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 Experiment Video

Updated: Jun 27, 2026

Artificial Intelligence-Based System for Detecting Attention Levels in Students
06:37

Artificial Intelligence-Based System for Detecting Attention Levels in Students

Published on: December 15, 2023

A Camera-Based Visual Sensor Pipeline for Fine-Grained Human Activity Recognition in Classroom Scenes.

Cheng Sun1, Danning Wu1, Zihao Wu1

  • 1School of Computer Science, Changsha University of Science and Technology, Changsha 410114, China.

Sensors (Basel, Switzerland)
|June 26, 2026
PubMed
Summary

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

A Novel Evaluation Strategy to Artificial Neural Network Model Based on Bionics.

Journal of bionic engineering·2021
See all related articles

This study introduces RepYOLOv5-SF3D for accurate student behavior recognition in classrooms. The novel framework enhances intelligent education systems by improving detection in complex scenes.

Area of Science:

  • Computer Vision
  • Artificial Intelligence
  • Educational Technology

Background:

  • Student behavior recognition is crucial for assessing teaching quality and advancing intelligent education.
  • Challenges include dense student populations, occlusions, scale variations, and subtle actions in classrooms.

Purpose of the Study:

  • To propose RepYOLOv5-SF3D, a cascaded visual perception framework for fine-grained student behavior recognition in complex classroom settings.
  • To enhance robustness and accuracy in challenging visual conditions.

Main Methods:

  • Integrated a lightweight RepYOLOv5m detector with a dual-stream SlowFast-3D recognition branch.
  • Employed a decoupled training strategy and introduced Spatiotemporal Depthwise-Separable 3D (SDS3D) and Normalization-Based Temporal Attention Mechanism (NTAM) modules.
Keywords:
classroom video analysisfine-grained action recognitionspatiotemporal modelingstudent behavior recognitionvisual sensing

Related Experiment Videos

Last Updated: Jun 27, 2026

Artificial Intelligence-Based System for Detecting Attention Levels in Students
06:37

Artificial Intelligence-Based System for Detecting Attention Levels in Students

Published on: December 15, 2023

  • Utilized a cascaded visual perception framework for automated inference from video to behavior labels.

    • Main Results:

    • Achieved a mean average precision (mAP) of 88.83% on a real classroom dataset.
    • Outperformed baseline SlowFast by 3.36% and LSTC by 2.05%.
    • Maintained a low front-end inference latency of 12.5 ms/frame and a model size of 151.46 MB.

    Conclusions:

    • RepYOLOv5-SF3D demonstrates a strong balance between fine-grained recognition accuracy and edge-deployment efficiency.
    • The framework is suitable for practical classroom visual sensing applications.
    • Offers improved performance for intelligent education systems in complex environments.