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 25, 2026

Strain Sensing Based on Multiscale Composite Materials Reinforced with Graphene Nanoplatelets
09:38

Strain Sensing Based on Multiscale Composite Materials Reinforced with Graphene Nanoplatelets

Published on: November 7, 2016

GAN-PD: generative adversarial networks for fabric defect generation towards precise detection.

Yuyang Xia1, Yong Yin1, Fengkai Luan1

  • 1School of Information Engineering, Wuhan University of Technology, Wuhan, 430070, China.

Scientific Reports
|May 18, 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

Molecular epidemiology of clinical <i>Staphylococcus aureus</i> infections: circulating CC398 lineage with human-adapted genomic signatures and virulent ST5-SCC<i>mec</i> II MRSA.

Frontiers in microbiology·2026
Same author

Temporomandibular disorder symptoms in Chinese and the U.S. University students: A cross-sectional study of oral behaviors, anxiety, depression, and country-level moderation.

Cranio : the journal of craniomandibular practice·2026
Same author

Age modifies the association between oral behaviors and orofacial pain in painful TMD patients: a cross-sectional study.

BMC oral health·2026
Same author

Cranial Base Characteristics in Growing Patients With Class III Skeletal Pattern: A Systematic Review and Meta-Analysis.

International dental journal·2026
Same author

Case Report: Zoledronic acid-induced hyperprogression in oncology: a case study of RET-driven neuroendocrine carcinoma treated with Selpercatinib.

Frontiers in oncology·2026
Same author

Dynamic Ensemble Learning with Transfer Learning for Fatigue Performance Prediction in Ni-Based Superalloys.

Materials (Basel, Switzerland)·2026

This study introduces Generative Adversarial Networks for Fabric Defect Generation towards Precise Detection (GAN-PD) to create realistic textile defects for improved automated inspection. GAN-PD enhances defect detection accuracy by focusing on structural realism and detector relevance.

Area of Science:

  • Textile Engineering
  • Computer Vision
  • Artificial Intelligence

Background:

  • Automated textile inspection faces challenges due to limited, geometrically complex defect samples, particularly small scratches.
  • Current generative models often prioritize visual realism over the geometric and contextual fidelity needed for robust defect detection.

Purpose of the Study:

  • To develop a task-oriented generative adversarial framework (GAN-PD) for synthesizing fabric defects that are both structurally realistic and relevant for detection systems.
  • To improve the accuracy and robustness of automated textile inspection by generating high-fidelity defect data.

Main Methods:

  • The proposed Generative Adversarial Networks for Fabric Defect Generation towards Precise Detection (GAN-PD) framework utilizes a generator with Local-Feature Residual modules.
Keywords:
Data augmentationDefect detectionFabric defect synthesisGenerative adversarial networksTask-aware defect synthesis

More Related Videos

A Simple and Scalable Fabrication Method for Organic Electronic Devices on Textiles
06:21

A Simple and Scalable Fabrication Method for Organic Electronic Devices on Textiles

Published on: March 13, 2017

Related Experiment Videos

Last Updated: Jun 25, 2026

Strain Sensing Based on Multiscale Composite Materials Reinforced with Graphene Nanoplatelets
09:38

Strain Sensing Based on Multiscale Composite Materials Reinforced with Graphene Nanoplatelets

Published on: November 7, 2016

A Simple and Scalable Fabrication Method for Organic Electronic Devices on Textiles
06:21

A Simple and Scalable Fabrication Method for Organic Electronic Devices on Textiles

Published on: March 13, 2017

  • The discriminator in GAN-PD incorporates columnar pooling to ensure directional continuity along defect axes.
  • This approach preserves high-frequency details, enforces linearity, and reduces artifacts from isotropic convolutions.

    • Main Results:

    • GAN-PD generates sharper linear anomalies and achieves more consistent texture fusion compared to existing methods.
    • The framework significantly improves detection accuracy for both one-stage and two-stage detectors on benchmark datasets.
    • Experimental results demonstrate superior performance over state-of-the-art Generative Adversarial Networks (GANs).

    Conclusions:

    • Reframing defect synthesis around downstream detection utility, rather than solely visual fidelity, markedly enhances automated textile inspection.
    • GAN-PD offers a novel approach to generating relevant synthetic data, addressing the scarcity of real-world textile defects.
    • The study validates the effectiveness of GAN-PD in improving the robustness and accuracy of automated quality control in the textile industry.