Modeling the product manifold of posture and motion

Ankur Datta,Yaser Sheikh,Takeo Kanade

doi:10.1109/iccvw.2009.5457588

Abstract

Long-term human motion is composed of an ensemble of different activities with varying complexity. This makes it challenging to develop models to accurately estimate human motion. In this paper, we exploit the dependencies that exist between posture and motion for long-term human motion estimation. We propose to model the nonlinear motion manifold as a collection of local linear models, noting that given a particular posture, the variation in motion for that posture can be well-approximated by a linear model. A collection of local linear models is easy to fit and also has the expressiveness to encode several activities in any arbitrary order. Furthermore, to account for the varying complexity of different activities, each local linear model can have a different dimensionality. A collection of local linear models, thus, avoids the limitation of global models that require a uniform dimensionality for the latent motion manifold. This model allows us to linearly regularize motion estimation algorithms over the nonlinear human motion manifold. Our results demonstrate that a collection of local linear models provides an effective representation for the motion manifold when compared to other global models such as the bilinear model [18] and the Principal Component Analysis [14].

Highlights

Human motion is influenced by the internal intentions of the actor and the external constraints of the actor’s environment
We used 500 clusters with 400 nearest neighbors to learn a collection of local linear models (LLMs) using weighted Principal Component Analysis (PCA)
We have proposed an approach that exploits the dependencies between posture and motion for human motion estimation

Summary

Introduction

Human motion is influenced by the internal intentions of the actor and the external constraints of the actor’s environment. The series of activities performed by the driver depends on what the driver wants to do (e.g. turn on the air conditioner, turn up the volume, or adjust the seat height), and what driving requires (e.g. shift gears, adjust the rear view mirror, and turn the steering wheel). The challenge in estimating motion in such settings is that both the intentions of the driver and the nature of environmental constraints usually remain hidden. Human motion manifests itself as a series of activities with varying complexity whose order is arbitrary. Activities have been modeled globally by low-dimensional manifolds. To handle arbitrary orderings of activities, extending global models would require a combinatorial increase in the training data

Objectives

Results

Conclusion