Articulated Object Modelling using Neural Gas Networks

S. García, D. Rowe, J. Gonzàlez, and J.J. Villanueva (Spain)


Segmentation, human motion, self-organizing maps, and neural gas networks


One of the main problems of human motion analysis is to find a structured model which describes movements. This paper proposes to apply Neural Gas networks to model articulated objects or actors. Our algorithm decomposes objects into regions which are used to find articulation points. This approach will be explained within the Human Sequence Evaluation (HSE) framework.

Important Links:

Go Back