Scholarlink

<p>This article proposes a web-based integration framework for complex aerobics movements to improve training efficiency and movement recognition. Through big data analytics, the approach assesses kinematic features of aerobic movements gathered from large-scale multi-sensor networks. A feature classification mechanism is implemented to lead adaptive resource allocation and motion recording. Aided by Web Technology (WT), a Fuzzy Scheduling (FS) strategy is designed to achieve real-time adaptive information integration. A multi-level integration paradigm is then established to continue enhancing the synchronization and decomposition of compound aerobics routines. Simulation tests in MATLAB demonstrate that the new method significantly outperforms conventional methods in recognition rates, achieving a figure of 99.4% based on a dataset comprising 600 samples. The findings demonstrate improvements in Information Fusion (IF) performance, enhanced Feature Extraction (FE) capabilities, and greater efficacy in motion recognition and decomposition.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Web-based integration of high-difficulty aerobics movements for real-time human activity recognition using fuzzy logic and big data analytics

  • 期刊论文

  • 发表日期：

    2025年10月29日

查看全文
  • Zhaoxia Ji,
  • Mingsheng Xu,
  • Fei Zhang,
  • Ai Zhang,
  • Yonglin Zhao

摘要

This article proposes a web-based integration framework for complex aerobics movements to improve training efficiency and movement recognition. Through big data analytics, the approach assesses kinematic features of aerobic movements gathered from large-scale multi-sensor networks. A feature classification mechanism is implemented to lead adaptive resource allocation and motion recording. Aided by Web Technology (WT), a Fuzzy Scheduling (FS) strategy is designed to achieve real-time adaptive information integration. A multi-level integration paradigm is then established to continue enhancing the synchronization and decomposition of compound aerobics routines. Simulation tests in MATLAB demonstrate that the new method significantly outperforms conventional methods in recognition rates, achieving a figure of 99.4% based on a dataset comprising 600 samples. The findings demonstrate improvements in Information Fusion (IF) performance, enhanced Feature Extraction (FE) capabilities, and greater efficacy in motion recognition and decomposition.