Scholarlink

This paper introduces a data glove prototype equipped with inertial measurement units (IMUs) for real-time resistance classification. Three pins with different resistance levels were used to evaluate the proposed system. A shallow one-dimensional convolutional neural network was employed to process the raw data from the integrated sensors. The network was deployed on a low-cost, constrained device. The system achieved a real-time inference classification accuracy of 92.22% with a latency of 19.70 ms while consuming 1,397.76 mJ of energy. Overall, the proposed system is a promising candidate for various biomedical applications requiring real-time inference, such as stroke rehabilitation.

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

Data Glove System: Real-Time Resistance Classification with Embedded Machine Learning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Mohamad Yaacoub,
  • Matthias Vanlaeres,
  • Daniella Shebly,
  • Maurizio Valle,
  • Christian Gianoglio

摘要

This paper introduces a data glove prototype equipped with inertial measurement units (IMUs) for real-time resistance classification. Three pins with different resistance levels were used to evaluate the proposed system. A shallow one-dimensional convolutional neural network was employed to process the raw data from the integrated sensors. The network was deployed on a low-cost, constrained device. The system achieved a real-time inference classification accuracy of 92.22% with a latency of 19.70 ms while consuming 1,397.76 mJ of energy. Overall, the proposed system is a promising candidate for various biomedical applications requiring real-time inference, such as stroke rehabilitation.