People Image Analysis Consortium
 

Home -> Projects ->Human Detection and Tracking Member Login  

Jump to: Components | Demos

 Description

In order to effectively detect people in images or videos, we break down the problem of detecting whole human body into different parts. Specificly, we develop algorithms and techniques to detect human Head, Face, Torso, Arms and Legs by probabilistic body representation.

 Components

The system include the following components.
1. Probabilistic Human Body Representation
2. Head/Face/Torso/Background Region Segmentation
3. Arm Localization
4. Omega Shape Head/Shoulder Outline Curve Detection
5. Human detection with the 2nd moment matrix

 Demos

Whole Human Profile Tracking
A movie to show the tracking of Whole Human Profile.
Download: avi [5.1 MB]
 
Arm Localization
A movie to show arm detection and tracking.
Download: avi [9.7 MB]
Arm Location 2
A movie to show arm detection and tracking.
Download: avi [43.9MB]

 Releases [ Members Only ]



PIA Consortium is a part of Robotics Institute, School of Computer Science, Carnegie Mellon University