Exercise Recognition System using Facial Image Information
加藤花歩,Chengshuo Xia,杉浦裕太
Kaho Kato, Chengshuo Xia, Yuta Sugiura

[Reference /引用はこちら]
Kaho Kato, Chengshuo Xia, Yuta Sugiura, Exercise Recognition System using Facial Image Information from a Mobile Device, The 2021 IEEE 2nd Global Conference on Life Sciences and Technologies (LifeTech 2021), IEEE, March 9-11 2021, Nara, Japan. [DOI]

健康寿命の延伸に向けて日常的な運動が重要であるが、運動の継続が困難な人は多い。本稿では手軽な運動管理や運動ゲーム作成のための、モバイル端末内蔵カメラを利用した顔情報による運動識別システムを提案する。具体的には、カメラで運動中のユーザの顔画像を取得し、顔上の追跡点を抽出する。追跡点より算出した特徴量をサポートベクタマシンで学習することで9 種類の運動を識別した。ユーザごとの精度検証の結果、平均識別精度は97.2%となった。

Daily exercise has played a significant role for people to stay healthy, however, some people cannot do moderate exercise continuously. In this paper, we propose an exercise recognition system using facial image information for making exercise management convenient. The proposed system gets facial image information from a built-in camera on a mobile device, can recognize and classify multiple kinds of exercises. When a user exercises in a condition where the face is within an angle of view, the system extracts features from the facial images. Via the evaluation experiment with the data of a user, the average classification accuracy can reach up to 97.2%. To improve the operation of the designed system, we also evaluated the suitable window size, the dimension reduction, and the influence of the user’s standing position.