Accurate tracking of the human hand using a particle based tracker
Our model-based human motion tracker is currently optimized for tracking the full human body. When investigating pick-and-place actions and human interactions the human hand plays a key role. Therefore we want to adapt our existing framework for the full human body to a special hand model.
Project type:
Bachelor Thesis
Project overview: We are using an particle filter based approach to track the human body by using a fully adaptable, high dimensional model with up to 50 degrees of freedoms. For further background information about the project context see www9.in.tum.de/research/memoman.
Task description: Apply our existing methods for the full body to a special hand model. Especially the model initialization and using a good appearance model could be central tasks. The topic is quite large and can be adjusted to your interests.
Prerequisites: Good programming skills with C/C++. Experienced and/or interested in one of the following topics: image processing, simulations, OpenGL, CUDA/OpenCL/GPU acceleration.
Professor: Prof. Beetz, Michael, PhD
Supervisor: Weikersdorfer, David
State: Running
