| Presentation | 2018-12-14 Correcting Outputs of Ensemble LSTMs by CRF for Robust Activity Recognition Haruka Abe, Yuta Hayakawa, Takuya Hino, Motohide Sugihara, Hiroki Ikeya, Masamichi Shimosaka, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Recognizing the activities of the construction vehicle helps to assess the skill of the workers or give a technical guidance to improve the workers' skill. Many research employ LSTM which can utilize long-term dependency for the activity recognition tasks. When we apply LSTM to the activity recognition tasks, we obtain multiple prediction results for data at the same frame. Conventional methods tend to employ an output only from the last frame of LSTM and others are disposed. Prediction accuracy can be improved by integrating these multiple outputs, but there is no current study. In this research, we tackle this issue by employing CRF. We'll show the proposed method overcomes the other methods throughout the experiment. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Activity recognition / Construction vehicle / LSTM / CRF / Deep learning |
| Paper # | PRMU2018-94 |
| Date of Issue | 2018-12-06 (PRMU) |
| Conference Information | |
| Committee | PRMU |
|---|---|
| Conference Date | 2018/12/13(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Shinichi Sato(NII) |
| Vice Chair | Yoshihisa Ijiri(Omron) / Toru Tamaki(Hiroshima Univ.) |
| Secretary | Yoshihisa Ijiri(NEC) / Toru Tamaki(Osaka Univ.) |
| Assistant | Go Irie(NTT) / Yoshitaka Ushiku(OSX) |
| Paper Information | |
| Registration To | Technical Committee on Pattern Recognition and Media Understanding |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Correcting Outputs of Ensemble LSTMs by CRF for Robust Activity Recognition |
| Sub Title (in English) | |
| Keyword(1) | Activity recognition |
| Keyword(2) | Construction vehicle |
| Keyword(3) | LSTM |
| Keyword(4) | CRF |
| Keyword(5) | Deep learning |
| 1st Author's Name | Haruka Abe |
| 1st Author's Affiliation | Tokyo Institute of Technology(Tokyo Tech) |
| 2nd Author's Name | Yuta Hayakawa |
| 2nd Author's Affiliation | Tokyo Institute of Technology(Tokyo Tech) |
| 3rd Author's Name | Takuya Hino |
| 3rd Author's Affiliation | Komatsu Ltd.(KOMATSU) |
| 4th Author's Name | Motohide Sugihara |
| 4th Author's Affiliation | Komatsu Ltd.(KOMATSU) |
| 5th Author's Name | Hiroki Ikeya |
| 5th Author's Affiliation | Komatsu Ltd.(KOMATSU) |
| 6th Author's Name | Masamichi Shimosaka |
| 6th Author's Affiliation | Tokyo Institute of Technology(Tokyo Tech) |
| Date | 2018-12-14 |
| Paper # | PRMU2018-94 |
| Volume (vol) | vol.118 |
| Number (no) | PRMU-362 |
| Page | pp.pp.103-108(PRMU), |
| #Pages | 6 |
| Date of Issue | 2018-12-06 (PRMU) |