DG-GAN: A High Quality Defect Image Generation Method for Defect Detection
The surface defect detection of industrial products has become a crucial link in industrial manufacturing. It has a series of chain effects on the control of product quality, the safety of the subsequent use of products, the reputation of products, and production efficiency. However, in actual produ...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2023-06-01
|
Series: | Sensors |
Subjects: | |
Online Access: | https://www.mdpi.com/1424-8220/23/13/5922 |
_version_ | 1797590833998856192 |
---|---|
author | Xiangjie He Zhongqiang Luo Quanyang Li Hongbo Chen Feng Li |
author_facet | Xiangjie He Zhongqiang Luo Quanyang Li Hongbo Chen Feng Li |
author_sort | Xiangjie He |
collection | DOAJ |
description | The surface defect detection of industrial products has become a crucial link in industrial manufacturing. It has a series of chain effects on the control of product quality, the safety of the subsequent use of products, the reputation of products, and production efficiency. However, in actual production, it is often difficult to collect defect image samples. Without a sufficient number of defect image samples, training defect detection models is difficult to achieve. In this paper, a defect image generation method DG-GAN is proposed for defect detection. Based on the idea of the progressive generative adversarial, <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>D</mi><mn>2</mn></mrow></semantics></math></inline-formula> adversarial loss function, cyclic consistency loss function, a data augmentation module, and a self-attention mechanism are introduced to improve the training stability and generative ability of the network. The DG-GAN method can generate high-quality and high-diversity surface defect images. The surface defect image generated by the model can be used to train the defect detection model and improve the convergence stability and detection accuracy of the defect detection model. Validation was performed on two data sets. Compared to the previous methods, the FID score of the generated defect images was significantly reduced (mean reductions of 16.17 and 20.06, respectively). The YOLOX detection accuracy was significantly improved with the increase in generated defect images (the highest increases were 6.1% and 20.4%, respectively). Experimental results showed that the DG-GAN model is effective in surface defect detection tasks. |
first_indexed | 2024-03-11T01:29:05Z |
format | Article |
id | doaj.art-18b71b70aaa549a89d4172510a985a06 |
institution | Directory Open Access Journal |
issn | 1424-8220 |
language | English |
last_indexed | 2024-03-11T01:29:05Z |
publishDate | 2023-06-01 |
publisher | MDPI AG |
record_format | Article |
series | Sensors |
spelling | doaj.art-18b71b70aaa549a89d4172510a985a062023-11-18T17:28:45ZengMDPI AGSensors1424-82202023-06-012313592210.3390/s23135922DG-GAN: A High Quality Defect Image Generation Method for Defect DetectionXiangjie He0Zhongqiang Luo1Quanyang Li2Hongbo Chen3Feng Li4School of Automation and Information Engineering, Sichuan University of Science and Engineering, Yibin 644000, ChinaSchool of Automation and Information Engineering, Sichuan University of Science and Engineering, Yibin 644000, ChinaSchool of Automation and Information Engineering, Sichuan University of Science and Engineering, Yibin 644000, ChinaSichuan Shuneng Electric Power Company Ltd., Chengdu 610000, ChinaSchool of Engineering and Technology, The Open University of Sichuan, Chengdu 610073, ChinaThe surface defect detection of industrial products has become a crucial link in industrial manufacturing. It has a series of chain effects on the control of product quality, the safety of the subsequent use of products, the reputation of products, and production efficiency. However, in actual production, it is often difficult to collect defect image samples. Without a sufficient number of defect image samples, training defect detection models is difficult to achieve. In this paper, a defect image generation method DG-GAN is proposed for defect detection. Based on the idea of the progressive generative adversarial, <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mrow><mi>D</mi><mn>2</mn></mrow></semantics></math></inline-formula> adversarial loss function, cyclic consistency loss function, a data augmentation module, and a self-attention mechanism are introduced to improve the training stability and generative ability of the network. The DG-GAN method can generate high-quality and high-diversity surface defect images. The surface defect image generated by the model can be used to train the defect detection model and improve the convergence stability and detection accuracy of the defect detection model. Validation was performed on two data sets. Compared to the previous methods, the FID score of the generated defect images was significantly reduced (mean reductions of 16.17 and 20.06, respectively). The YOLOX detection accuracy was significantly improved with the increase in generated defect images (the highest increases were 6.1% and 20.4%, respectively). Experimental results showed that the DG-GAN model is effective in surface defect detection tasks.https://www.mdpi.com/1424-8220/23/13/5922deep learninggenerating adversarial networksdefect image generationdefect detection |
spellingShingle | Xiangjie He Zhongqiang Luo Quanyang Li Hongbo Chen Feng Li DG-GAN: A High Quality Defect Image Generation Method for Defect Detection Sensors deep learning generating adversarial networks defect image generation defect detection |
title | DG-GAN: A High Quality Defect Image Generation Method for Defect Detection |
title_full | DG-GAN: A High Quality Defect Image Generation Method for Defect Detection |
title_fullStr | DG-GAN: A High Quality Defect Image Generation Method for Defect Detection |
title_full_unstemmed | DG-GAN: A High Quality Defect Image Generation Method for Defect Detection |
title_short | DG-GAN: A High Quality Defect Image Generation Method for Defect Detection |
title_sort | dg gan a high quality defect image generation method for defect detection |
topic | deep learning generating adversarial networks defect image generation defect detection |
url | https://www.mdpi.com/1424-8220/23/13/5922 |
work_keys_str_mv | AT xiangjiehe dgganahighqualitydefectimagegenerationmethodfordefectdetection AT zhongqiangluo dgganahighqualitydefectimagegenerationmethodfordefectdetection AT quanyangli dgganahighqualitydefectimagegenerationmethodfordefectdetection AT hongbochen dgganahighqualitydefectimagegenerationmethodfordefectdetection AT fengli dgganahighqualitydefectimagegenerationmethodfordefectdetection |