Human motion recognition from 3D pose information - Trisarea: a new pose-based feature
Author
Vinagre, M.; Aranda, J.; Casals, A.
Type of activity
Presentation of work at congresses
Name of edition
10th Internacional Conference on Informatics in Control, Automation and Robotics
Date of publication
2013
Presentation's date
2013-07-30
Book of congress proceedings
ICINCO 2013 - Proceedings of the 10th International Conference on Informatics in Control, Automation and Robotics, Volume 1, Reykjavík, Iceland, 29 - 31 July, 2013
First page
74
Last page
82
Abstract
The use of pose-based features has demonstrated to be a promising approach for human motion recognition. Encouraged by the results achieved, a new relational pose-based feature, Trisarea, based on geometric relationship between human joints, is proposed and analysed. This feature is defined as the area of the triangle formed by connecting three joints. The paper shows how the variation of a selected set of Trisarea features over time constitutes a descriptor of human motion. It also demonstrates...
The use of pose-based features has demonstrated to be a promising approach for human motion recognition. Encouraged by the results achieved, a new relational pose-based feature, Trisarea, based on geometric relationship between human joints, is proposed and analysed. This feature is defined as the area of the triangle formed by connecting three joints. The paper shows how the variation of a selected set of Trisarea features over time constitutes a descriptor of human motion. It also demonstrates how this motion descriptor based
on Trisarea features can provide useful information in terms of human motion for its application to action recognition tasks.