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

Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography
11:34

Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography

Published on: May 15, 2017

Active Verification for Missing-Annotation-Aware Tiny Surface Defect Detection in Resistors.

Chengdi Zhang1, Mingxuan Yu2, Wenzhang Dong2

  • 1School of Mechanical and Electrical Engineering, Wenzhou University, Wenzhou 325000, 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

Lipid-based nanosystems for atherosclerosis treatment: Evolving approaches and novel therapeutic strategies.

Acta pharmaceutica Sinica. B·2026
Same author

Viologen Derivatives in Aqueous Organic Redox Flow Batteries: Progress and Perspectives.

Advanced materials (Deerfield Beach, Fla.)·2025
Same author

Serum exosomal miR-200c is a potential diagnostic biomarker for breast cancer.

Biomarkers : biochemical indicators of exposure, response, and susceptibility to chemicals·2024
Same author

miR-192 family in breast cancer: Regulatory mechanisms and diagnostic value.

Biomedicine & pharmacotherapy = Biomedecine & pharmacotherapie·2024
Same author

Deciphering the Allosteric Activation Mechanism of SIRT6 Using Molecular Dynamics Simulations.

Journal of chemical information and modeling·2023
Same author

Effects of physical exercise on attention deficit and other major symptoms in children with ADHD: A meta-analysis.

Psychiatry research·2022
This summary is machine-generated.

This study enhances defect detection in resistor images by correcting inaccurate labels and implementing a novel training method. The approach significantly improves detection accuracy, particularly for subtle defects, boosting recall rates.

Area of Science:

  • Computer Vision
  • Machine Learning
  • Materials Science

Background:

  • Automated visual inspection of electronic components like resistors is crucial for quality control.
  • Subtle defects, such as weak coating marks, are often missed in standard imaging and annotation processes.
  • Inaccurate annotations lead to poor performance in deep learning-based defect detectors.

Purpose of the Study:

  • To improve the accuracy of defect detection in resistor images, especially for subtle and easily missed defects.
  • To address the challenge of inaccurate or incomplete annotations in training deep learning models.
  • To develop a robust method for enhancing the performance of object detection models in industrial inspection tasks.

Main Methods:

  • Corrected inaccurate and incomplete defect annotations by using an initial YOLOv2 model to nominate candidates, followed by manual verification.
Keywords:
active verificationlabel repairmissing annotationprototype consistencyresistor surface inspectiontiny defect detection

More Related Videos

Comprehensive Characterization of Extended Defects in Semiconductor Materials by a Scanning Electron Microscope
11:14

Comprehensive Characterization of Extended Defects in Semiconductor Materials by a Scanning Electron Microscope

Published on: May 28, 2016

Related Experiment Videos

Last Updated: Jun 27, 2026

Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography
11:34

Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography

Published on: May 15, 2017

Comprehensive Characterization of Extended Defects in Semiconductor Materials by a Scanning Electron Microscope
11:14

Comprehensive Characterization of Extended Defects in Semiconductor Materials by a Scanning Electron Microscope

Published on: May 28, 2016

  • Introduced a scale-gated prototype consistency term during training to mitigate bias towards dominant tiny defect classes.
  • Evaluated the method on a fixed corrected benchmark and in an end-to-end deployment scenario, including transfer learning on the MVTec AD dataset.

    • Main Results:

    • Mean Average Precision (mAP50) improved from 28.14% to 63.20%, and Recall increased from 18.42% to 62.20% on the corrected benchmark.
    • In end-to-end deployment, mAP50 rose from 43.66% to 63.15%, and Recall improved from 30.01% to 62.24%.
    • Recall for normal-sized defects increased substantially from 26.09% to 56.52%.

    Conclusions:

    • The proposed label correction and training strategy significantly enhance defect detection performance, particularly for challenging, subtle defects.
    • The method demonstrates robustness and generalizability, as shown by performance on public benchmarks and the absence of additional inference complexity.
    • This approach offers a practical solution for improving automated quality control in manufacturing processes involving visual inspection.