基于计算卸载的智能电视运动评测系统设计与实现

Design and Implementation of Motion Evaluation System Based on Computing Offloadinng

在家庭视频业务环境下,随着计算能力的提升和视频内容的网络化推进,用户对于智能电视的复杂交互需求逐渐提高。针对用户在家庭场景下家庭健身、康复等方面的运动行为评价,构建基于智能电视运动数据行为分析业务环境。在实现上,根据智能电视运动识别业务场景中视频采集设备的特点,采用基于OpenPose的神经网络运动估计方法。同时,针对智能电视计算能力不足的问题,在家庭环境下构建了以时延最小化为目标的计算卸载优化方法。通过应用的构建,丰富了用户基于智能电视的应用途径,增强了智能电视的功能。

In the home video business environment,with the improvement of computing power and the advancement of video content networking,users’complex interactive requirements for smart TV are gradually increasing.In this paper,for the user in the family scene,the user’s personal exercise behavior evaluation based on family fitness,rehabilitation and other aspects,build a business environment based on smart TV sports data behavior analysis.In the implementation,according to the characteristics of video acquisition equipment in the smart TV motion recognition business scene,the neural network motion estimation method based on openpose is adopted.At the same time,in order to solve the problem of insufficient computing power of smart TV,an optimization method of computing offloading with the goal of minimizing time delay is constructed in the home environment.Through the construction of the application,it enriches the user’s application way based on the smart TV,and enhances the function of the smart TV.