Related references
Note: Only part of the references are listed.
Article
Automation & Control Systems
Shuai Ma et al.
Summary: Deep neural networks have shown great improvement in rail surface defect segmentation when the distribution of test samples is the same as that of training samples. However, in practical inspection scenarios, the appearance of the rail surface varies due to different service times and natural conditions. Conventional deep learning models have limited generalization in scenes with distribution differences. To overcome this challenge, we propose a novel one-shot unsupervised domain adaptation framework that aligns the pixel-level distribution between training and test images using a shape-consistent style transfer module. Our method reconstructs the training image based on the one-shot test image to have the same appearance, and employs multitask learning to prevent content distortion. Furthermore, we design an edge-aware defect segmentation model and train it using the reconstructed training images to improve the model's robustness to distribution differences.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2023)
Article
Engineering, Civil
Yanyan Wang et al.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2023)
Article
Computer Science, Interdisciplinary Applications
Tongzhi Niu et al.
Summary: Surface defect detection is a crucial task in manufacturing, and addressing imbalanced data is a challenge. In this study, we propose a neural network that compares templates and testing samples to address the lack of positive sample data and the difficulty in predicting new batch data distribution. We introduce the Dual-Attention Mechanism (DAM) for noise-free defect feature extraction and the Recurrent Residual Attention Mechanism (RRAM) for noise shielding and multi-scale feature fusion. Our method outperforms existing state-of-the-art methods in Optical Communication Devices (OCDs), Printed Circuit Boards (PCBs), and Motor Commutator Surface Defects (MCSD) datasets. This work provides a promising direction for surface defect detection in OCDs and can be generalized to other flexible manufacturing systems (FMS).
COMPUTERS IN INDUSTRY
(2023)
Article
Computer Science, Artificial Intelligence
Wei Zhu et al.
Summary: In this study, a new network architecture called LSwin Transformer is proposed for steel-surface defect detection. Various techniques, such as convolutional embedding module, attention patch merging module, and depth multilayer perceptron module, are introduced to enhance the adaptability of the Swin Transformer model. Experimental results demonstrate that the proposed model outperforms other methods with a mean average precision of 81.2%.
ADVANCED ENGINEERING INFORMATICS
(2023)
Article
Engineering, Electrical & Electronic
Mingxu Li et al.
Summary: Inspection of railway defects is crucial for train safety and efficiency. In this article, a rendering-based fully convolutional network is proposed to achieve a balance between efficiency and precision by using a coarse-to-fine approach. The network utilizes multiple scales of the feature map and employs residual connections to improve low-level feature detection. Experimental results demonstrate that the proposed method outperforms other state-of-the-art image segmentation methods in terms of frame rate and performance.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2023)
Article
Automation & Control Systems
Tongzhi Niu et al.
Summary: In this study, a new memory-augmented adversarial autoencoder (MAA) is proposed for real-time defect detection and localization using defect-free samples only. By introducing a memory module and redesigning the reconstruction loss function, the proposed MAA avoids missing defect detection. The results show that the proposed method is effective, accurate, and robust, suitable for real-time industrial production.
IEEE-ASME TRANSACTIONS ON MECHATRONICS
(2022)
Article
Automation & Control Systems
Xuefeng Ni et al.
Summary: In this article, a attention neural network based rail surface defect detection method is proposed, which addresses the challenges of complex background and data imbalance through techniques such as key-point estimation and cross-stage fusion. Experimental results demonstrate that the proposed method outperforms competitive methods on four surface defect datasets.
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
(2022)
Article
Engineering, Civil
Zhengxing Chen et al.
Summary: This paper proposes a multi-source data fusion algorithm for rail surface defect detection in both camera-based rail inspection images and ultrasound B-scan images. The algorithm utilizes image processing and feature extraction networks to achieve rail surface segmentation and feature extraction, and uses a feature fusion network to fuse the feature information from different data sources. Experimental results demonstrate that the algorithm achieves high accuracy in rail surface defect detection.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Computer Science, Software Engineering
Wenhai Wang et al.
Summary: This work presents the improved Pyramid Vision Transformer v2 (PVT v2) by adding three designs, achieving significant improvements in fundamental vision tasks. PVT v2 performs comparably or better than recent work such as the Swin transformer.
COMPUTATIONAL VISUAL MEDIA
(2022)
Article
Engineering, Civil
Hui Zhang et al.
Summary: This study proposes a multi-model rail surface defect detection system based on convolutional neural networks, which can rapidly and accurately identify various defects on rail surfaces and improve the detection performance.
IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS
(2022)
Article
Engineering, Mechanical
Guizhong Fu et al.
Summary: This paper proposes a multi-light source illumination/acquisition system and a multi-stream CNN model for high-accuracy surface defect classification on highly reflective metal. By fusing features extracted from multi-light source illuminated images, more accurate recognition results can be generated. In addition, the authors also propose individual stream deep supervision and channel attention-based feature re-calibration techniques.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2022)
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)
Proceedings Paper
Computer Science, Artificial Intelligence
Lukas Hoyer et al.
Summary: As acquiring pixel-wise annotations for real-world images is costly, this paper explores the use of synthetic data and unsupervised domain adaptation (UDA) to train a model that can adapt to real images without annotations. The authors benchmark different network architectures for UDA and find the potential of Transformers for UDA semantic segmentation. They propose a novel UDA method called DAFormer, which achieves significant improvements in state-of-the-art results.
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
(2022)
Article
Engineering, Electrical & Electronic
Ihor Konovalenko et al.
Summary: This study investigates the impact of illumination level on the quantitative indicators of mechanical damage of rolled metal strips. Through a physical model experiment and analysis using a neural network model, the study identifies the differences in damage recognition under different illumination levels and provides insights for further adjustments of industrial systems.
Article
Computer Science, Artificial Intelligence
Yucheng Wang et al.
Summary: Surface defect recognition is crucial for improving surface quality, and CNN-based ProtoCNN method utilizing prototype vectors from the training dataset enhances recognition accuracy.
ADVANCED ENGINEERING INFORMATICS
(2021)
Proceedings Paper
Engineering, Biomedical
Xingchen Zhao et al.
Summary: This study explores the application of machine learning models to unseen medical imaging data, focusing on the challenge of Domain Generalization. By using two distinct Domain Generalization approaches, researchers achieved significant improvements in predicting white matter hyperintensity (WMH) on an unseen target domain.
2021 IEEE 18TH INTERNATIONAL SYMPOSIUM ON BIOMEDICAL IMAGING (ISBI)
(2021)
Article
Computer Science, Artificial Intelligence
Wenting Dai et al.
ADVANCED ENGINEERING INFORMATICS
(2020)
Article
Optics
Guorong Song et al.
OPTICS AND LASERS IN ENGINEERING
(2020)
Article
Computer Science, Interdisciplinary Applications
Ling Zhang et al.
IEEE TRANSACTIONS ON MEDICAL IMAGING
(2020)
Article
Materials Science, Multidisciplinary
Ihor Konovalenko et al.
Article
Engineering, Electrical & Electronic
Haomin Yu et al.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2019)
Article
Computer Science, Artificial Intelligence
Abdul Mujeeb et al.
ADVANCED ENGINEERING INFORMATICS
(2019)
Article
Optics
Guizhong Fu et al.
OPTICS AND LASERS IN ENGINEERING
(2019)
Article
Computer Science, Artificial Intelligence
Agisilaos Chartsias et al.
MEDICAL IMAGE ANALYSIS
(2019)
Proceedings Paper
Computer Science, Artificial Intelligence
Xun Huang et al.
2017 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV)
(2017)