Scholarlink

Boxing, a popular yet high-risk sport, lacks comprehensive fatigue assessments crucial for performance analysis. Previous studies have used low-cost inertial sensors like accelerometers and electromyography to analyze performance under fatigue. This paper introduces a novel approach to fatigue classification using smartphone sensors to evaluate punch consistency through acceleration and hand speed, using time intervals between punches. A framework is established to model action patterns from IMU data and assess various deep learning models, including CNNs, LSTMs, bi-LSTMs, attention-based LSTMs, and CNN-LSTM combinations. The CNN-LSTM model achieved 99% accuracy in classifying fatigue levels, demonstrating potential for further optimization and integration in wearable devices.

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

Evaluating the Usage of Smartphone Sensor to Evaluate Fatigue in Boxing in Real-Time Training Using Deep-Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • J. Brindha,
  • G. Nallavan,
  • L. Balaji

摘要

Boxing, a popular yet high-risk sport, lacks comprehensive fatigue assessments crucial for performance analysis. Previous studies have used low-cost inertial sensors like accelerometers and electromyography to analyze performance under fatigue. This paper introduces a novel approach to fatigue classification using smartphone sensors to evaluate punch consistency through acceleration and hand speed, using time intervals between punches. A framework is established to model action patterns from IMU data and assess various deep learning models, including CNNs, LSTMs, bi-LSTMs, attention-based LSTMs, and CNN-LSTM combinations. The CNN-LSTM model achieved 99% accuracy in classifying fatigue levels, demonstrating potential for further optimization and integration in wearable devices.