Scholarlink

<p>This research investigates human activity recognition (HAR) by proposing and evaluating a multi-resolution CNN-LSTM architecture for analyzing sensor data. The distinctive feature of this architecture is the integration of CNN and LSTM components, enabling comprehensive extraction of both spatial and temporal features. The research assesses the model’s performance across multiple datasets, demonstrating its robustness in diverse real-world scenarios. Results indicate high accuracy levels: 88% for HARTH, 92% for HAR70+, 97% for UCI-Sensor and 98% for Opportunity datasets. The inclusion of multi-resolution elements allows the model to capture complex activity patterns effectively, highlighting its capability to understand intricate human behavior. Overall, this research advances HAR methodologies by demonstrating the potential of multi-resolution CNN-LSTM architectures for precise and adaptable recognition of human activities using sensor data.</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

A multi-resolution CNN-LSTM architecture for sensor-based human activity recognition

  • 期刊论文

  • 发表日期：

    2026年2月19日

查看全文
  • Gunjan Pareek,
  • Rajiv Singh,
  • Swati Nigam

摘要

This research investigates human activity recognition (HAR) by proposing and evaluating a multi-resolution CNN-LSTM architecture for analyzing sensor data. The distinctive feature of this architecture is the integration of CNN and LSTM components, enabling comprehensive extraction of both spatial and temporal features. The research assesses the model’s performance across multiple datasets, demonstrating its robustness in diverse real-world scenarios. Results indicate high accuracy levels: 88% for HARTH, 92% for HAR70+, 97% for UCI-Sensor and 98% for Opportunity datasets. The inclusion of multi-resolution elements allows the model to capture complex activity patterns effectively, highlighting its capability to understand intricate human behavior. Overall, this research advances HAR methodologies by demonstrating the potential of multi-resolution CNN-LSTM architectures for precise and adaptable recognition of human activities using sensor data.