing of the top 20 institutions which, according to our
Special Topic on face recognition, attracted the highest
total citations to their papers published on the topic
inThomson
Reuters-indexed journals.
These institutions are the top 20 ranked by total cites out of a pool
of 2,674 institutions publishing on this topic, based on a keyword
search for "face recog*" OR "face ident*" OR "face detect*" OR "view-based
recog*" OR "object recog*" in the titles, abstracts, and keywords sections
of original articles, reviews, and proceedings papers. The data pool was
also narrowed by field.
The resulting list of institutions shows that face recognition technology
is of international interest; fourteen of the institutions are US-based,
three are from Europe, two from Asia, and one is from Canada. Research is
being conducted by universities, government organizations, and the private
sector. Fifteen of the organizations are academic, two are governmental,
two are industrial, and one is an academic/governmental partnership.
Seven of these institutions also made an appearance in the
ScienceWatch.com feature on the top 20 institutions in the field
of Computer Science from the Essential Science
IndicatorsSM from Clarivate database. These are: MIT,
the University of California, San Diego, the University of Illinois,
Carnegie Mellon University, IBM, AT&T, and the University of
California, Berkeley.
Ranking at #1 is MIT, with 114 papers cited a total of 2,429
times—almost 1,000 more citations than the next institution on the
list. One of MIT's top researchers in this area is Tomaso Poggio, who ranks
at #5 on our list of the top 20 scientists. Areas under exploration at MIT
include multidimensional morphable models, view-based human face detection,
cortex-like mechanisms, and object detection by components.
Coming in at #2 is Michigan State University, with 43 papers cited a total
of 1,467 times. An overwhelming majority of these papers were authored or
co-authored by Anil K. Jain, who ranks at #1 by citations on the top 20
scientists list. Among Michigan's highly cited research are papers on data
clustering, statistical pattern recognition, face detection in color
images, the use of faces and fingerprints for personal identification, and
kernel principal component analysis.
The institution ranked at #3 is the University of California, San Diego,
with 50 papers cited a total of 1,404 times. UCSD's papers examine the use
of shape contexts for object recognition, slow feature analysis,
classifying facial actions, and face recognition using independent
component analysis. David Kriegman, who ranks at #6 on our list of the top
20 scientists, is affiliated with UCSD—he is also affiliated with the
next institution on our list.
The #4 slot belongs to the University of Illinois, with 121 papers cited a
total of 906 times. Illumination cone models, geometrical face models,
detecting objects with sparse partial representation, and graph embedding
and extensions are some of the areas being researched here.
Carnegie Mellon University ranks at #5, with 158 papers cited a total of
904 times. Highly cited papers for CMU include such topics as neural
network based face detection, the use of spin images, the LDA algorithm,
and deciphering objects in complex or cluttered 3D scenes.
Coming in at #6 is the US Army, with 41 papers cited 896 times. The Army's
most-cited work concerns the FERET evaluation methodology, a program that
was managed by Jonathon Phillips, who is now at the National Institute of
Standards and Technology (NIST). Other Army projects involve targeting and
intelligence recognition programs.
Ranking at #7 is the first of the European institutions: Delft University
of Technology with 15 papers cited a total of 722 times. Delft's research
includes automated analysis of facial expressions, image processing with
neural networks, auto video logo detection, and object recognition through
one-class learning.
We're back in the US with #8—Ohio State University with 37 papers
cited 702 times. Neural computing, precision range image recognition, and
analyses of the LDA algorithm are in focus at OSU.
The University of Amsterdam ranks at #9 in our analysis, with 32 papers
cited 697 times. Amsterdam's research includes human-computer interactions,
face detection by aggregated Bayesian network classifiers, and dynamic
programming for matching shapes.
Rounding out the top 10 is the University of British Columbia, with 29
papers cited a total of 662 times. Image features from scale-invariant
keypoints, matching words and pictures, vision-based navigation, and
probabilistic models for object recognition are among the topics being
explored at this institution.
The remaining institutions on the list include four US universities (the
University of Connecticut at #15, George Mason University at #17, CUNY at
#18, and the University of California, Berkeley at #19), one US government
agency (NIST at #11), two US corporations (IBM at #12 and AT&T at #14),
the third European organization (Max Planck Society at #13), and the two
Asian institutions (Hong Kong Polytechnic University at #16 and Nanjing
University of Science & Technology at #20).
The full list of the top 20 institutions in face recognition by total cites
is as follows: