EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

AWARE-Net: A Lightweight Joint Optimization Framework for Robust Sensor-Based Human Activity Recognition

Pei He iD, Yuyan Wang, Pengxin Ren, Xiaodong Wang, Lishuai Xie, Yangming Guo

Sensors · 18.07.2026 · 26 (14) · 4566

DOI: 10.3390/s26144566 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 18.07.2026. Verfasst von Pei He und 5 weiteren Autor:innen. Beteiligte Einrichtung: School of Computer Science, Northwestern Polytechnical University, Xi'an 710072, China. Hauptthemen laut MEDLINE: Human Activities. Die Arbeit nennt 39 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Sensor-based human activity recognition (SHAR) serves as a core research direction in pervasive computing, mobile health, and related fields. Although existing deep learning methods have achieved promising progress in SHAR tasks, most optimize from a single dimension only. They struggle to simultaneously balance recognition accuracy, noise robustness, adaptation to class imbalance, and lightweight deployment requirements, leading to performance bottlenecks in real-world scenarios. To address these challenges, this paper proposes a lightweight joint optimization framework named AWARE-Net. Leveraging the lightweight TS-ResNet as a backbone encoder, the framework integrates spatiotemporal dynamic convolution feature encoding with a global loss function that fuses class-balanced loss, contrastive learning auxiliary loss, and temporal smooth regularization to achieve multi-objective joint optimization. Extensive experiments on three widely used SHAR benchmark datasets, namely OPPORTUNITY, PAMAP2, and USC-HAD, demonstrate that the proposed AWARE-Net achieves competitive performance compared with representative state-of-the-art HAR methods.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Human ActivitiesHumansAlgorithmsDeep LearningConvolutional Neural NetworksSoft Computing

Stichwörter der Autor:innen

class-balanced losscontrastive learninglightweight deep learningsensor-based human activity recognition (SHAR)total variation regularization

Autor:innen und Institutionen

  • School of Computer Science, Northwestern Polytechnical University, Xi'an 710072, China.
  • College of Computer and Information Science, Guangdong University of Technology, Guangzhou 510006, China.
  • Ningbo Institute of Northwestern Polytechnical University, Ningbo 315103, China.
  • Jiangsu Key Laboratory of Advanced Structural Materials and Application Technology, Nanjing Institute of Technology, Nanjing 211167, China.

ORCID-Profile: Pei He

Zitieren

APA

He, P., Wang, Y., Ren, P., Wang, X., Xie, L., & Guo, Y. (2026). AWARE-Net: A Lightweight Joint Optimization Framework for Robust Sensor-Based Human Activity Recognition. Sensors, 26(14), 4566. https://doi.org/10.3390/s26144566

Vancouver

He P, Wang Y, Ren P, Wang X, Xie L, Guo Y. AWARE-Net: A Lightweight Joint Optimization Framework for Robust Sensor-Based Human Activity Recognition. Sensors. 2026;26(14):4566. doi:10.3390/s26144566

BibTeX
@article{He2026,
  author = {He, Pei and Wang, Yuyan and Ren, Pengxin and Wang, Xiaodong and Xie, Lishuai and Guo, Yangming},
  title = {AWARE-Net: A Lightweight Joint Optimization Framework for Robust Sensor-Based Human Activity Recognition},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {14},
  pages = {4566},
  doi = {10.3390/s26144566}
}
Kostenlos starten →