检索结果-内蒙古大学图书馆

Event-Based Weak Oil Impurity Detection

IEEE ACCESS 2024年 12卷 18868-18876页

作者： Li, Ming Wang, Huiyu Chen, Qianghua Zong, Zhaoxiang Shanghai Dianji Univ Sch Elect Informat Shanghai 201306 Peoples R China

Machine vision-based methods are usually applied to detect impurities in transparent liquids like water or alcohols. Oils have higher viscosity and lower light transmittance, so traditional vision detection methods do not work well. A weak oil impurity detection algorithm was proposed using event stream data. In this method, the binary image of the oil was firstly captured using an event camera. To reduce the noise interference from the event camera, image filtering and morphological operations were applied. Then, image algebra operations were used to remove the oil container's bottom pattern. Finally, impurity detection was performed through the YOLOv5 network. Three common edible oils serve as the experimental samples. Small flying insects, raw material fragments, metal fragments, hair strands, and tin beads of various sizes are selected as the weak impurities. Experiments were performed on a dataset containing 3000 sample images. To the best of our knowledge, existing algorithms can detect impurities with the minimum size of 0.4mm, and most of the experimental samples are transparent liquids. The proposed method can be applied to detect impurities in water and oils, and the detectable impurity size limit is 0.2mm.

关键词： Impurities Feature extraction Cameras Oils Liquids classification algorithms YOLO Event detection Machine vision Impurity detection dynamic vision event camera

来源：评论

学校读者我要写书评

暂无评论

Detection of Bolt Defects on Transmission Lines Based on Multi-Scale YOLOv7

引用

IEEE ACCESS 2024年 12卷 156639-156650页

作者： Peng, Lincong Wang, Kerui Zhou, Hao Li, Haiyan Yu, Pengfei Yunnan Univ Sch informat Kunming 650504 Peoples R China

Bolts are indispensable in power systems, serving a critical function in securely fastening various components and ensuring stable operation of transmission lines. To address the challenges associated with detecting small bolt objects in images captured by unmanned aerial vehicles (UAVs), we propose an enhanced multi-scale YOLOv7 model called multi-scale YOLOv7 (MS-YOLOv7). First, a multi-scale path aggregation network (M-PANet) is developed, which seamlessly integrates multi-scale information and features four detection heads, thereby reducing the miss rate of small objects. Second, a channel-coordinate attention (CCA) mechanism is introduced to enhance the network's focus on spatial and channel information, thereby augmenting object localization accuracy. Furthermore, in the feature fusion process, the selected convolutions are replaced with a fully self-calibrated convolutional network (FSCNet) to expand the receptive field and enable richer feature extraction. Finally, standard max pooling is replaced with space-to-depth convolution (SPD-Conv) to preserve object detail information during the downsampling phase. Experimental results on a proprietary bolt defect dataset demonstrated significant improvements over the original YOLOv7 model, with a 2.52% increase in mean Average Precision (mAP), a 3.07% improvement in recall, a 1.54% improvement in accuracy, and a 2.41% increase in F1 score. The algorithm also exhibited robust performance on the CSUST Chinese traffic sign detection benchmark (CCTSDB) dataset, demonstrating its strong generalization ability and robustness.

关键词： Fasteners Feature extraction Power transmission lines Defect detection Object detection Convolutional neural networks Power capacitors classification algorithms Accuracy Transforms Channel coordinate attention fully self calibrated convolutional network multi-scale fusion object detection transmission line defect detection

来源：评论

学校读者我要写书评

暂无评论

ML-Powered Handwriting Analysis for Early Detection of Alzheimer's Disease

引用

IEEE ACCESS 2024年 12卷 69031-69050页

作者： Mitra, Uddalak Ul Rehman, Shafiq Siliguri Inst Technol Siliguri 734009 West Bengal India Kingdom Univ Coll Informat Technol Riffa 3903 Bahrain

Alzheimer's disease (AD) is a progressive, incurable condition leading to decline of nerve cells and cognitive functions over time. Early detection is essential for improving quality of life, as treatment strategies aim to decelerate its progression. AD also impacts fine motor control, including handwriting. Utilizing machine learning (ML) with efficient data analysis methods for early detection of Alzheimer's disease (AD) through handwriting analysis holds considerable promise for clinical diagnosis, albeit a challenging undertaking. In this study, we address this complexity by employing ensemble machine learning, which amalgamates diverse ML algorithms to enhance predictive performance. Our approach involves developing an ensemble model for analysis of handwriting kinetics, utilizing the stacking technique to integrate multiple base-level classifiers. The study encompasses 174 individuals, including 89 diagnosed with Alzheimer's disease and 85 healthy participants, drawn from the DARWIN dataset (Diagnosis AlzheimeR WIth haNdwriting). To discern the most effective base classifiers, we employ both Repeated-k-fold and Monte-Carlo Cross Validation techniques. Subsequently, top k features are selected for each best-performing base classifier using analysis of variance (ANOVA) and recursive feature elimination (RFE). The final step involves consolidating predictions from the base classifiers through a stacking ensemble, resulting in an impressive performance. The ensemble model achieves 97.14% accuracy, 95% sensitivity, 100% specificity, 100% precision, 97.44% F1-score, 94.37% Matthews Correlation Coefficient (MCC), 94.21% Cohen Kappa, and 97.5% Area Under the Receiver Operating Characteristic Curve (AUC-ROC). Comparative performance analysis demonstrates that our proposed model surpasses all state-of-the-art models based on the DARWIN dataset for Alzheimer's disease prediction. These findings underscore the potential of machine learning to offer highly accurate predict

关键词： Alzheimer's disease Task analysis Feature extraction Solid modeling classification algorithms Predictive models Prediction algorithms Alzheimer's disease prediction diagnosis AlzheimeR WIth handwriting ensemble machine learning handwriting analysis for identifying neurodegenerative disease machine learning for disease prediction machine learning based Alzheimer's disease prediction

来源：评论

学校读者我要写书评

暂无评论

Automated identification of clinically meaningful biomechanical phenotypes in cerebral palsy through multicenter gait data

引用

CLINICAL BIOMECHANICS 2025年 125卷 106501-106501页

作者： Graf, Adam Krzak, Joseph J. Kruger, Karen M. Davids, Jon Smith, Ryan Steinlein, Brandon Bagley, Anita Shriners Childrens Chicago Mot Anal Ctr Engineer 2211 North Oak Pk Ave Chicago IL 60707 USA Shriners Childrens Sacramento Mot Anal Lab Orthopaed Surg 2425 Stockton Blvd Sacramento CA 95817 USA Univ Calif Davis Ben Ali Chair Pediat Orthopaed Dept Orthopaed Surg 4860 & St Suite 3800 Sacramento CA 95817 USA Midwestern Univ 555 31st St Downers Grove IL 60515 USA Marquette Univ Mot Anal Ctr Director 1250 W Wisconsin Ave Milwaukee WI 53233 USA

Background: Cerebral palsy is the most prevalent motor disability in childhood, encompassing various movement disorders that affect walking. Researchers have described gait patterns in cerebral palsy, but these are often subjective and based on clinician experience. This study introduces an automated approach to objectively identify clinically meaningful biomechanical phenotypes in cerebral palsy and test it on multicenter gait data. Utilizing instrumented gait analysis, this research aims to improve treatment strategies for gait dysfunction. This study addresses whether classification algorithms can objectively identify clinically meaningful gait patterns and if severe gait deviations are more frequent in advanced forms of cerebral palsy. Methods: Two novel classification algorithms (sagittal and transverse planes) were developed and automated in Python. These were based on previous work and refined using clinical expertise and data from four motion analysis centers in the Shriners Children's system, including 700 patients with cerebral palsy. The patient's gait data was applied to the treatment algorithms, and the percentage of each phenotype is presented. Findings. Novel sagittal and transverse plane gait phenotype algorithms were created. When applied to the cerebral palsy cohort, we found that more severe gait deviations, or combinations of deviations, were more apparent in the more severe forms of cerebral palsy. Interpretations. Classifying a patient's biomechanical phenotype provides valuable insights into therapeutic interventions. The results allow for the automation of data-driven classification algorithms, leading to efficient, accurate, and reliable classifications of biomechanical phenotypes that support evidence-based, personalized treatment decisions and clinical management.

关键词： Cerebral palsy Biomechanical phenotypes classification algorithms Motion analysis Instrumented gait analysis Pediatric Orthopaedics

来源：评论

学校读者我要写书评

暂无评论

Point Cloud Segmentation Algorithm Based on Improved Euclidean Clustering

引用

IEEE ACCESS 2024年 12卷 152959-152971页

作者： Chen, Fangrui Xie, Feifei Sun, Lin Gu, Yuchao Zhang, Zhipeng Chen, Jinpeng Zhang, Jinrui Yi, Mingzhe Shandong Univ Sci & Technol Coll Geodesy & Geomat Qingdao 266590 Peoples R China Zhejiang Zhengyuan Geog Informat Co Ltd Hangzhou 313200 Zhejiang Peoples R China

Point cloud segmentation is a crucial technique for object recognition and localization, widely employed in various applications such as point cloud registration, 3D reconstruction, object recognition, and robotic grasping. However, in practical scenarios, challenges such as sparse object features, significant pose variations, or instances of object adhesion, stacking, and occlusion can lead to difficulties and poor stability in point cloud segmentation. In response to these challenges, this paper proposes a novel point cloud segmentation algorithm based on enhanced Euclidean clustering. To address the issue of over-segmentation and under-segmentation in cases of object adhesion and stacking in unordered scenes, a method is introduced for discriminating collision points based on normal angle constraints. Subsequently, an adaptive search radius is employed to determine the clustering distance threshold, enhancing the algorithm's ability to handle different object arrangements. Finally, the collision points are reintegrated into the clustering results to ensure the completeness of the segmented target point cloud. Experimental results using the ROBI public dataset and real-world point cloud segmentation scenarios demonstrate that the proposed algorithm effectively resolves challenges associated with the segmentation of adhering and stacking objects. The improved algorithm exhibits higher accuracy and robustness in multi-object segmentation across diverse scenes.

关键词： Point cloud compression Clustering algorithms Three-dimensional displays Surface treatment Noise measurement classification algorithms Object recognition Stacking Solid modeling Partitioning algorithms 3D point cloud Euclidean clustering unordered scenes point cloud segmentation

来源：评论

学校读者我要写书评

暂无评论

A Novel Hybrid Ordinal Learning Model With Health Care Application

引用

IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ENGINEERING 2025年 22卷 339-352页

作者： Wang, Lujia Wang, Hairong Su, Yi Lure, Fleming Li, Jing Georgia Inst Technol H Milton Stewart Sch Ind & Syst Engn Atlanta GA 30332 USA Banner Alzheimers Inst Phoenix AZ 85006 USA MS Technol Corp Rockville MD 20850 USA

Ordinal learning (OL) is a type of machine learning models with broad utility in health care applications such as diagnosis of different grades of a disease (e.g., mild, modest, severe) and prediction of the speed of disease progression (e.g., very fast, fast, moderate, slow). This paper aims to tackle a situation when precisely labeled samples are limited in the training set due to cost or availability constraints, whereas there could be an abundance of samples with imprecise labels. We focus on imprecise labels that are intervals, i.e., one can know that the a sample belongs to an interval of labels but cannot know which unique label it has. This situation is quite common in health care datasets due to limitations of the diagnostic instrument, sparse clinical visits, or/and patient dropout. Limited research has been done to develop OL models with imprecise/interval labels. We propose a new Hybrid Ordinal Learner (HOL) to integrate samples with both precise and interval labels to train a robust OL model. We also develop a tractable and efficient optimization algorithm to solve the HOL formulation. We compare HOL with several recently developed OL methods on four benchmarking datasets, which demonstrate the superior performance of HOL. Finally, we apply HOL to a real-world dataset for predicting the speed of progressing to Alzheimer's Disease (AD) for individuals with Mild Cognitive Impairment (MCI) based on a combination of multi-modality neuroimaging and demographic/clinical datasets. HOL achieves high accuracy in the prediction and outperforms existing methods. The capability of accurately predicting the speed of progression to AD for each individual with MCI has the potential for helping facilitate more individually-optimized interventional strategies.

关键词： Modeling Neuroimaging classification algorithms Training Prediction algorithms Alzheimer's disease Machine learning algorithms Machine learning ordinal learning health care imprecise labels

来源：评论

学校读者我要写书评

暂无评论

Dimensionality Reduction for Partial Label Learning: A Unified and Adaptive Approach

引用

IEEE TRANSACTIONS ON KNOWLEDGE AND DATA ENGINEERING 2024年第8期36卷 3765-3782页

作者： Yu, Xiang-Ru Wang, Deng-Bao Zhang, Min-Ling Southeast Univ Sch Cyber Sci & Engn Nanjing 210096 Peoples R China Southeast Univ Key Lab Comp Network & Informat Integrat Minist Educ Nanjing 210018 Peoples R China Southeast Univ Sch Comp Sci & Engn Nanjing 210096 Peoples R China

Partial label learning learns from instances with weak supervision, where each instance is associated with a set of candidate labels, among which only one is valid. Recently, dimensionality reduction has emerged as an effective preprocessing strategy to improve generalization performance. Existing approaches mainly tackle this problem through supervised or unsupervised dimensionality reduction. However, the former requires ground-truth labels, which are concealed in candidate label sets. Consequently, methods in this line may suffer from overfitting due to false positive labels in candidate label set. Conversely, the latter overlooks weakly supervised information in training instances, leading to performance degradation. In this paper, we propose an approach called partial label Dimensionality Reduction via Adaptive Weight (Draw) to leverage the strengths of Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA). Specifically, our approach tends to exploit unsupervised and data-driven nature of PCA to capture underlying structure of instances in initial stage. As the ground-truth label is gradually identified, our method increasingly relies on the discriminative ability of LDA to enhance the separation between different classes. Through extensive experiments on diverse partial label datasets, we validate that the proposed dimensionality reduction approach significantly improves classification performance of well-established partial label learning algorithms.

关键词： Dimensionality reduction Training Principal component analysis Faces Supervised learning Covariance matrices classification algorithms linear discriminant analysis partial label learning principal component analysis

来源：评论

学校读者我要写书评

暂无评论

In-Line Measurement of Particle Weight of Pulses Such as Chickpea by Piezoelectric Sensor

引用

IEEE ACCESS 2024年 12卷 156401-156411页

作者： Cankaya, Nihat Necmettin Erbakan Univ TR-42140 Konya Turkiye

Piezoelectric-based sensors are used as well as optical sensors for homogeneous sowing of seeds in the fields and for calculating the grain loss rate in harvesting and cleaning with harvesters in agricultural production. In this study, a piezoelectric-based sensor design has been developed that can measure properties such as size and weight during the storage of grains in silos in addition to these applications. Two different impact sensors using piezoelectric disks and piezoelectric ceramic elements were designed. Impact experiments were performed with chickpea grains in three different weight and size classes, and the energies of the electrical signals generated by the piezo elements because of the impact were calculated. This energy data was used in the training of ten various classification algorithms and their prediction success was calculated. While the highest classification accuracy for sensor data using piezo disk elements was obtained with the neural network algorithm with 92%, the highest classification accuracy for sensor data using piezo ceramic elements was obtained with the neural network and adaptive boosting algorithms with 68%. These data show that the sensor design using piezo disk elements could be used to measure the weights of grains such as chickpea during storage in silos.

关键词： classification algorithms ceramic ceramic disc particle weight particle weight piezoelectric piezoelectric particle weight piezoelectric

来源：评论

学校读者我要写书评

暂无评论

Ensemble Learning Approach With Class Rotation for Three-Dimensional classification on Direction-of-Arrival Estimation

引用

IEEE ACCESS 2022年 10卷 108185-108193页

作者： Mendoza-Velazquez, Israel Perez-Meana, Hector Haneda, Yoichi Univ Electrocommun Grad Sch Informat Tokyo 1828585 Japan Inst Politecn Nacl ESIME Culhuacan Mexico City 04260 DF Mexico

Over the past decade, neural networks have been widely used for direction-of-arrival (DoA) estimation owing to their high accuracy in noisy and reverberant environments. Classes of singular-model classifiers generally correspond to discretized DoA candidate angles, which in the case of a three-dimensional estimation, are bounded by the grid derived from uniform sampling over the unit sphere. Motivated by this, we propose an ensemble learning approach for classification tasks to improve estimation accuracy, as the ensemble criterion outperforms the individual criterion. First, individual networks that make up the ensemble differ slightly by grid rotation according to the Euler rotation theorem to complement discrete directional information. Score fusion was performed in the spherical harmonic domain for a more stable ensemble classification because the grid rotation also involves an angular mismatch. Moreover, to achieve a more accurate DoA estimation, interpolation over the fused scores was performed. Performance analysis and a comparison of state-of-the-art parametric and deep learning-based methods in several acoustic situations were conducted to determine the accuracy of the ensemble and to analyze the gradual angular error reduction as a function of noise and reverberation levels as more networks were added.

关键词： Direction-of-arrival estimation Estimation Harmonic analysis Learning systems classification algorithms Spectrogram Multiple signal classification Performance analysis Neural networks Reverberation Noise measurement Convolutional recurrent neural networks spherical harmonics three-dimensional microphone arrays direction of arrival ensemble learning

来源：评论

学校读者我要写书评

暂无评论

Robust Stacking Ensemble Model for Traffic Sign Detection and Recognition

引用

IEEE ACCESS 2024年 12卷 178941-178950页

作者： Wang, Yichen Wang, Qianjin Tsinghua Univ Shenzhen Int Grad Sch Div Logist & Transportat Shenzhen 518055 Peoples R China Jiangsu Ocean Univ Sch Comp Engn Lianyungang 222005 Jiangsu Peoples R China

As autonomous driving technology rapidly advances, the accurate detection and classification of traffic signs have become pivotal in ensuring driver safety and supporting the evolution of autonomous vehicles. Nonetheless, individual models possess inherent limitations. YOLOv8 is renowned for its swift detection capabilities and proficiency in identifying distant objects. However, due to the constraints imposed by its grid cell architecture, it exhibits suboptimal performance in detecting small, close-range targets. Conversely, Mask R-CNN demonstrates high precision in the detection of objects at close range, yet there remains a need for improvement in terms of distant object detection and the speed of detection. To overcome these obstacles, we introduce a novel model which integrates the strengths of Mask Region-based Convolutional Neural Network (Mask R-CNN) and the YOLOv8 model using a stacking ensemble technique. Our model was evaluated on the CCTSDB dataset and the MTSD dataset, demonstrating superior performance across various conditions. The experimental results on the MTSD dataset show a 3.63% improvement in mean Average Precision (mAP) and a 2.35% increase in Frames Per Second (FPS) compared to the Mask R-CNN, achieving a 3.20% increase in mAP over the YOLOv8. Moreover, the proposed model exhibited notable precision in challenging scenarios such as ultra-long-distance detections, shadow occlusions, motion blur, and complex environments with diverse sign categories. These findings not only showcase the model's robustness but also serve as a cornerstone in propelling the evolution of intelligent transportation systems and autonomous driving technology.

关键词： Feature extraction Accuracy Image recognition Target recognition Predictive models Mathematical models Deep learning classification algorithms YOLO Data models Traffic sign detection stacking ensemble model mask region-based convolutional neural network (Mask R-CNN) YOLOv8

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：