|
Published Articles >> Table of Contents >> Abstract
IEEE International Workshop on Analysis and Modeling of Faces and Gestures
p. 36
Human Recognition of Familiar and Unfamiliar People in Naturalistic Video
D. A. Roark, University of Texas at Dallas
A. J. O'Toole, University of Texas at Dallas
H. Abdi, University of Texas at Dallas
Full Article Text:
  
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/AMFG.2003.1240821
Send link to a friend
| Abstract |
|
Understanding the human performance factors that mediate
successful person identification can be helpful in the development
of automatic face recognition algorithms. Face
familiarity and facial motion are two factors that seem especially
useful when subjects make recognition decisions
from challenging viewing formats. We tested the effects of
these two factors on person recognition from naturalistic,
surveillance-like video. Subjects learned faces from either
static photographs or facial speech videos and were asked
to recognize people from whole body gait videos. We found
that the more experience participants had with a face during
learning (i.e., 1-view, 2-view, and 4-view conditions),
the better their recognition performance for people in the
whole body video gait clips. Thus, familiarizing subjects
with high-resolution images or videos of faces was sufficient
to improve recognition from low-resolution, whole-body images.
Moreover, participants who learned faces from dynamic
video clips were more accurate than participants who
learned the faces from static images, but only when they
were familiar with the faces. Facial motion and face familiarity
may therefore play a role in understanding recognition
when there are photometric inconsistencies between
learning and test stimuli.
|
 Additional Information
|
Citation:
D. A. Roark, A. J. O'Toole, H. Abdi,
"Human Recognition of Familiar and Unfamiliar People in Naturalistic Video,"
amfg,
p. 36,
IEEE International Workshop on Analysis and Modeling of Faces and Gestures,
2003
|
|
