ビデオからの3次元姿勢推定と機械学習を用いた行動認識の試み

Kohei Adachi, Tsuyoshi Okita, Sozo Inoue,
日本知能情報ファジィ学会九州支部学術講演会予稿集
(Not Available)
(Not Available)
40-43
2019-11-30
Iizuka
http://soft-kyushu.org/2019/conf/index.html
MotionCapture system achieved promising results in Human Activity Recognition(HAR). However, it was time consuming such as preparation and post processing. Thus, we thought use 3D estimation method as alternative one. However, some error is occurred in 2D and 3D estimation and it is possible to some affect in HAR. In this paper, we investigated method of HAR using 2D and 3D pose estimation. Then, we compared result of HAR using 2D and 3D estimation pose.

Data Files

  • スクリーンショット_2019-11-30_11.09.34.png