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
L'analyse vidéo de volleyball joue un rôle important en fournissant des données pour les contenus télévisés et en développant des stratégies. Parmi tous les sujets d'analyse du volleyball, la reconnaissance qualitative des actions des joueurs est essentielle car elle fournit potentiellement non seulement l'action exécutée, mais également la qualité, c'est-à-dire la qualité de l'action. Cependant, la plupart des recherches sur la reconnaissance d’actions se concentrent sur la discrimination entre différentes actions. Jusqu'à présent, la qualité d'une action, utile à l'évaluation et à l'entraînement des compétences du joueur, n'a reçu que peu d'attention. Les problèmes vitaux dans la reconnaissance qualitative des actions comprennent l'occlusion, les petites différences inter-classes et divers types d'apparence provoqués par le changement de joueur. Cet article propose un cadre de reconnaissance basé sur une combinaison de caractéristiques locales globales et multi-vues 3D avec une fonctionnalité de formation d'équipe globale, une fonctionnalité d'état de balle et des fonctionnalités de pose abrupte. Les problèmes ci-dessus sont résolus par la combinaison de fonctionnalités globales 3D (qui masquent la fonctionnalité de mouvement 2D instable et incomplète causée par l'occlusion) et de fonctionnalités locales multi-vues (qui obtiennent des fonctionnalités de mouvement local détaillées des parties du corps dans plusieurs points de vue). Premièrement, la formation de l'équipe extrait les trajectoires 3D de l'ensemble des membres de l'équipe plutôt que d'un seul joueur cible. Cette proposition se concentre davantage sur l’ensemble de la fonctionnalité tout en éliminant l’effet personnel. Deuxièmement, la fonction d’état de mouvement de la balle extrait les caractéristiques de la trajectoire de la balle en 3D. Le mouvement de la balle n'est pas affecté par l'apparence personnelle, donc cette proposition ignore l'influence de l'apparence du joueur et la rend plus robuste au changement de joueur cible. Enfin, la fonction de pose abrupte se compose de deux parties : la pose du cadre de frappe abrupte (qui extrait la forme du contour de la pose du joueur au moment du coup) et la variation de pose abrupte (qui extrait la variation de pose entre la pose de préparation et la pose de fin pendant l'action). Ces deux caractéristiques permettent de mieux distinguer la qualité de chaque action en se concentrant sur le standard de mouvement et la stabilité entre les différentes actions de qualité. Des expériences sont menées sur des vidéos de jeux de la demi-finale et de la finale des Jeux inter-lycées japonais de volleyball masculin 2014 au Tokyo Metropolitan Gymnasium. Les résultats expérimentaux montrent que la précision atteint 97.26 %, une amélioration de 11.33 % pour la discrimination des actions et de 91.76 %, et une amélioration de 13.72 % pour l'évaluation de la qualité des actions.
Xina CHENG
Xidian University,Waseda University
Yang LIU
Waseda University
Takeshi IKENAGA
Waseda 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
Xina CHENG, Yang LIU, Takeshi IKENAGA, "3D Global and Multi-View Local Features Combination Based Qualitative Action Recognition for Volleyball Game Analysis" in IEICE TRANSACTIONS on Fundamentals,
vol. E102-A, no. 12, pp. 1891-1899, December 2019, doi: 10.1587/transfun.E102.A.1891.
Abstract: Volleyball video analysis plays important roles in providing data for TV contents and developing strategies. Among all the topics of volleyball analysis, qualitative player action recognition is essential because it potentially provides not only the action that being performed but also the quality, which means how well the action is performed. However, most action recognition researches focus on the discrimination between different actions. The quality of an action, which is helpful for evaluation and training of the player skill, has only received little attention so far. The vital problems in qualitative action recognition include occlusion, small inter-class difference and various kinds of appearance caused by the player change. This paper proposes a 3D global and multi-view local features combination based recognition framework with global team formation feature, ball state feature and abrupt pose features. The above problems are solved by the combination of 3D global features (which hide the unstable and incomplete 2D motion feature caused by occlusion) and the multi-view local features (which get detailed local motion features of body parts in multiple viewpoints). Firstly, the team formation extracts the 3D trajectories from the whole team members rather than a single target player. This proposal focuses more on the entire feature while eliminating the personal effect. Secondly, the ball motion state feature extracts features from the 3D ball trajectory. The ball motion is not affected by the personal appearance, so this proposal ignores the influence of the players appearance and makes it more robust to target player change. At last, the abrupt pose feature consists of two parts: the abrupt hit frame pose (which extracts the contour shape of the player's pose at the hit time) and abrupt pose variation (which extracts the pose variation between the preparation pose and ending pose during the action). These two features make difference of each action quality more distinguishable by focusing on the motion standard and stability between different quality actions. Experiments are conducted on game videos from the Semifinal and Final Game of 2014 Japan Inter High School Games of Men's Volleyball in Tokyo Metropolitan Gymnasium. The experimental results show the accuracy achieves 97.26%, improving 11.33% for action discrimination and 91.76%, and improving 13.72% for action quality evaluation.
URL: https://global.ieice.org/en_transactions/fundamentals/10.1587/transfun.E102.A.1891/_p
Copier
@ARTICLE{e102-a_12_1891,
author={Xina CHENG, Yang LIU, Takeshi IKENAGA, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={3D Global and Multi-View Local Features Combination Based Qualitative Action Recognition for Volleyball Game Analysis},
year={2019},
volume={E102-A},
number={12},
pages={1891-1899},
abstract={Volleyball video analysis plays important roles in providing data for TV contents and developing strategies. Among all the topics of volleyball analysis, qualitative player action recognition is essential because it potentially provides not only the action that being performed but also the quality, which means how well the action is performed. However, most action recognition researches focus on the discrimination between different actions. The quality of an action, which is helpful for evaluation and training of the player skill, has only received little attention so far. The vital problems in qualitative action recognition include occlusion, small inter-class difference and various kinds of appearance caused by the player change. This paper proposes a 3D global and multi-view local features combination based recognition framework with global team formation feature, ball state feature and abrupt pose features. The above problems are solved by the combination of 3D global features (which hide the unstable and incomplete 2D motion feature caused by occlusion) and the multi-view local features (which get detailed local motion features of body parts in multiple viewpoints). Firstly, the team formation extracts the 3D trajectories from the whole team members rather than a single target player. This proposal focuses more on the entire feature while eliminating the personal effect. Secondly, the ball motion state feature extracts features from the 3D ball trajectory. The ball motion is not affected by the personal appearance, so this proposal ignores the influence of the players appearance and makes it more robust to target player change. At last, the abrupt pose feature consists of two parts: the abrupt hit frame pose (which extracts the contour shape of the player's pose at the hit time) and abrupt pose variation (which extracts the pose variation between the preparation pose and ending pose during the action). These two features make difference of each action quality more distinguishable by focusing on the motion standard and stability between different quality actions. Experiments are conducted on game videos from the Semifinal and Final Game of 2014 Japan Inter High School Games of Men's Volleyball in Tokyo Metropolitan Gymnasium. The experimental results show the accuracy achieves 97.26%, improving 11.33% for action discrimination and 91.76%, and improving 13.72% for action quality evaluation.},
keywords={},
doi={10.1587/transfun.E102.A.1891},
ISSN={1745-1337},
month={December},}
Copier
TY - JOUR
TI - 3D Global and Multi-View Local Features Combination Based Qualitative Action Recognition for Volleyball Game Analysis
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 1891
EP - 1899
AU - Xina CHENG
AU - Yang LIU
AU - Takeshi IKENAGA
PY - 2019
DO - 10.1587/transfun.E102.A.1891
JO - IEICE TRANSACTIONS on Fundamentals
SN - 1745-1337
VL - E102-A
IS - 12
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - December 2019
AB - Volleyball video analysis plays important roles in providing data for TV contents and developing strategies. Among all the topics of volleyball analysis, qualitative player action recognition is essential because it potentially provides not only the action that being performed but also the quality, which means how well the action is performed. However, most action recognition researches focus on the discrimination between different actions. The quality of an action, which is helpful for evaluation and training of the player skill, has only received little attention so far. The vital problems in qualitative action recognition include occlusion, small inter-class difference and various kinds of appearance caused by the player change. This paper proposes a 3D global and multi-view local features combination based recognition framework with global team formation feature, ball state feature and abrupt pose features. The above problems are solved by the combination of 3D global features (which hide the unstable and incomplete 2D motion feature caused by occlusion) and the multi-view local features (which get detailed local motion features of body parts in multiple viewpoints). Firstly, the team formation extracts the 3D trajectories from the whole team members rather than a single target player. This proposal focuses more on the entire feature while eliminating the personal effect. Secondly, the ball motion state feature extracts features from the 3D ball trajectory. The ball motion is not affected by the personal appearance, so this proposal ignores the influence of the players appearance and makes it more robust to target player change. At last, the abrupt pose feature consists of two parts: the abrupt hit frame pose (which extracts the contour shape of the player's pose at the hit time) and abrupt pose variation (which extracts the pose variation between the preparation pose and ending pose during the action). These two features make difference of each action quality more distinguishable by focusing on the motion standard and stability between different quality actions. Experiments are conducted on game videos from the Semifinal and Final Game of 2014 Japan Inter High School Games of Men's Volleyball in Tokyo Metropolitan Gymnasium. The experimental results show the accuracy achieves 97.26%, improving 11.33% for action discrimination and 91.76%, and improving 13.72% for action quality evaluation.
ER -