Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Dual-Stream Architecture Enhanced by Soft-Attention Mechanism for Plant Species Classification.

Plants (Basel, Switzerland)·2024
Same author

Visual Intelligence in Precision Agriculture: Exploring Plant Disease Detection via Efficient Vision Transformers.

Sensors (Basel, Switzerland)·2023
Same author

Human Activity Recognition via Hybrid Deep Learning Based Model.

Sensors (Basel, Switzerland)·2022
Same author

Adaptive Node Clustering for Underwater Sensor Networks.

Sensors (Basel, Switzerland)·2021
Same author

DLNR-SIQA: Deep Learning-Based No-Reference Stitched Image Quality Assessment.

Sensors (Basel, Switzerland)·2020
Same author

Automatic 360° Mono-Stereo Panorama Generation Using a Cost-Effective Multi-Camera System.

Sensors (Basel, Switzerland)·2020

Related Experiment Video

Updated: Jun 29, 2025

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

3.7K

PAR-Net: An Enhanced Dual-Stream CNN-ESN Architecture for Human Physical Activity Recognition.

Imran Ullah Khan1, Jong Weon Lee1

  • 1Mixed Reality and Interaction Lab, Department of Software, Sejong University, Seoul 05006, Republic of Korea.

Sensors (Basel, Switzerland)
|March 28, 2024
PubMed
Summary

This study introduces PAR-Net, an AI model for recognizing physical activities. It effectively captures complex spatiotemporal patterns, improving accuracy in human activity recognition.

Keywords:
deep learningecho state networksmachine learningphysical activity recognitionskeleton data

More Related Videos

Evaluation of a Smartphone-based Human Activity Recognition System in a Daily Living Environment
06:49

Evaluation of a Smartphone-based Human Activity Recognition System in a Daily Living Environment

Published on: December 11, 2015

8.9K
Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
04:48

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

395

Related Experiment Videos

Last Updated: Jun 29, 2025

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

3.7K
Evaluation of a Smartphone-based Human Activity Recognition System in a Daily Living Environment
06:49

Evaluation of a Smartphone-based Human Activity Recognition System in a Daily Living Environment

Published on: December 11, 2015

8.9K
Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique
04:48

Swin-PSAxialNet: An Efficient Multi-Organ Segmentation Technique

Published on: July 5, 2024

395

Area of Science:

  • Computer Science
  • Artificial Intelligence
  • Biomedical Engineering

Background:

  • AI-driven techniques are used for human physical activity recognition.
  • Existing methods struggle with temporal and spatial data features and complex activity patterns.
  • There is a need for advanced architectures to learn spatiotemporal dependencies for improved accuracy.

Purpose of the Study:

  • To develop an attention-enhanced dual-stream network (PAR-Net) for physical activity recognition.
  • To extract spatial and temporal features simultaneously for enhanced recognition accuracy.
  • To improve the comprehension of complex activity patterns over different periods.

Main Methods:

  • The proposed PAR-Net integrates Convolutional Neural Networks (CNNs) and Echo State Networks (ESNs).
  • A self-attention mechanism is employed for optimal feature selection and targeted attention on significant features.
  • A dual-stream feature extraction mechanism learns spatiotemporal dependencies from data.

Main Results:

  • PAR-Net achieved higher performance compared to baseline methods on two benchmark datasets.
  • The model demonstrated improved identification of nuanced activity patterns.
  • A thorough ablation study confirmed the optimal model configuration.

Conclusions:

  • PAR-Net effectively extracts spatial and temporal features for accurate physical activity recognition.
  • The attention-enhanced dual-stream network advances the field of human activity recognition.
  • The proposed architecture offers a promising solution for complex activity pattern analysis.