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

Versatile Synthesis of Mesoporous Nanomaterials via Monomicelle-Directed Assembly.

Advanced materials (Deerfield Beach, Fla.)·2026
Same author

The Levels, Prevalence, and Related Factors of Toxic Leadership Among Nurses Outside the United States: A Systematic Review and Meta-Analysis.

The Journal of nursing administration·2026
Same author

Nursing Students' Pain Management Knowledge and Attitudes: A Systematic Review and Meta-Analysis.

Pain management nursing : official journal of the American Society of Pain Management Nurses·2026
Same author

Letter to the Editor: proposal of a novel model for identifying complete response and nonoperative management in DMMR colon cancer following neoadjuvant immunotherapy: a retrospective cohort study.

International journal of surgery (London, England)·2025
Same author

Machine learning for predicting burnout among healthcare workers: a systematic review and meta-analysis.

Contemporary nurse·2025
Same author

Moral resilience among nurses: A systematic review and meta-analysis.

Nursing ethics·2025

Related Experiment Video

Updated: Jun 21, 2025

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis
05:41

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis

Published on: February 6, 2020

9.4K

Abnormal Behavior Recognition Based on 3D Dense Connections.

Wei Chen1, Zhanhe Yu2, Chaochao Yang1

  • 1School of Electrical and Control Engineering, North China University of Technology, Beijing 100144, P. R. China.

International Journal of Neural Systems
|July 16, 2024
PubMed
Summary

This study introduces a novel abnormal behavior recognition algorithm using 3D dense connections for enhanced real-time detection. The method improves urban safety by quickly identifying deviations from normal patterns with high accuracy and efficiency.

Keywords:
Abnormal behavior recognitionGRUadaptive soft thresholddense connectionmulti-instance learning

More Related Videos

Measuring Connectivity in the Primary Visual Pathway in Human Albinism Using Diffusion Tensor Imaging and Tractography
13:26

Measuring Connectivity in the Primary Visual Pathway in Human Albinism Using Diffusion Tensor Imaging and Tractography

Published on: August 11, 2016

12.3K
Automatic Identification of Dendritic Branches and their Orientation
06:08

Automatic Identification of Dendritic Branches and their Orientation

Published on: September 17, 2021

1.9K

Related Experiment Videos

Last Updated: Jun 21, 2025

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis
05:41

A Step-by-Step Implementation of DeepBehavior, Deep Learning Toolbox for Automated Behavior Analysis

Published on: February 6, 2020

9.4K
Measuring Connectivity in the Primary Visual Pathway in Human Albinism Using Diffusion Tensor Imaging and Tractography
13:26

Measuring Connectivity in the Primary Visual Pathway in Human Albinism Using Diffusion Tensor Imaging and Tractography

Published on: August 11, 2016

12.3K
Automatic Identification of Dendritic Branches and their Orientation
06:08

Automatic Identification of Dendritic Branches and their Orientation

Published on: September 17, 2021

1.9K

Area of Science:

  • Computer Science
  • Artificial Intelligence
  • Machine Learning

Background:

  • Abnormal behavior recognition is vital for security and fraud detection.
  • Deep Convolution Networks (ConvNets) show promise but lack real-time capabilities.
  • Existing methods prioritize accuracy over speed, hindering immediate threat identification.

Purpose of the Study:

  • To develop a real-time abnormal behavior recognition algorithm.
  • To enhance urban public safety through rapid detection of anomalous activities.
  • To improve upon existing ConvNet-based methods by addressing computational efficiency.

Main Methods:

  • Proposed an algorithm based on three-dimensional (3D) dense connections.
  • Utilized a multi-instance learning strategy for classifying diverse abnormal behaviors.
  • Employed dense connection modules and soft-threshold attention mechanisms to optimize model parameters and computational efficiency.
  • Implemented attention allocation to reduce redundant sequential information.

Main Results:

  • Achieved a recognition accuracy of 95.61% on the UCF-crime dataset.
  • Demonstrated strong performance in both recognition accuracy and processing speed compared to existing models.
  • Successfully reduced model parameter count and enhanced network computational efficiency.

Conclusions:

  • The proposed 3D dense connection algorithm offers an effective solution for real-time abnormal behavior recognition.
  • The method balances high accuracy with improved computational efficiency, crucial for public safety applications.
  • Attention mechanisms and dense connections contribute to mitigating performance impacts from redundant data.