|
|
Learning Graphical Model for Human Motion Characterization Using Genetic Optimization |
INVESTIGATORS
Dr. Hau San WONG, Ms. Huiyang QU
KEYWORDS:
genetic algorithm, graphical model
FUNDING AGENCY
City University of Hong Kong
BRIEF DESCRIPTION
We present a novel method of using genetic algorithm (GA) to learn a graphical model which is used for human motion characterization. The modeling of human movements will involve a high dimensional joint probability density function. With this graphical model, the joint probability distribution can be decomposed into a number of low dimensional distributions which are represented as tree models and triangulated models. To automatically search for such a model from a database of cases is a NP-hard problem. We use GA to solve this problem, which can optimize both the ordering structure and the conditional independence relationship of the graphical model. The searched graphical models are used to classify different types of human motions. The experimental results demonstrate that, compared with a previous greedy search algorithm, the GA is more effective for optimization of the graphical model.
Given two motion sequences: walking on uneven terrain and walking while dribbling basketball, we learn the graphical models using GA and greedy algorithm. Then we use these graphical models to classify new motion data.
(a) Walking unevenly (b) Walking and dribbling basketball
Figure 1. The two motion sequences (from CMU motion capture database).
(a) Graphical model of walking on uneven terrain by the greedy. (b) Graphical model of walking on uneven terrain by the GA.
(c) Graphical model of walking and dribbling basketball by the greedy. (d) Graphical model of walking and dribbling basketball by the GA.
Figure 2. Searched graphical models.
Greedy
Walking unevenly
Walking and dribbling
Walking unevenly
1880
120
Walking and dribbling
230
1770
(a) Classification using greedy searched graph.
GA
Walking unevenly
Walking and dribbling
Walking unevenly
1891
109
Walking and dribbling
112
1888
(b) Classification using GA searched graph.
Figure 3. Confusion matrices of classification.
PUBLICATIONS
Huiyang Qu, Hau San Wong, Bo Ma. Learning Graphical Model for Human Motion Characterization Using Genetic Optimization. The 9th International Conference on Control, Automation, Robotics and Vision (ICARCV 2006).
