Human
Detection, Tracking, and Activity Understanding
The
Intelligent Systems Laboratory at RPI has long performed research related to
human activity modeling and recognition. Specifically, our related efforts include
two areas: computer vision and probabilistic graphical models. In computer
vision, we have performed research in face detection and tracking, human
detection and tracking, human activity modeling and recognition. These efforts
have led to algorithms and software for real time multi-view face detection and
tracking, multi-target human body detection and tracking, human action and
facial activity recognition.
In
probabilistic graphical models, we have significant experience in developing
theories for graphical model learning and efficient inference for different
graphical models including Hidden Markov Models, Conditional Random Fields,
Dynamic Bayesian Networks, and the Influence Diagrams. In addition, we have
applied graphical models to different applications including information
fusion, human action modeling and recognition, facial activity understanding,
image and video segmentation, and decision making. The latest research in human
activity modeling and recognition has demonstrated the superior performance of
the probabilistic graphical models over other competing methodologies. We are
currently combining various graphical models to develop a unified hierarchical
probabilistic model for activity modeling, representation, and inference.
Our
current efforts include person-object activity/event recognition (DARPA VIRAT
program), persistent detection, tracking and activity recognition of dismounts
(sponsored by Army Research Office) and face recognition from distance for
maritime applications (Sponsored by ONR). Our efforts in these areas have been
supported by different governmental agencies including DARPA, ARO, ONR, AFOSR,
and NSF.