Modeling the Behaviors of Participants in Meetings for Decision Making Using OpenPose
Modeling the Behaviors of Participants in Meetings for Decision Making Using OpenPose
复制标题
使用 OpenPose 对会议参与者的行为进行建模以进行决策
DOI：
10.1007/978-3-030-40274-7_3
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Takashi Ozeki and Takeshi Kohama
中科院分区：
文献类型：
--
作者：
Eiji Watanabe;Takashi Ozeki and Takeshi Kohama
本文讨论了决策会议中参与者行为之间的交互建模。首先，我们采用行为（即，眼睛，面部和手部运动）由OpenPose [9]检测为使用单个相机的骨架检测算法。其次，提出了一种基于神经网络的参与者行为建模方法。在此基础上，讨论了多层神经网络中参与者行为与模型参数之间的关系。值得注意的是，我们表明，参与者的特征表示的指数的基础上，上述模型的参数。
This paper discusses the modeling for interactions between the behaviors of participants in a meeting for decision making. First, we adopt the behaviors (i.e., eye, facial, and hand movements) detected by OpenPose [9] as a skeleton detection algorithm using a single camera. Second, we propose a modeling method for the participant behaviors based on neural networks. Furthermore, we discuss the relationships between the participant behaviors and the model parameters in multi-layered neural networks based on the experimental results. Notably, we show that the participant characteristics are represented by the indices based on the parameters of the abovementioned models.