S. Kevin Zhou
Email: s.kevin.zhou AT gmail.com
Web: http://www.cfar.umd.edu/~shaohua/
I am interested in building unconstrained biometrics recognition and surveillance system. Below itemizes my efforts along this line: Face recognition from videos and under illumination/pose variations, gait modeling, and surveillance tracking and metrology, etc.
| Face recognition from videos | |
 |
Still-to-video
&
video-to-video face recognition
[HumanID project] Probabilistic identity characterization [DARPA ONR project] |
|
|
|
 |
Adaptive
visual tracking and recognition [CTA project, also see video demo & MERL website] |
|
|
|
 |
Review of FR from multiple stills and videos |
|
|
|
| Face recognition under variations | |
 |
Generalized photometric stereo [HumanID project] |
|
|
|
 |
Face recognition across illuminations and poses [NSF ITR project] |
|
|
|
| Gait modeling & object recognition | |
 |
Pedestrian segmentation via manifold learning |
|
|
|
 |
Appearance modeling under geometric context [VACE project] |
|
|
|
| Surveillance tracking & metrology | |
 |
Adaptive
visual tracking
[CTA project, also see video demo
& MERL
website] |
|
|
|
 |
Robust two-view tracking with homography [VIVID project, also see video demo] |
|
|
|
 |
Video tracking by foreground and background modeling [VIVID project, also see video demo] |
|
|
|
 |
Surveillance video metrology |