講演名 2020-03-16
Egocentric pedestrian motion prediction by separately modeling body pose and position
Donghao Wu(the Univ. of Tokyo), Takuma Yagi(the Univ. of Tokyo), Yusuke Matsui(the Univ. of Tokyo), Yoichi Sato(the Univ. of Tokyo),
PDFダウンロードページ PDFダウンロードページへ
抄録(和) 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.
抄録(英) 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.
キーワード(和) motion forecasting / egocentric vision / human dynamics / deep learning / neural network
キーワード(英) motion forecasting / egocentric vision / human dynamics / deep learning / neural network
資料番号 PRMU2019-72
発行日 2020-03-09 (PRMU)

研究会情報
研究会 PRMU / IPSJ-CVIM
開催期間 2020/3/16(から2日開催)
開催地(和) 京都大学
開催地(英)
テーマ(和) 安全安心、セキュリティ・防災
テーマ(英)
委員長氏名(和) 佐藤 洋一(東大)
委員長氏名(英) Yoichi Sato(Univ. of Tokyo)
副委員長氏名(和) 玉木 徹(広島大) / 木村 昭悟(NTT)
副委員長氏名(英) Toru Tamaki(Hiroshima Univ.) / Akisato Kimura(NTT)
幹事氏名(和) 入江 豪(NTT) / 牛久 祥孝(オムロンサイニックエックス)
幹事氏名(英) Go Irie(NTT) / Yoshitaka Ushiku(OMRON SINICX)
幹事補佐氏名(和) 内田 祐介(DeNA) / 山下 隆義(中部大)
幹事補佐氏名(英) Yusuke Uchida(DeNA) / Takayoshi Yamashita(Chubu Univ.)

講演論文情報詳細
申込み研究会 Technical Committee on Pattern Recognition and Media Understanding / Special Interest Group on Computer Vision and Image Media
本文の言語 ENG
タイトル(和)
サブタイトル(和)
タイトル(英) Egocentric pedestrian motion prediction by separately modeling body pose and position
サブタイトル(和)
キーワード(1)(和/英) motion forecasting / motion forecasting
キーワード(2)(和/英) egocentric vision / egocentric vision
キーワード(3)(和/英) human dynamics / human dynamics
キーワード(4)(和/英) deep learning / deep learning
キーワード(5)(和/英) neural network / neural network
第 1 著者 氏名(和/英) Donghao Wu / Donghao Wu
第 1 著者 所属(和/英) The University of Tokyo(略称:the Univ. of Tokyo)
The University of Tokyo(略称:the Univ. of Tokyo)
第 2 著者 氏名(和/英) Takuma Yagi / Takuma Yagi
第 2 著者 所属(和/英) The University of Tokyo(略称:the Univ. of Tokyo)
The University of Tokyo(略称:the Univ. of Tokyo)
第 3 著者 氏名(和/英) Yusuke Matsui / Yusuke Matsui
第 3 著者 所属(和/英) The University of Tokyo(略称:the Univ. of Tokyo)
The University of Tokyo(略称:the Univ. of Tokyo)
第 4 著者 氏名(和/英) Yoichi Sato / Yoichi Sato
第 4 著者 所属(和/英) The University of Tokyo(略称:the Univ. of Tokyo)
The University of Tokyo(略称:the Univ. of Tokyo)
発表年月日 2020-03-16
資料番号 PRMU2019-72
巻番号(vol) vol.119
号番号(no) PRMU-481
ページ範囲 pp.39-44(PRMU),
ページ数 6
発行日 2020-03-09 (PRMU)