Scholarlink

This study presents a lightweight 1D-Convolutional Neural Network (1D-CNN) for object recognition on edge devices utilizing Inertial Measurement Unit (IMU) sensors positioned on the thumb and index finger to capture motion data. The proposed system achieves a classification accuracy of 91%, underscoring its effectiveness in distinguishing four object shapes. Optimized for embedded applications, the model provides an inference time of 110.1 ms and energy consumption of 2.27 mJ per inference, highlighting its efficiency and suitability for real-time deployment. These results demonstrate the potential of the developed solution for enabling intelligent and energy-efficient wearable systems.

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

Object Recognition with IMUs and Convolutional Neural Networks for Wearable Systems

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Daniella Shebly,
  • Christian Gianoglio,
  • Hussein Chibli,
  • Maurizio Valle

摘要

This study presents a lightweight 1D-Convolutional Neural Network (1D-CNN) for object recognition on edge devices utilizing Inertial Measurement Unit (IMU) sensors positioned on the thumb and index finger to capture motion data. The proposed system achieves a classification accuracy of 91%, underscoring its effectiveness in distinguishing four object shapes. Optimized for embedded applications, the model provides an inference time of 110.1 ms and energy consumption of 2.27 mJ per inference, highlighting its efficiency and suitability for real-time deployment. These results demonstrate the potential of the developed solution for enabling intelligent and energy-efficient wearable systems.