Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

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

Mechanomyography is more sensitive than EMG in detecting age-related sarcopenia.

Journal of biomechanics·2009

Same author

Phylogenomic analyses predict sistergroup relationship of nucleariids and fungi and paraphyly of zygomycetes with significant support.

BMC evolutionary biology·2009

Same author

Evolution of an X-linked primate-specific micro RNA cluster.

Molecular biology and evolution·2009

Same author

mTOR regulation and therapeutic rejuvenation of aging hematopoietic stem cells.

Science signaling·2009

Same author

Proteomic analysis of mitochondria reveals a metabolic switch from fatty acid oxidation to glycolysis in the failing heart.

Science in China. Series C, Life sciences·2009

Same author

Characterization and catalytic performance of porous carbon prepared using in situ-formed aluminophosphate framework as template.

Journal of colloid and interface science·2009

Same journal

MT-MRI for detection of renal interstitial fibrosis in renovascular disease.

Scientific reports·2026

Same journal

Detection of underground objects from GPR data using a lightweight YOLO-based approach.

Scientific reports·2026

Same journal

Early systemic inflammatory-metabolic trajectory phenotypes are associated with survival outcomes in metastatic renal cell carcinoma treated with nivolumab.

Scientific reports·2026

Same journal

Water balance components in a dry-seeded rice-wheat system: Untangling the effects of tillage and mulching practices.

Scientific reports·2026

Same journal

Topological approaches to quantum tensor train compression via ZX-calculus and SVD.

Scientific reports·2026

Same journal

determinants of flood impacts and adaptive capacity among market vendors in Walukuba-Masese, Jinja city, Uganda.

Scientific reports·2026

See all related articles

Search research articles

Related Experiment Video

Updated: May 27, 2025

Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography

Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography

Published on: May 15, 2017

A steel surface defect detection method based on improved RetinaNet.

Zhanglin Yang¹, Yu Liu²

¹College of Mechanical and Automotive Engineering, ChuZhou Polytechnic, Chuzhou, 239000, China.

Scientific Reports

|February 19, 2025

Summary

This summary is machine-generated.

This study introduces an improved RetinaNet for steel surface defect detection, enhancing accuracy by adapting feature extraction and fusion. The new method significantly boosts detection performance and reduces model parameters for practical applications.

Keywords:

Defect detection Deformable convolution RetinaNet Steel surface defects

More Related Videos

Scanning Electron Microscopic Evaluation of Surface Defects of Remover Retreatment File After Single and Multiple Uses

Scanning Electron Microscopic Evaluation of Surface Defects of Remover Retreatment File After Single and Multiple Uses

Published on: October 11, 2024

Surface Enhanced Raman Spectroscopy Detection of Biomolecules Using EBL Fabricated Nanostructured Substrates

Surface Enhanced Raman Spectroscopy Detection of Biomolecules Using EBL Fabricated Nanostructured Substrates

Published on: March 20, 2015

Related Experiment Videos

Last Updated: May 27, 2025

Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography

Subsurface Defect Localization by Structured Heating Using Laser Projected Photothermal Thermography

Published on: May 15, 2017

Scanning Electron Microscopic Evaluation of Surface Defects of Remover Retreatment File After Single and Multiple Uses

Scanning Electron Microscopic Evaluation of Surface Defects of Remover Retreatment File After Single and Multiple Uses

Published on: October 11, 2024

Surface Enhanced Raman Spectroscopy Detection of Biomolecules Using EBL Fabricated Nanostructured Substrates

Surface Enhanced Raman Spectroscopy Detection of Biomolecules Using EBL Fabricated Nanostructured Substrates

Published on: March 20, 2015

Area of Science:

Materials Science
Computer Vision
Artificial Intelligence

Background:

Steel surface defects pose challenges for detection due to diverse types, shapes, and background similarity.
Existing methods often struggle with accuracy because of these complex visual characteristics.

Purpose of the Study:

To develop an advanced steel surface defect detection method with higher accuracy and efficiency.
To improve the adaptability of feature extraction and fusion for varied defect appearances.

Main Methods:

Integration of deformable convolutions into the ResNet backbone for adaptive feature extraction.
Implementation of a CA-BiFPN with attention mechanisms for enhanced feature fusion.
Introduction of an IA-BCELoss function to couple classification and regression for precise detection boxes.

Main Results:

The proposed method achieved a 6% improvement in mean Average Precision (mAP) over the original RetinaNet, reaching 81.5%.
It demonstrated superior performance compared to YOLOv7-X and YOLOX-L, with mAP increases of 5.2% and 5.3%, respectively.
The model also achieved significant parameter reduction, decreasing by 37.96M and 21.23M compared to YOLOv7-X and YOLOX-L.

Conclusions:

The improved RetinaNet method offers superior performance for steel surface defect detection.
The adaptive feature extraction, enhanced fusion, and coupled loss function contribute to higher accuracy and practical value.