Related references
Note: Only part of the references are listed.
Article
Optics
Yanlong Cao et al.
Summary: Visual surface defect inspection is important for assessing product quality in various industrial applications, but it remains challenging to develop deep-learning-based approaches for industrial product defect inspection. To overcome this challenge, we propose a light-weight defect classification model based on pre-trained SqueezeNet architecture and introduce three effective techniques to achieve high accuracy with limited defective training samples. Experimental results validate the effectiveness of our method.
OPTICS AND LASERS IN ENGINEERING
(2023)
Article
Optics
Jingtian Guan et al.
Summary: Automated defect inspection for specular surfaces is challenging due to their reflection property. Deflectometry has been widely used in defect detection by capturing fringe patterns. Traditional methods require hand-crafted features, but this study proposes a deep-learning-based approach using a benchmark dataset named SpecularDefect9. By combining light intensity contrast map and captured fringe pattern, a fusion network accurately classifies different defects.
OPTICS AND LASERS IN ENGINEERING
(2023)
Article
Computer Science, Artificial Intelligence
Zehua Lv et al.
Summary: On the premise of ensuring the defect segmentation precision of aluminum strip surfaces, a lightweight and efficient network is proposed for real-time defect segmentation. The network utilizes the lightweight GhostNet with a proposed dilation attention mechanism for multi-scale feature extraction. It also incorporates a lightweight fusion node and a novel boundary refinement block for efficient integration of features and improved localization ability. Experimental results show that the proposed network achieves a mean intersection over union of 85.51%, a speed of 68.86 fps, and a model volume of 9.38 MB. This network provides a good trade-off between segmentation speed and accuracy for aluminum strip surfaces, making it suitable for real-time segmentation on embedded systems.
JOURNAL OF REAL-TIME IMAGE PROCESSING
(2023)
Article
Computer Science, Artificial Intelligence
Jianyu Liu et al.
Summary: In industrial manufacturing, it is challenging to detect anomalies due to the lack of defective samples for training. This paper proposes a two-stage framework for anomaly detection by training a classification network and building a one-class classifier using learned representations. Experimental results show that our method achieves high AUROC scores of 99.3% and 96.2% and reduces missed detection of anomalous samples.
ADVANCED ENGINEERING INFORMATICS
(2023)
Article
Computer Science, Artificial Intelligence
Bingyu Lu et al.
Summary: Defect inspection is crucial for ensuring the quality of industrial products, with a growing demand for automatic defect detection algorithms due to drawbacks of human visual inspection methods. This paper proposes a deep-learning-based anomaly detection framework for detecting lace defects, utilizing lace videos in the weaving stage. Experimental results demonstrate the effectiveness of the framework in detecting fabric defects using videos.
ADVANCED ENGINEERING INFORMATICS
(2022)
Article
Engineering, Multidisciplinary
Rushuai Tian et al.
Summary: This paper introduces a steel surface defect detector, DCC-CenterNet, utilizing deep learning and keypoint estimation techniques to achieve a balance between speed and accuracy. Experimental results demonstrate that the detector efficiently detects steel surface defects on different datasets.
Article
Computer Science, Artificial Intelligence
Rongg Xu et al.
Summary: In this paper, a high-efficiency defect detector based on self-supervised learning strategy and image segmentation is proposed for surface defect detection. The self-supervised learning strategy and homographic enhancement are used to reduce the need for annotated defective samples. A new method for generating a surface defect simulation dataset is also introduced to address the problem of insufficient training data. Additionally, a lightweight structure with an attention module and a multi-task auxiliary strategy are employed to reduce computation cost and improve segmentation accuracy. Experimental results show that the proposed model achieves competitive performance with smaller computational consumption and higher running speed.
ADVANCED ENGINEERING INFORMATICS
(2022)
Article
Computer Science, Interdisciplinary Applications
Yunpeng Wu et al.
Summary: This paper introduces a new rail boundary guidance network (RBGNet) for salient RS detection, utilizing the complementarity between RS and RE, and enhancing accuracy through high-level RS information injection and hybrid loss. Experiments show that the system achieves a high detection rate and good adaptation capability in complex environments.
COMPUTER-AIDED CIVIL AND INFRASTRUCTURE ENGINEERING
(2022)
Article
Materials Science, Ceramics
Guang Wan et al.
Summary: In this study, a detection method based on improved YOLOv5s for ceramic tile detection was proposed. By deepening the network layer, adding attention mechanism module, introducing small-scale detection layer, enhancing network feature fusion, and replacing convolution, the problems caused by difficulty in extracting texture features and small defects were effectively solved.
CERAMICS INTERNATIONAL
(2022)
Article
Computer Science, Interdisciplinary Applications
Zhuxi Ma et al.
Summary: This paper proposes a lightweight detection method based on attention mechanism for aluminum strip defect inspection. The method achieves higher detection accuracy, smaller model size, and faster detection speed compared to the original YOLOv4 model.
COMPUTERS IN INDUSTRY
(2022)
Article
Computer Science, Artificial Intelligence
Chih-Kai Cheng et al.
Summary: The research introduces a multi-light source lighting strategy to enhance the detection capability of Automated Optical Inspection (AOI) system, optimizing inspection parameters intelligently to maximize the detection of diverse defects.
JOURNAL OF INTELLIGENT MANUFACTURING
(2022)
Article
Zhou OUYANG et al.
Xibei Gongye Daxue Xuebao/Journal of Northwestern Polytechnical University
(2022)
Article
Computer Science, Artificial Intelligence
Ying Liang et al.
Summary: In this article, a simple yet powerful image transformation network is proposed to remove textures and highlight defects at full resolution. The network utilizes a polynomial loss function combining perceptual loss, structural similarity loss, and image gradient loss to effectively suppress texture and emphasize defects. The method demonstrates superior performance in experiments and has been successfully applied to the surface defect online detection system of an aluminum ingot milling production line.
ADVANCED ENGINEERING INFORMATICS
(2022)
Article
Materials Science, Multidisciplinary
Yu He et al.
Summary: This study proposes a semi-supervised framework based on deep learning techniques for texture surface defect classification, utilizing generative adversarial networks and convolutional neural networks. A novel label assignment scheme is introduced to integrate unlabeled samples into semi-supervised learning for enhanced system performance.
Article
Construction & Building Technology
Yanyan Wang et al.
Summary: Current defect detection studies in the industrial fields primarily rely on supervised strategies, which require a large amount of annotated defective samples. However, it is difficult to meet such data requirements in actual industrial scenarios. To address this issue, this paper proposes a novel approach for defect detection using only non-defective samples, which can accurately detect defects with complex backgrounds and weak textures.
AUTOMATION IN CONSTRUCTION
(2022)
Article
Optics
Jie Liu et al.
Summary: In this research, a semi-supervised anomaly detection method based on Dual Prototype Auto-Encoder is proposed, which achieves effective results by introducing dual prototype loss and reconstruction loss to encourage latent vectors to stay closer.
OPTICS AND LASERS IN ENGINEERING
(2021)
Article
Optics
Zhenyu Liu et al.
Summary: This paper proposes a visual defect detection framework based on Convolutional Neural Networks, which elegantly addresses the challenges of large defect shape change, large-scale variation, and high-quality defect localization by introducing three components. Under the COCO evaluation metrics, the method significantly outperforms the Faster RCNN baseline with a large margin.
OPTICS AND LASERS IN ENGINEERING
(2021)
Article
Chemistry, Multidisciplinary
Chaowei Tang et al.
Summary: The paper proposes a semantic segmentation network for surface defect detection of an automobile wheel hub, which tackles the gridding effect in defect detection by using a high-resolution network and multi-scale features, accurately detecting defect boundaries through decoupling features. Experimental results show that the network performs better in various defect detection scenarios.
APPLIED SCIENCES-BASEL
(2021)
Article
Optics
Xiaoming Liu et al.
OPTICS AND LASERS IN ENGINEERING
(2020)
Article
Engineering, Multidisciplinary
Du-Ming Tsai et al.
Article
Chemistry, Physical
Ruofeng Wei et al.
Article
Optics
He Di et al.
OPTICS AND LASERS IN ENGINEERING
(2019)
Article
Computer Science, Information Systems
Guangzhong Cao et al.
Article
Engineering, Multidisciplinary
S. Satorres Martinez et al.
Article
Optics
Paulo M. F. Forte et al.
OPTICS AND LASERS IN ENGINEERING
(2017)
Article
Computer Science, Artificial Intelligence
Chuanxia Jian et al.
APPLIED SOFT COMPUTING
(2017)
Article
Computer Science, Information Systems
Maofu Liu et al.
JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION
(2016)
Article
Chemistry, Physical
Xiao-cui Yuan et al.
APPLIED SURFACE SCIENCE
(2015)
Article
Optics
Yan Zhang et al.
OPTICS AND LASER TECHNOLOGY
(2013)
Article
Transportation Science & Technology
Susan E. Hartfield-Wunsch et al.
SAE INTERNATIONAL JOURNAL OF MATERIALS AND MANUFACTURING
(2011)
Article
Computer Science, Artificial Intelligence
Z Wang et al.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2004)