Research Projects
Multi-Camera Action Recognition |
|
Dynamic selection of the best view(s) for recognition can improve accuracy and reduce computational overhead.
More details |
|
Multi-Camera, Multi-Person Detection |
|
By selectively verifying likely person locations with high-level detectors accuracy is improved with minimal impact on speed.
More details |
|
Multi-Camera Tracking |
|
By using single-camera trackers in a dynamic subset of the available views, we keep speed high for multi-camera tracking.
More details |
|
Gamesourcing |
|
We created a gamesourcing framework that can run side-by-side with any Microsoft Kinect-based PC game to collect annoted pose estimation data.
More details |
|
Biomechanical Visualization |
|
We created a framework combining interactive visual analysis with automated techniques
for investigating 3D biomechanical motion data.
More details |