相关参考文献
注意:仅列出部分参考文献,下载原文获取全部文献信息。
Article
Environmental Sciences
Fan Yang et al.
Summary: With the rise of human-robot collaboration and artificial intelligence in Industry 5.0, ultrasonic non-destructive testing (NDT) technology is being increasingly used for quality inspections. This paper presents a systematic comparison of different time-of-flight (ToF) algorithms and introduces an auto-diagnosis method based on the Defect Peaks Tracking Model (DPTM) for ultrasonic echo signal processing. The proposed DPTM, using Hilbert transform and wavelet denoising, can locate defects with sizes of 2-10 mm. The real-time auto-diagnosis feature of DPTM has the potential to be integrated with AI, such as machine learning and deep learning, for more intelligent industrial health inspection approaches.
Article
Chemistry, Multidisciplinary
Alexey N. Beskopylny et al.
Summary: The creation and training of artificial neural networks enable the identification of patterns and hidden relationships in the production of unique building materials, prediction of mechanical properties, and problem-solving in defect detection and classification.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Information Systems
Longlong Li et al.
Summary: The paper proposes a method named GBH-YOLOv5 for the detection of surface defects on PV panels. This method utilizes Ghost convolution, BottleneckCSP module, and a prediction head to improve the accuracy of detecting tiny defects. By compressing and cropping the original image, and incorporating feature extraction and classification networks, the proposed method significantly improves the mAP performance of PV panel defect detection.
Article
Chemistry, Analytical
Andrea Meoni et al.
Summary: A diffuse and continuous monitoring of the in-service structural response of buildings can allow for the early identification of the formation of cracks and collapse mechanisms before the occurrence of severe consequences. In the case of existing masonry constructions, the implementation of tailored Structural Health Monitoring (SHM) systems appears quite significant, given their well-known susceptibility to brittle failures. Recently, a new sensing technology based on smart bricks, i.e., piezoresistive brick-like sensors, was proposed in the literature for the SHM of masonry constructions. Overall, the effectiveness of smart bricks in strain monitoring and crack detection is demonstrated.
Article
Tareq Salem et al.
Article
Chemistry, Analytical
Xiangyang Xu et al.
Summary: The intelligent crack detection method is of great significance for intelligent operation and maintenance as well as traffic safety. This paper investigates the application of deep learning in intelligently detecting road cracks and compares and analyzes Faster R-CNN and Mask R-CNN. The results show that the joint training strategy is effective, but it degrades the effectiveness of the bounding box detected by Mask R-CNN.
Article
Chemistry, Analytical
Fu-Jun Du et al.
Summary: This paper proposes a lightweight target detection algorithm that enhances feature extraction and optimizes sample imbalance for accurate and reliable road defect detection. The proposed model outperforms other models in terms of mAP@.5 in evaluation tests.
Article
Chemistry, Analytical
Zexuan Guo et al.
Summary: With the development of artificial intelligence technology and the popularity of intelligent production projects, intelligent inspection systems have become a hot topic in the industrial field. This paper introduces the improved MSFT-YOLO model for object detection in the industry, which addresses challenges such as background interference and scale changes. The model achieves real-time detection and shows higher detection accuracy compared to baseline models, offering advantages and improvements.
Article
Environmental Sciences
Min Jae Park et al.
Summary: This study introduces a nondestructive and noncontact testing method using thermal images and machine learning to detect the depth of concrete cracks. By measuring the temperature of cracks and surfaces and considering relevant parameters, the crack depth can be accurately predicted. Different machine learning algorithms are compared to identify the best algorithm.
Article
Chemistry, Multidisciplinary
Chien-Yi Huang et al.
Summary: This study proposes an automatic defect detection system using deep learning to identify categories and locations of defects, leading to increased accuracy and reduced cost of human resources.
APPLIED SCIENCES-BASEL
(2022)
Article
Construction & Building Technology
Dimitrios Loverdos et al.
Summary: This paper aims to improve automation in brick segmentation and crack detection of masonry walls through image-based techniques and machine learning. It shows that machine learning provides better outcomes than typical image-processing applications for brick segmentation and crack detection.
AUTOMATION IN CONSTRUCTION
(2022)
Article
Green & Sustainable Science & Technology
Dhirendra Prasad Yadav et al.
Summary: Early crack detection is crucial for saving lives and preventing building collapses. This study proposes a deep convolutional neural network called 3SCNet, which achieves high sensitivity, specificity, and accuracy by using SLIC super pixel segmentation and LBP texture pattern recognition.
Article
Chemistry, Multidisciplinary
Nils Hutten et al.
Summary: Artificial intelligence has been considered as an approach to visual inspection in industrial applications for decades. Recent advances in deep learning, particularly in attention-based vision transformer architectures, have the potential to enable automated visual inspection even in complex environmental conditions. However, the application of vision transformers to real world visual inspection is still limited, possibly due to the assumption that they require large amounts of data to be effective.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Information Systems
Fatema Tuz Zohra et al.
Summary: This paper presents a conveyor belt structural health monitoring model using machine learning and IoT connectivity, which can detect crack width, orientation, and location with high accuracy and predict the region of cracks. The method has significant industrial significance in coal mines.
Article
Computer Science, Information Systems
Gang Li et al.
Summary: This paper proposes an industrial defect detection method based on an expanded perceptual field and feature fusion to address the low efficiency, high false detection rate, and poor real-time performance of current methods. The method enhances the network structure and critical information extraction, introduces a feature fusion method with a shallower feature map and dense multiscale weighting, and achieves fast and accurate detection of industrial surface defects.
Article
Chemistry, Multidisciplinary
Sergei Khotiaintsev et al.
Summary: This study examines the suitability and accuracy of using silica optical fibers bonded to the surface of brick masonry to detect structural cracks. The research successfully achieved reliable detection of cracks with a given minimum value by using arrays of optical fibers. The technique is simple, cost-effective, and can be applied to damage-detection systems in buildings in seismic zones.
APPLIED SCIENCES-BASEL
(2022)
Article
Chemistry, Multidisciplinary
Faquan Chen et al.
Summary: This article presents an adaptive convolution and anchor network, called ACA-Net, for metallic surface defect detection. ACA-Net improves the performance of defect detection by using adaptive convolution and anchor. Extensive experiments validate its effectiveness.
APPLIED SCIENCES-BASEL
(2022)
Article
Agronomy
Hongjun Wang et al.
Summary: This study proposes a strategy for robotic fruit picking to enhance the accuracy of the vision system by remote and precise location methods. Different algorithms are used to determine the location of fruit bunches and bifurcate stems, guiding the robotic arm for picking and achieving a success rate of 88.46%.
Article
Computer Science, Artificial Intelligence
Md Monirul Islam et al.
Summary: This research proposes a transfer learning-based method using Convolutional Neural Networks (CNN) for detecting cracks in concrete structures. The method achieves good performance on both public and external datasets, demonstrating its effectiveness in crack detection.
Article
Physics, Multidisciplinary
Shanyong Xu et al.
Summary: This study proposes an insulator defect detection algorithm based on an improved MobilenetV1-YOLOv4, which enhances the accuracy and detection speed by introducing lightweight modules and attention mechanisms into the feature extraction process.
Article
Chemistry, Analytical
Dehua Wei et al.
Summary: This article proposes an image captioning model for railway track line monitoring, which generates inspection reports by extracting information about key components from images, and applies techniques to improve defect description performance and image processing speed.
Article
Chemistry, Analytical
Jun Hu et al.
Summary: This paper proposes an improved YoLoX-Nano method for rail fastener defect detection by adding the CA attention mechanism and ASFF. The improved model achieves significant improvements on different types of fasteners and also shows a remarkable increase in detection speed. It can accurately and rapidly detect rail fastener defects, and has the advantage of being deployed on lightweight devices.
Review
Chemistry, Physical
Ali Jaber et al.
Summary: This article provides an overview of both conventional inspection methods and smart techniques used in defect detection. It also examines the opportunities for integrating non-destructive evaluation (NDE) methods with Industry 4.0 technologies. The challenges hindering progress in the field are discussed, along with potential solutions. A virtual inspection system is proposed as a means of deploying smart inspection.
Article
Chemistry, Physical
Sergey A. Stel'makh et al.
Summary: This study developed and trained a neural network and an ensemble model to predict the mechanical properties of lightweight fiber-reinforced concrete, achieving high accuracy. It is of great significance for predicting such heterogeneous materials.
Article
Chemistry, Physical
Vojtech Barton et al.
Summary: Historical buildings and monuments made of brickwork require attention to their durability. This study assesses the durability of solid-fired bricks through non-destructive measurements that identify defects in the material's internal structure, resulting in the determination of four durability classes.
Article
Chemistry, Multidisciplinary
Dong-Han Mo et al.
Summary: This study combines machine vision and drone technology to detect cracks in retaining walls in mountaineering areas or forest roads. Deep learning is used to extract the characteristics of the cracks and evaluate their danger level. It provides suggestions based on the detected gap images and improves the efficiency of gap identification with an expanded database.
APPLIED SCIENCES-BASEL
(2022)
Review
Chemistry, Multidisciplinary
Giuseppe Ciaburro et al.
Summary: Acoustic emission is a nondestructive control technique that analyzes structural damage by capturing ultrasonic signals emitted by materials. Recently, machine learning has been extensively used in acoustic emission technology for detecting and locating damage and predicting failure modes.
APPLIED SCIENCES-BASEL
(2022)
Article
Chemistry, Multidisciplinary
Alexey N. Beskopylny et al.
Summary: Machine learning methods have been applied in the construction industry to predict the mechanical properties of building materials, particularly in improving concrete production using artificial intelligence algorithms. This study developed and compared three machine learning algorithms for predicting the compressive strength of concrete, with the k-nearest neighbors algorithm showing the smallest errors and highest coefficient of determination. The developed models can be successfully implemented in the production process and quality control of building materials.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Information Systems
Eleni Vrochidou et al.
Summary: This work provides a comprehensive study on marble crack segmentation using deep learning techniques. The authors propose efficient network architectures and feature extraction methods, making an important contribution to addressing the problem of marble crack segmentation.
Article
Engineering, Civil
Wenjun Wang et al.
Summary: The automated detection of concrete surface defects is a deep learning-based method that uses a one-stage object detection network, combining the backbone network EfficientNetB0 and feature pyramid network. Training and testing on the concrete surface defects dataset showed that the method achieved high detection accuracy.
Article
Construction & Building Technology
L. Minh Dang et al.
Summary: This research focuses on implementing computer vision techniques and deep learning to automate crack segmentation and real-life crack length measurement of masonry walls. The experimental results demonstrate that deep learning-based crack segmentation outperforms previous approaches and can provide accurate measurements.
CONSTRUCTION AND BUILDING MATERIALS
(2022)
Article
Chemistry, Multidisciplinary
Shunyong Zhou et al.
Summary: A lightweight rolled steel strip surface defect detection model, YOLOv5s-GCE, is proposed to improve the efficiency and accuracy of industrialized rolled steel strip defect detection. Experimental results demonstrate that the model performs well in terms of accuracy, model size, and detection speed.
APPLIED SCIENCES-BASEL
(2022)
Article
Construction & Building Technology
Shuai Teng et al.
Summary: Automatic bridge surface defect detection is an important and efficient method for saving human resources and improving work efficiency. This study investigates the performance of different object detection algorithms in bridge surface defect detection, and proposes an improved YOLOv3 network as a decent detector for fast and real-time detection of bridge defects.
Article
Engineering, Electrical & Electronic
Luya Yang et al.
Summary: This paper proposes an improved intelligent detection system that can effectively identify defects on the surface of steel plates and improve production quality. Compared to other methods, this approach shows better overall performance and meets the industry's requirements for accuracy and real-time performance.
Article
Optics
Yongzhong Fu et al.
Summary: A deep-learning-based quality assessment algorithm for MLC-Si wafers is proposed to address challenges in defect detection in PV cells using traditional machine vision methods. Experimental results show that the prediction error of the improved network model is reduced by 63%, with a reasoning speed of 10.22 FPS, indicating good detection performance.
Article
Chemistry, Multidisciplinary
Levon R. Mailyan et al.
Summary: The article develops methods and methodology for experimental studies on concrete products with annular cross-section, demonstrating the advantages of vibro-centrifugation over centrifugation and vibration techniques in terms of structure and performance. It has been shown through experimental studies that the outer layers of centrifuged and vibro-centrifuged concretes have the best characteristics, while the inner layers have the worst characteristics. The variatropic structure of these concretes has been experimentally confirmed with different strength and deformation characteristics in different layers.
APPLIED SCIENCES-BASEL
(2021)
Article
Chemistry, Analytical
Mitchell J. Hallee et al.
Summary: The study focused on crack detection in masonry using various machine learning methods, with the CNN showing better performance in domain adaptation from lab to real-world images compared to simple classifiers. Conversely, the simple classifiers performed significantly better in the reverse domain adaptation task.
Article
Environmental Sciences
Linlin Zhu et al.
Summary: This paper improves the YOLOv5 object detection method with attention mechanism and designs a pyramid based approach to detect boulders from planetary images, achieving a 3.4% increase in precision. The improved method extracts multiple layers of images with different resolutions and detects boulders of various scales, contributing to the analysis of boulder distribution on Bennu asteroid.
Article
Chemistry, Multidisciplinary
Rajagopalan-Sam Rajadurai et al.
Summary: This study utilized deep convolutional neural networks and transfer learning to detect cracks, achieving a high accuracy rate of 99.9% with the trained model. By fine-tuning the architecture and augmenting the image datasets, the model successfully achieved precise detection and classification of cracks.
APPLIED SCIENCES-BASEL
(2021)
Article
Chemistry, Physical
Alexey Beskopylny et al.