| Presentation | 2010-10-21 3D human motion tracking from monocular image with hierarchical learning model Ayumi MATSUMOTO, Xiaojun WU, Kaoru WAKABAYASHI, Norihiko MATSUURA, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | We propose a robust method to estimate 3D human motion from monocular camera. In recent years, nonlinear and probabilistic dimensional compression technique such as GPLVM and GPDM are used as prior knowledge to estimate 3D human motion from monocular camera. However, in these traditional methods, one prior model on full body made from one 3D human motion. From 2D pictures, generaly, it is diffucult to track the whole body motion, which damages the estimation results greatly. We learn two model for one 3D human motion data by GPDM, full body's model and part body's model. Furthermore, we formulate the connection of two models by a gauss process. We aim at realizing 3D human motion estimate of the whole body when only part information of the human body was given from a 2D picture as observation data. By this report, we formulate learning between two models and show what can rebuild the full body's 3D human motion from the model that I learned from part body's model. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Model-Based Human Pose estimation / monocular image / human tracking / hierarchical learning model / Gaussian Process |
| Paper # | MVE2010-56 |
| Date of Issue |
| Conference Information | |
| Committee | MVE |
|---|---|
| Conference Date | 2010/10/14(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| Paper Information | |
| Registration To | Media Experience and Virtual Environment (MVE) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | 3D human motion tracking from monocular image with hierarchical learning model |
| Sub Title (in English) | |
| Keyword(1) | Model-Based Human Pose estimation |
| Keyword(2) | monocular image |
| Keyword(3) | human tracking |
| Keyword(4) | hierarchical learning model |
| Keyword(5) | Gaussian Process |
| 1st Author's Name | Ayumi MATSUMOTO |
| 1st Author's Affiliation | NTT Cyber Space Laboratories, NTT Corporation() |
| 2nd Author's Name | Xiaojun WU |
| 2nd Author's Affiliation | NTT Cyber Space Laboratories, NTT Corporation |
| 3rd Author's Name | Kaoru WAKABAYASHI |
| 3rd Author's Affiliation | NTT Cyber Space Laboratories, NTT Corporation |
| 4th Author's Name | Norihiko MATSUURA |
| 4th Author's Affiliation | NTT Cyber Space Laboratories, NTT Corporation |
| Date | 2010-10-21 |
| Paper # | MVE2010-56 |
| Volume (vol) | vol.110 |
| Number (no) | 238 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |