| Presentation | 2020-03-16 Egocentric pedestrian motion prediction by separately modeling body pose and position Donghao Wu, Takuma Yagi, Yusuke Matsui, Yoichi Sato, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | We study the problem of forecasting human's motion captured from egocentric videos. We propose a novel learning approach by separately modeling human pose and its corresponding scale and position with two deep learning modules, whose outputs are later combined to make the final prediction. Our proposed method successfully forecasts the position and body pose of the target person with an ideal scale, relieving from the mean convergence problem. The experiment is evaluated based on First-Person Locomotion (FPL) dataset. The predictions show the separate modeling approach has plausible-looking visualization results upon egocentric settings, outperforming the state-of-the-art methods which only consider modeling single pose granularity of human motion that suffers from the mean convergence results. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | motion forecasting / egocentric vision / human dynamics / deep learning / neural network |
| Paper # | PRMU2019-72 |
| Date of Issue | 2020-03-09 (PRMU) |
| Conference Information | |
| Committee | PRMU / IPSJ-CVIM |
|---|---|
| Conference Date | 2020/3/16(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Yoichi Sato(Univ. of Tokyo) |
| Vice Chair | Toru Tamaki(Hiroshima Univ.) / Akisato Kimura(NTT) |
| Secretary | Toru Tamaki(NTT) / Akisato Kimura(OMRON SINICX) |
| Assistant | Yusuke Uchida(DeNA) / Takayoshi Yamashita(Chubu Univ.) |
| Paper Information | |
| Registration To | Technical Committee on Pattern Recognition and Media Understanding / Special Interest Group on Computer Vision and Image Media |
|---|---|
| Language | ENG |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Egocentric pedestrian motion prediction by separately modeling body pose and position |
| Sub Title (in English) | |
| Keyword(1) | motion forecasting |
| Keyword(2) | egocentric vision |
| Keyword(3) | human dynamics |
| Keyword(4) | deep learning |
| Keyword(5) | neural network |
| 1st Author's Name | Donghao Wu |
| 1st Author's Affiliation | The University of Tokyo(the Univ. of Tokyo) |
| 2nd Author's Name | Takuma Yagi |
| 2nd Author's Affiliation | The University of Tokyo(the Univ. of Tokyo) |
| 3rd Author's Name | Yusuke Matsui |
| 3rd Author's Affiliation | The University of Tokyo(the Univ. of Tokyo) |
| 4th Author's Name | Yoichi Sato |
| 4th Author's Affiliation | The University of Tokyo(the Univ. of Tokyo) |
| Date | 2020-03-16 |
| Paper # | PRMU2019-72 |
| Volume (vol) | vol.119 |
| Number (no) | PRMU-481 |
| Page | pp.pp.39-44(PRMU), |
| #Pages | 6 |
| Date of Issue | 2020-03-09 (PRMU) |