The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
La reconnaissance d'actions à l'aide de données de squelette (coordonnées 3D d'articulations humaines) est un sujet attrayant en raison de sa robustesse face à l'apparence de l'acteur, au point de vue de la caméra, à l'éclairage et à d'autres conditions environnementales. Cependant, les données squelettes doivent être mesurées par un capteur de profondeur ou extraites des données vidéo à l’aide d’un algorithme d’estimation, ce qui risque d’entraîner des erreurs d’extraction et du bruit. Dans ce travail, pour une reconnaissance d'action robuste basée sur un squelette, nous proposons un modèle d'espace d'état profond (DSSM). Le DSSM est un modèle génératif profond de la dynamique sous-jacente d'une séquence observable. Nous avons appliqué le DSSM proposé aux données squelettes et les résultats démontrent qu'il améliore les performances de classification d'une méthode de base. De plus, nous confirmons que l’extraction de caractéristiques avec le DSSM proposé rend les classifications ultérieures robustes au bruit et aux valeurs manquantes. Dans de tels contextes expérimentaux, le DSSM proposé surpasse une méthode de pointe.
Kazuki KAWAMURA
Kobe University
Takashi MATSUBARA
Kobe University
Kuniaki UEHARA
Kobe University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copier
Kazuki KAWAMURA, Takashi MATSUBARA, Kuniaki UEHARA, "Deep State-Space Model for Noise Tolerant Skeleton-Based Action Recognition" in IEICE TRANSACTIONS on Information,
vol. E103-D, no. 6, pp. 1217-1225, June 2020, doi: 10.1587/transinf.2019MVP0012.
Abstract: Action recognition using skeleton data (3D coordinates of human joints) is an attractive topic due to its robustness to the actor's appearance, camera's viewpoint, illumination, and other environmental conditions. However, skeleton data must be measured by a depth sensor or extracted from video data using an estimation algorithm, and doing so risks extraction errors and noise. In this work, for robust skeleton-based action recognition, we propose a deep state-space model (DSSM). The DSSM is a deep generative model of the underlying dynamics of an observable sequence. We applied the proposed DSSM to skeleton data, and the results demonstrate that it improves the classification performance of a baseline method. Moreover, we confirm that feature extraction with the proposed DSSM renders subsequent classifications robust to noise and missing values. In such experimental settings, the proposed DSSM outperforms a state-of-the-art method.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2019MVP0012/_p
Copier
@ARTICLE{e103-d_6_1217,
author={Kazuki KAWAMURA, Takashi MATSUBARA, Kuniaki UEHARA, },
journal={IEICE TRANSACTIONS on Information},
title={Deep State-Space Model for Noise Tolerant Skeleton-Based Action Recognition},
year={2020},
volume={E103-D},
number={6},
pages={1217-1225},
abstract={Action recognition using skeleton data (3D coordinates of human joints) is an attractive topic due to its robustness to the actor's appearance, camera's viewpoint, illumination, and other environmental conditions. However, skeleton data must be measured by a depth sensor or extracted from video data using an estimation algorithm, and doing so risks extraction errors and noise. In this work, for robust skeleton-based action recognition, we propose a deep state-space model (DSSM). The DSSM is a deep generative model of the underlying dynamics of an observable sequence. We applied the proposed DSSM to skeleton data, and the results demonstrate that it improves the classification performance of a baseline method. Moreover, we confirm that feature extraction with the proposed DSSM renders subsequent classifications robust to noise and missing values. In such experimental settings, the proposed DSSM outperforms a state-of-the-art method.},
keywords={},
doi={10.1587/transinf.2019MVP0012},
ISSN={1745-1361},
month={June},}
Copier
TY - JOUR
TI - Deep State-Space Model for Noise Tolerant Skeleton-Based Action Recognition
T2 - IEICE TRANSACTIONS on Information
SP - 1217
EP - 1225
AU - Kazuki KAWAMURA
AU - Takashi MATSUBARA
AU - Kuniaki UEHARA
PY - 2020
DO - 10.1587/transinf.2019MVP0012
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E103-D
IS - 6
JA - IEICE TRANSACTIONS on Information
Y1 - June 2020
AB - Action recognition using skeleton data (3D coordinates of human joints) is an attractive topic due to its robustness to the actor's appearance, camera's viewpoint, illumination, and other environmental conditions. However, skeleton data must be measured by a depth sensor or extracted from video data using an estimation algorithm, and doing so risks extraction errors and noise. In this work, for robust skeleton-based action recognition, we propose a deep state-space model (DSSM). The DSSM is a deep generative model of the underlying dynamics of an observable sequence. We applied the proposed DSSM to skeleton data, and the results demonstrate that it improves the classification performance of a baseline method. Moreover, we confirm that feature extraction with the proposed DSSM renders subsequent classifications robust to noise and missing values. In such experimental settings, the proposed DSSM outperforms a state-of-the-art method.
ER -