A Method for Classification of Surface Defect on Metal Workpieces Based on Twin Attention Mechanism Generative Adversarial Network

Hu, Jinghua; Yan, Ping; Su, Yingtao; Wu, Dayuan; Zhou, Hui

doi:10.1109/jsen.2021.3066603

Cited by 21 publications

(13 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to alternate collaborative AM on the basis of cascading attention, Hu et al designed and implemented a dynamic attention network. The iterative process allowed the model to recover from the initial local maximum associated with the incorrect response and produced positive outcomes in the machine reading task of question-answering [ 15 ]. In order to apply the Encoder-Decoder framework based on the attention model to the field of speech recognition, Pang S et al [ 16 ] used AM to identify the correspondence between speech and words.…”

Section: Related Workmentioning

confidence: 99%

[Retracted] Analysis of the Basic Characteristics and Teaching Environment and Mode of Music Appreciation Course Based on Core Literacy

2022

Journal of Environmental and Public Health

View full text Add to dashboard Cite

The main goals of a music appreciation course are to develop college students’ emotions, improve their musical abilities, and strengthen their conceptual framework. This essay primarily introduces the fundamental elements and instructional approach of a music appreciation course built on core literacy. In order to clarify the need for and orientation of basic literacy in college music appreciation teaching, as well as to understand the special value of the subject, we conducted research on the relationship between basic literacy and college music appreciation teaching. Based on this, a new detection algorithm that fuses feature fusion and AM (attention mechanism) is proposed, and visual AM is added to the algorithm based on AM features to help the agent learn the best course of action quickly. The findings indicate that this method’s cumulative normalized discount gain increases by about 0.07 and that the recommended list’s overall quality has significantly improved. The hit ranking is decreased by about 0.05 when compared to the collaborative neural network filtering method without AM.

show abstract

Section: Related Workmentioning

confidence: 99%

[Retracted] Analysis of the Basic Characteristics and Teaching Environment and Mode of Music Appreciation Course Based on Core Literacy

2022

Journal of Environmental and Public Health

View full text Add to dashboard Cite

show abstract

“…This machine learning algorithm has a somewhat lower classification accuracy compared to the deep learning algorithm and there are not many types of defects used in the experiments, which do not have a high practicality. Hu et al (2021) applied the twin attention mechanism to generative adversarial networks and proposed the twin attention mechanism generative adversarial network (TARGAN) can generate high-quality defective images. The spatial attention module and the channel attention module are connected in parallel to form the twin attention mechanism, and the ability of the twin attention mechanism to improve the refinement of defective features is verified by ablation experiments.…”

Section: Classification Of Steel Strip Surface Defectsmentioning

confidence: 99%

Surface defect classification of hot-rolled steel strip based on mixed attention mechanism

Fan

Dong

Guo

2023

RIA

View full text Add to dashboard Cite

Purpose This paper aims to propose a classification method for steel strip surface defects based on a mixed attention mechanism to achieve fast and accurate classification performance. The traditional method of classifying surface defects of hot-rolled steel strips has the problems of low recognition accuracy and low efficiency in the industrial complex production environment. Design/methodology/approach The authors selected min–max scaling comparison method to filter the training results of multiple network models on the steel strip surface defect data set. Then, the best comprehensive performance model EfficientNet-B0 was refined. Based on this, the authors proposed two mixed attention addition methods, which include squeeze-excitation spatial mixed module and multilayer mixed attention mechanism (MMAM) module, respectively. Findings With these two methods, the authors achieved 96.72% and 97.70% recognition accuracy on the steel strip data set after data augmentation for adapting to the complex production environment, respectively. Using the transfer learning method, the EfficientNet-B0 based on MMAM obtained 100% recognition accuracy. Originality/value This study not only focuses on improving the recognition accuracy of the network model itself but also considers other performance indicators of the network, which are rarely considered by many researchers. The authors further improve the intelligent production technique and address this issue. Both methods proposed in this paper can be applied to embedded equipment, which can effectively improve steel strip factory production efficiency and reduce material and time loss.

show abstract

“…A total of six feature maps could be extracted, including Conv7, Conv8_2, Conv9_2, Conv10_2, Conv11_2, and Conv4_3. The sizes of the six feature maps are (38, 38), (19,19), (10, 10), (5,5), (3,3), and (1, 1), respectively. The scale and aspect ratio of the priori box were configured based on the linear increasing rule.…”

Section: Model Constructionmentioning

confidence: 99%

“…[18] designed a composite evaluation metric for the image quality of image dataset, and applied it to train the defect detection model. Hu et al [19] presented a generative adversarial network with the relative average discriminator, which is driven by dual attention mechanisms, to generate high-quality defect images. The network provides a desired solution to the lack of samples and imbalanced classes in the surface defect dataset of metal workpieces in industrial production.…”

Section: Introductionmentioning

confidence: 99%

Detection of Various Types of Metal Surface Defects Based on Image Processing

Xue¹,

Jiang²,

Pang³

2021

View full text Add to dashboard Cite

Machine vision is a promising technique to promote intelligent production. It strikes a balance between product quality and production efficiency. However, the existing metal surface defect detection algorithms are too general, and deviate from electrical production equipment in the level of response time to the target image. To address the two problems, this paper designs a detection algorithm for various types of metal surface defects based on image processing. Firstly, each metal surface image was preprocessed through average graying and nonlocal means filtering. Next, the principle of the composite model scale expansion was explained, and an improved EfficientNet was constructed to classify metal surface defects, which couples spatial attention mechanism. Finally, the backbone network of the single shot multi-box detector (SSD) network was improved, and used to fuse the features of the target image. The proposed model was proved effective through experiments.

show abstract

A Method for Classification of Surface Defect on Metal Workpieces Based on Twin Attention Mechanism Generative Adversarial Network

Cited by 21 publications

References 20 publications

[Retracted] Analysis of the Basic Characteristics and Teaching Environment and Mode of Music Appreciation Course Based on Core Literacy

[Retracted] Analysis of the Basic Characteristics and Teaching Environment and Mode of Music Appreciation Course Based on Core Literacy

Surface defect classification of hot-rolled steel strip based on mixed attention mechanism

Detection of Various Types of Metal Surface Defects Based on Image Processing

Contact Info

Product

Resources

About