期刊
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
卷 19, 期 1, 页码 144-152出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TII.2022.3165875
关键词
Sensors; Deep learning; Feature extraction; Convolutional neural networks; Activity recognition; Residual neural networks; Informatics; Deep learning (DL); human activity recognition; Industry 5; 0; Internet of Things (IoT); recurrent neural network (RNN); wearable sensors
This study developed a deep learning architecture based on a multi-level residual network for human activity recognition, which improved recognition accuracy through model integration and feature extraction, and was evaluated and compared on multiple datasets, showing significant performance.
Human activity recognition (HAR) applications have received much attention due to their necessary implementations in various domains, including Industry 5.0 applications such as smart homes, e-health, and various Internet of Things applications. Deep learning (DL) techniques have shown impressive performance in different classification tasks, including HAR. Accordingly, in this article, we develop a comprehensive HAR system based on a novel DL architecture called Multi-ResAtt (multilevel residual network with attention). This model incorporates initial blocks and residual modules aligned in parallel. Multi-ResAtt learns data representations on the inertial measurement units level. Multi-ResAtt integrates a recurrent neural network with attention to extract time-series features and perform activity recognition. We consider complex human activities collected from wearable sensors to evaluate the Multi-ResAtt using three public datasets, Opportunity; UniMiB-SHAR; and PAMAP2. Additionally, we compared the proposed Multi-ResAtt to several DL models and existing HAR systems, and it achieved significant performance.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据