The Marker-less MoCap Research group
| home |
research (Thesis) |
research
(PostDoc-NZ) |
research
(PostDoc-MPI) |
publications | curriculum vitae | private |
Currently I am leading a small research group at the Max Planck Center in Saarbruecken. The group members (currently) are
- Bodo Rosenhahn
- Juergen Gall
- Martin Sunkel
- Nils Hasler
- Christian Schmaltz (Joachim Weickert)
Click on the images for some example videos !
Pose Estimation - Segmentation
|
This project deals with embedding 3D shape priors in the segmentation process. For segmentation we follow an approach based on level set functions. It basically allows us to segment and track objects in complex environments, e.g. cluttered back ground, rapid movements, changing lighting conditions, etc. See (here), (here) and (here) for recent publications. |
|
This project deals with Marker-less Motion Capture in Outdoor environments. For successful tracking, the works in e.g.
(here),
(here), and (here)
are applied.
|
|
This project deals with the integration of clothing models during tracking, which can also be reanimated with other fabrics. See
here
and here for recent articles.
|
|
This project deals with the integration of motion priors during tracking. Here a Parzen-Rosenblatt density is computed and integrated in the tracking process. This allows to compensate for heavy noise and occlusions during tracking, as can be seen in the left. A comparison with a marker based tracking system still shows a suitable accuracy during tracking. See here, here and here for recent articles. |
|
This project deals with the integration of geometric priors during tracking. This can be important when there is interest in tracking people interacting with the environment or with special equipment, e.g. arising from sporting activities such as snow boarding, biking or just walking on the ground floor. See here, here and here for recent articles. |
Other projects deal with extensions of my PostDoc. News will be presented (hopefully :-) ) soon.