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: May 9, 2026

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
03:31

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

Published on: December 15, 2023

MBDANet: a multi-scale balanced dynamic alignment detection network based on knowledge distillation for small

Fengdong Shi1, Xin Zhao2, Liankun Sun3

  • 1National Demonstration Center for Experimental Engineering Training Education, Tiangong University, Tianjin, 300387, China.

Scientific Reports
|May 7, 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

Fighting Evolving Spam With ARTMAP Models: A Noise-Resilient Online Detection Framework.

IEEE transactions on neural networks and learning systems·2026
Same author

Reshaping of extracellular polymeric substance scaffolds, microbiota and metabolism mitigates bio-clogging in microbial fuel cell-coupled constructed wetlands.

Bioresource technology·2026
Same author

From electron supply to electron economy: a unified framework for N<sub>2</sub>O control in membrane aerated biofilm reactors.

Water research·2026
Same author

A Pheromone-Dominant Odorant-Binding Protein Mediates Mating and Host Volatile Perception in <i>Anomala corpulenta</i>.

Journal of agricultural and food chemistry·2026
Same author

Covalent Organic Framework Mediated Solvation Regulation Reinforces Anion-Derived Interface in Quasi-Solid-State Electrolytes for High-Performance Silicon-Based Lithium-Ion Batteries.

ChemSusChem·2026
Same author

Balancing electron competition in nitrogen oxide reduction via membrane-aerated electroactive biofilms for electron flow regulation and N<sub>2</sub>O mitigation.

Water research·2026

This study introduces MBDANet, a novel network for detecting small object defects. It enhances accuracy and reduces model size and complexity, improving industrial manufacturing quality control.

Area of Science:

  • Computer Vision
  • Machine Learning
  • Industrial Automation

Background:

  • Small object defect detection is crucial for industrial quality control, but current methods struggle with accuracy, model size, and computational complexity.
  • Existing techniques often face a trade-off dilemma, limiting their effectiveness in real-world manufacturing scenarios.

Purpose of the Study:

  • To develop an advanced deep learning model for improved small object defect detection.
  • To address the limitations of existing methods by balancing detection accuracy, computational complexity, and model size.

Main Methods:

  • Proposes the Multi-Scale Balanced Dynamic Alignment Detection Network (MBDANet), an enhanced YOLOv8n architecture.
  • Incorporates a robust feature downsampling module, multi-scale feature fusion, and a task dynamic alignment detection head.
Keywords:
Defect detectionKnowledge distillationPrinted circuit boardSteel surface defectYOLOv8n

Related Experiment Videos

Last Updated: May 9, 2026

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
03:31

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

Published on: December 15, 2023

  • Utilizes a knowledge distillation method with a teacher model to further boost detection accuracy.

    • Main Results:

    • MBDANet achieved significant improvements in mean Average Precision (mAP) on benchmark datasets (PCB, DeepPCB, Steel Surface).
    • Demonstrated a 25% reduction in model size, 27% fewer parameters, and 15% lower computational complexity compared to the baseline.
    • Effectively enhances attention to small objects, improving detection performance.

    Conclusions:

    • MBDANet offers a superior solution for small object defect detection in industrial settings.
    • The network effectively balances detection speed and accuracy, providing robust technical support for manufacturing equipment.
    • This research contributes to proactive defect prevention and stable industrial operations.