摘要
通过人工智能技术与分布式技术的结合,提出了一个基于姿态估计算法的分布式动作教学平台设计方案,能够应用于教学太极拳、广播体操、健身动作等领域,为用户在无人指导的环境下提供智能的动作教练。该平台采用前后端分离开发,利用分布式与集群技术,解决高并发与可用性问题。平台通过Tensorflow.js框架,在前端实时进行动作识别、动作分析、动作指导,避免了前后端数据交互中容易产生的隐私泄露以及服务器负载过大等问题。通过评分及对比算法,使得用户在使用过程中不断纠正自己的动作,达到动作学习的正反馈效果。
Through the combination of artificial intelligence technology and distributed technology,a design scheme of distributed action teaching platform based on pose estimation algorithm is proposed,which can be applied to teaching Tai Chi,radio gymnastics,fitness movement,and other fields,and provides users with an unsupervised environment.Provide intelligent action coaches.The platform uses front⁃end and back⁃end separation development and uses distributed and cluster technologies to solve high concurrency and availability issues.The platform uses the Tensorflow.js framework to perform motion recognition,motion analysis,and motion guidance on the front⁃end in real⁃time,avoiding problems such as privacy leaks and server overload that are prone to occur in front⁃end and back⁃end data interactions.Through the scoring and comparison algorithms,users can continuously correct their actions during use and achieve positive feedback on action learning.
作者
段诩
张文辉
DUAN Xu;ZHANG Wenhui(School of Computer and Information Security,Guilin University of Electronic Technology,Guilin 541004,China)
出处
《电子设计工程》
2020年第23期63-67,72,共6页
Electronic Design Engineering
基金
桂林电子科技大学国家级大学生创新训练项目(201810595037)。
关键词
分布式架构
动作识别
动作教学
深度学习
distributed architecture
action recognition
action teaching
deep learning
