Article
Computer Science, Software Engineering
Cheng Wei, Jiuzhen Liang, Hao Liu, Zhenjie Hou, Zhan Huan
Summary: This paper proposes a multi-stage unsupervised fabric defect detection method based on DCGAN. Through image reconstruction and pixel-level detection, different types of defects can be accurately detected. In addition, the introduction of a classifier training phase and likelihood map further improves the accuracy of defect detection.
Article
Chemistry, Applied
Hongwei Zhang, Xiwei Chen, Shuai Lu, Le Yao, Xia Chen
Summary: The unsupervised defect-detection method for colour-patterned fabric has attracted wide attention. In this paper, a Contrastive Learning-based Attention Generative Adversarial Network (CLAGAN) is proposed for defect detection in colour-patterned fabrics, which effectively improves the detection accuracy.
COLORATION TECHNOLOGY
(2023)
Article
Computer Science, Information Systems
Gwo-Jiun Horng, Min-Xiang Liu, Chien-Chin Hsu
Summary: The treatment of brain ischemia with tissue plasminogen activator may lead to brain hemorrhage risk; decision-making under rescue time pressure is crucial. Lack of benchmarks due to uncertainties post-treatment; Utilization of adaptive deep autoencoder model to learn features of specific data. Preprocessing of data proposed with methods like K-means and image denoising for maximum area preservation; Use of VAE WGAN-GP to generate 3D medical images for insufficient training data, with focus on real data preprocessing and image generation techniques.
COMPUTER COMMUNICATIONS
(2021)
Article
Automation & Control Systems
Benyi Yang, Zhenyu Liu, Guifang Duan, Jianrong Tan
Summary: This article proposes a new data augmentation algorithm called Mask2Defect for metal surface defect inspection. It can generate defects with varied features by infusing prior knowledge-based data. Experimental results show that the synthesized image quality of our method outperforms other methods, and the performance of the inspection model in defect classification and localization has also been improved.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2022)
Article
Chemistry, Analytical
Chao-Ching Ho, Wei-Chi Chou, Eugene Su
Summary: The research focuses on using deep learning and Bayesian optimization algorithm for network pruning to automatically adjust and optimize the defect detection network, significantly improving both inference speed and accuracy.
Article
Computer Science, Information Systems
Danpeng Cheng, Wuxin Sha, Zuo Xu, Shide Li, Zhigao Yin, Yuling Lang, Shun Tang, Yuan-Cheng Cao
Summary: Traditional electron microscope analyses relying on time-consuming human operations are not suitable for large-scale image analysis. To overcome this, AtomGAN, a robust unsupervised learning method, is developed to automatically segment defects in MoS2/WS2. The model is trained on unpaired simulated data and achieves a measurement precision of 96.9%.
SCIENCE CHINA-INFORMATION SCIENCES
(2023)
Article
Computer Science, Information Systems
Xiangjie He, Zhengwei Chang, Linghao Zhang, Houdong Xu, Hongbo Chen, Zhongqiang Luo
Summary: With the development of science and technology, defect detection has become an indispensable part of manufacturing. However, the application of supervised deep learning algorithms in this field is limited due to the difficulty of obtaining defect samples. In recent years, unsupervised deep learning algorithms like GAN have gained attention and been widely used in defect detection due to their strong generation ability. This paper reviews the theoretical basis, technical development, and practical application of GAN-based defect detection, discusses current issues and future research directions, and provides technical information for researchers interested in utilizing GAN for defect detection tasks.
Article
Materials Science, Textiles
Hongwei Zhang, Guanhua Qiao, Shuai Lu, Le Yao, Xia Chen
Summary: In this article, an Attention-based Feature Fusion Generative Adversarial Network framework is proposed for the unsupervised defect detection of yarn-dyed fabrics. The framework utilizes a modified Feature Pyramid Network to fuse multi-level information and an attention mechanism to enhance feature representation capabilities. Experimental results demonstrate the effectiveness of the proposed method compared to other methods on public datasets.
TEXTILE RESEARCH JOURNAL
(2023)
Article
Computer Science, Information Systems
Xiaoxu Cai, Gaige Wang, Jianwen Lou, Muwei Jian, Junyu Dong, Rung-Ching Chen, Brett Stevens, Hui Yu
Summary: This paper introduces a novel approach for saliency detection using a generative adversarial network guided by perceptual loss. The proposed method utilizes shape information to shape the perceptual saliency cues and demonstrates competitive performance on six benchmark datasets, regardless of color images or grayscale images.
INFORMATION SCIENCES
(2024)
Article
Polymer Science
Kaixin Liu, Fumin Wang, Yuxiang He, Yi Liu, Jianguo Yang, Yuan Yao
Summary: In this article, a novel generative manifold learning thermography (GMLT) method is proposed for defect detection and evaluation of composites. The method utilizes spectral normalized generative adversarial networks as an image augmentation strategy, to learn the thermal image distribution and generate virtual images to enrich the dataset. Manifold learning is employed for unsupervised dimensionality reduction, and partial least squares regression is used for defect visualization. Probability density maps and quantitative metrics are introduced to evaluate and explain the defect detection performance. Experimental results demonstrate the superiority of GMLT compared to other methods.
Article
Chemistry, Multidisciplinary
Chuan-Wang Chang, Chuan-Yu Chang, Yuan-Yi Lin, Wei-Wen Su, Henry Shen-Lih Chen
Summary: This paper proposes a method of using a generative adversarial network (GAN) to generate corresponding OCT images from fundus images to assist family doctors in judging whether further examination is needed based on the generated OCT images to achieve early detection and treatment of glaucoma. In addition, in order to improve the classification accuracy of the system deployed in different hospitals or clinics, this paper also proposes to use the incremental training method to fine-tune the model. Experimental results show the effectiveness and feasibility of our proposed method.
APPLIED SCIENCES-BASEL
(2023)
Article
Construction & Building Technology
Gaowei Zhang, Yue Pan, Limao Zhang
Summary: This research introduces a semisupervised generative adversarial network (SSGAN) with two sub-networks for automatic defect detection, utilizing attention mechanism and dual loss functions to enhance segmentation performance and reduce data labeling efforts.
AUTOMATION IN CONSTRUCTION
(2021)
Article
Green & Sustainable Science & Technology
Sertan Serte, Mehmet Alp Dirik, Fadi Al-Turjman
Summary: Healthcare is enhanced through the Internet of things, with machine learning-based systems providing faster services and doctors utilizing artificial intelligence to analyze X-rays and CT scans. This paper proposes a data-efficient deep network that generates synthetic CT scans using a generative adversarial network (GAN) to increase the available data. The GAN-based deep learning model shows superior performance in COVID-19 detection compared to classic models, as evaluated on the COVID19-CT and Mosmed datasets.
Article
Engineering, Multidisciplinary
Haipeng Peng, Jie Zhao, Lixiang Li, Yeqing Ren, Shanshan Zhao
Summary: This article proposes a new one-class adversarial fraud detection model called CS-OCAN, which improves the detection accuracy and stability by modifying autoencoders and GANs.
IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING
(2023)
Article
Engineering, Electrical & Electronic
Kaiyou Song, Hua Yang, Zhouping Yin
Summary: Texture defect inspection remains challenging due to extreme variations in textures and defects. In this study, a novel anomaly composition and decomposition network (ACDN) is proposed for accurate inspection of various texture defects.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)