检索结果-内蒙古大学图书馆

Road surface classification using accelerometer and speed data: evaluation of a convolutional neural network model

neural COMPUTING & APPLICATIONS 2023年第19期35卷 14183-14194页

作者： Sabapathy, Ashwin Biswas, Avik Danlaw Inc 1 PrimparkPrimrose Rd Bangalore 560025 India

Assessing the quality of road surfaces for maintenance has typically been an expensive exercise for local government agencies. This paper evaluates the feasibility of using accelerometer and speed data collected from on-board diagnostic (OBD-II) devices and labelled using the PASER system of road classification to classify road quality and identify poor stretches. An ordinal logistic model and a support vector machine (SVM) classifier were first trained on features created from raw data. The SVM and an artificial neural network (ANN) were then trained separately on the raw data parameters followed by a convolutional neural network (CNN) model. All models have comparable results in terms of overall classification accuracy on validation datasets (around 67%) but the ANN and CNN models were superior in their ability to correctly identify 'Poor' and 'Good' stretches. The CNN model had the best performance among all models evaluated. The approach presented offers a low-cost method to map poor roads and identify stretches that require more attention as an alternative to more expensive traditional methods.

关键词： On-board diagnostics (OBD-II) devices Pavement surface Evaluation and rating (PASER) Ordinal logistic model convolutional neural network model

来源：评论

学校读者我要写书评

暂无评论

Estimation of lung volume changes from frontal and lateral views of dynamic chest radiography using a convolutional neural network model: a computational phantom study

Estimation of lung volume changes from frontal and lateral v...

引用

Medical Imaging Conference - Physics of Medical Imaging

作者： Ishihara, Nozomi Tanaka, Rie Segars, William Paul Abadi, Ehsan Samei, Ehsan Kanazawa Univ Kanazawa Ishikawa 9200942 Japan Carl E Ravin Adv Imaging Labs Durham NC 27705 USA

ISBN: (纸本)9781510640207

We aimed to estimate respiratory changes in lung volumes (Alung volume) using frontal and lateral dynamic chest radiography (DCR) by employing a convolutional neural network (CNN) learning approach trained and tested using the four-dimensional (4D) extended cardiac-torso (XCAT) phantom. Twenty XCAT phantoms of males (5 normal, 5 overweight, and 5 obese) and females (5 normal) were generated to obtain 4D computed tomography (CT) of a virtual patient. XCAT phantoms were projected in frontal and lateral directions. We estimated lung volumes of the XCAT phantoms using CNN learning techniques. One dataset consisted of a right- or left-half frontal view, a lateral view, and ground truth (GT) knowledge of each phantom in the same respiratory phase. Alung volume were calculated by subtracting the lung volume estimated at the maximum exhale from that at the maximal inhale, and was compared with Alung volume calculated from the known GT. Alung volume was successfully estimated from frontal and lateral DCR images of XCAT phantoms by a CNN learning approach. There was a correlation for Alung volume between GT and estimation in both lungs. There were no significant differences in the estimation error between the right and left lungs, males and females, and males having different physiques. We confirmed that DCR has potential use in the estimation of Alung volume, which corresponds to vital capacity (VC) in pulmonary function tests (PFT). Pulmonary function could be assessed by DCR even in patients with infectious diseases who can't do PFT using a spirometer.

关键词： Dynamic chest radiography Flat-panel detector XCAT phantom Lung volume Respiratory changes convolutional neural network model Prediction method Simulation study

来源：评论

学校读者我要写书评

暂无评论

Toward efficient runs of nonlinear gyrokinetic simulations assisted by a convolutional neural network model recognizing wavenumber-space images

引用

NUCLEAR FUSION 2022年第8期62卷 086037-086037页

作者： Narita, E. Honda, M. Maeyama, S. Watanabe, T-H Natl Inst Quantum Sci & Thchnol Naka Ibaraki 3110193 Japan Kyoto Univ Grad Sch Engn Nishikyo Ku Kyoto 6158530 Japan Nagoya Univ Dept Phys Nagoya Aichi 4648602 Japan

A neural-network based innovative model recognizing the wavenumber space images has been developed to accurately forecast when the saturation of turbulent heat fluxes commences, i.e., the saturation time, in nonlinear gyrokinetic simulations. The wavenumber space images of the perturbed distribution function are focused on, which better represent the characteristics of turbulence. The model exploiting the state-of-the-art convolutional neural network model is capable of detecting minuscule differences between the images. Once the wavenumber space image is fed into the developed model, it can quickly and almost perfectly classify which phase of the turbulence evolution in nonlinear gyrokinetic simulations the image is in: the linearly and nonlinearly growing phases and the saturation phase. It can also predict the simulation time at which the image was processed with significantly high accuracy. The model enables us to forecast the saturation time of the gyrokinetic simulation in question by feeding an image at an early stage of the simulation and receiving the degree of progress toward the saturation. The ability of the model makes it possible to easily search out a desirable initial condition that rapidly conducts the simulation to a saturation phase. Such a pre-prediction model is important for running long time simulations on a large scale supercomputer like Fugaku in view of the efficient use of computational resources. In order to improve the predictive capability for the simulation that is going to be performed, several prediction models are trained by data with different major instabilities. The best predictor is selected to be in use based on the result of the pre-performed linear stability calculation with low computational cost.

关键词： turbulent transport nonlinear gyrokinetic simulations wavenumber-space convolutional neural network model deep learning large scale supercomputers

来源：评论

学校读者我要写书评

暂无评论

Expressway Project Cost Estimation With a convolutional neural network model

引用

IEEE ACCESS 2020年 8卷 217848-217866页

作者： Xue, Xiaojuan Jia, Yuanhua Tang, Yuanjie Beijing Jiaotong Univ Sch Traff & Transportat Beijing 100044 Peoples R China CCCC First Highway Consultants Co Ltd Xian 710075 Peoples R China

With the development of the economy, the prediction of expressway project costs has gained increasing research attention. In this study, based on the convolution neural network (CNN) algorithm, the prediction of the expressway construction cost was analyzed with respect to the conceptual design stage. By summarizing the existing research results, 10 new factors related to the bridge and tunnel are creatively introduced into the cost-prediction index of the expressway conceptual design stage. In addition, the data structure of the expressway project cost prediction is defined and a CNN model is established. Finally, the project information of 415 expressways in China collected in this study is used to verify the research results. The results of the case analysis show that the 10 new indexes related to the bridge and tunnel can improve the prediction accuracy of the model. In addition, the CNN model is more suitable for solving the high-dimensional nonlinear problem of expressway cost prediction than the conventional artificial-neural-network and regression-analysis models, and it can improve the prediction accuracy. The findings of this study can be used to devise financial plans in the early stage of expressway construction and facilitate cost management at the conceptual design stage to help investors acquire project funds in advance.

关键词： Predictive models Meters Licenses Analytical models Prediction algorithms Linear regression Geology Expressway project cost prediction convolutional neural network model artificial neural network regression analysis

来源：评论

学校读者我要写书评

暂无评论

Research on Recognition of Multiple Partial Discharge Sources in Switchgear Based on the Combination of GST-TEV and ResNet-18

引用

IET SCIENCE MEASUREMENT & TECHNOLOGY 2025年第1期19卷

作者： Wang, Feng Wang, Ning Zhong, Lipeng Chen, She Sun, Qiuqin Han, Xutao Xu, Puming Liu, Sanwei Peng, Tao Yan, Ying Deng, Xiao Hunan Univ Coll Elect & Informat Engn Changsha Peoples R China Xi An Jiao Tong Univ State Key Lab Elect Insulat & Power Equipment Xian Peoples R China State Grid Hunan Elect Power Co Ltd Res Inst Changsha Peoples R China

Switchgear can develop insulation defects due to electrical, thermal, and chemical stresses during manufacturing or prolonged operation. Moreover, as voltage levels rise, multiple insulation defects can coexist within the switchgear. Traditional partial discharge (PD) recognition methods often suffer from poor generalisation and low accuracy, limiting their practical applications. This paper proposes a method to identify multiple PD sources by combining generalised S-transform (GST) with the ResNet-18 network. PD tests confirm that the designed monitoring device effectively detects transient earth voltage (TEV) signals from diverse single and mixed insulation defects. Given the non-stationary nature of TEV signals, this paper employs the generalised S-transform (GST) for time-frequency analysis. The findings demonstrate that the GST method offers high time-frequency resolution, significantly improving the feature extraction of various partial discharge sources. Additionally, deep learning algorithms are employed to classify the time-frequency image dataset derived from GST-TEV. The results demonstrate that, compared to traditional manual feature extraction methods, the ResNet-18 network efficiently extracts GST-TEV features from both single and mixed partial discharge sources, achieving a recognition accuracy of 99.41%. This study provides new methods and theoretical support for identifying multiple partial discharge sources in switchgear.

关键词： convolutional neural network model GST-TEV multiple partial discharge sources switchgear pattern recognition

来源：评论

学校读者我要写书评

暂无评论

An Ensemble of Vision-Language Transformer-Based Captioning model With Rotatory Positional Embeddings

引用

IEEE ACCESS 2025年 13卷 59841-59865页

作者： Sathyanarayana, K. B. Naik, Dinesh Natl Inst Technol Karnataka Dept Informat Technol Surathkal 575025 Karnataka India JNN Coll Engn Dept Informat Sci & Engn Shivamogga 577204 Karnataka India

Image captioning is a dynamic and crucial research area focused on automatically generating image textual descriptions. Traditional models, primarily employing an encoder-decoder framework with convolutional neural networks (CNNs), often struggle to capture the complex spatial and sequential relationships inherent in visual data. This gap in performance underscores the necessity for more sophisticated solutions. The proposed work introduces a groundbreaking ensemble model that integrates CNN, Graph convolutional network (GCN), Bidirectional Long Short-Term Memory (BiLSTM), and Transformer architectures. Our approach achieves an outstanding 97% increase in CIDEr scores on the Flickr30K dataset and a remarkable 28.6% improvement on the Flickr8K dataset, thanks to the innovative implementation of Rotary Positional Encoding (RoPE). By strategically incorporating GCN and BiLSTM layers, our model adeptly captures essential relationships within the data. This groundbreaking research effectively addresses the challenges of image captioning, leveraging a powerful combination of advanced architectures. As a result, our model significantly enhances the generation of accurate and contextually rich captions, positioning it as a game-changer for automated image-to-text applications. The proposed Ensemble model with RoPE, achieved impressive performance on the Flickr8k and Flickr30k datasets, with scores of 80.62 and 95.0 for BLEU-1, 72.01 and 90.51 for BLEU-2, 63.12 and 81.24 for BLEU-3, 48.32 and 68.8 for BLEU-4, 74.26 and 81.89 for METEOR, 80.24 and 84.29 for ROUGE-L, 118.94 and 155.77 for CIDEr, and 48.7 and 39.0 for SPICE, respectively.

关键词： Transformers Feature extraction Computational modeling Computer architecture Bidirectional long short term memory Semantics Decoding Adaptation models Accuracy Data models Attention mechanisms bidirectional long short term memory model convolutional neural network model graph convolution network image caption generation positional embedding rotary positional embedding transformer

来源：评论

学校读者我要写书评

暂无评论

Automated lepidopteran pest developmental stages classification via transfer learning framework

引用

ENVIRONMENTAL ENTOMOLOGY 2024年第6期53卷 1062-1077页

作者： Qin, Wei-bo Abbas, Arzlan Abbas, Sohail Alam, Aleena Chen, De-hui Hafeez, Faisal Ali, Jamin Romano, Donato Chen, Ri-Zhao Jilin Agr Univ Coll Plant Protect 2888 Xincheng Rd Changchun 130118 Jilin Peoples R China Ayub Agr Res Inst Entomol Res Inst Faisalabad 37000 Punjab Pakistan BioRobot Inst St Anna Sch Adv Studies Viale Rinaldo Piaggio 34 I-56025 Pontedera Italy

The maize crop is highly susceptible to damage caused by its primary pests, which poses considerable challenges in manually identifying and controlling them at various larval developmental stages. To mitigate this issue, we propose an automated classification system aimed at identifying the different larval developmental stages of 23 instars of 4 major lepidopteran pests: the Asian corn borer, Ostrinia furnacalis (Guen & eacute;e;Lepidoptera: Crambidae), the fall armyworm, Spodoptera frugiperda (J.E. Smith;Lepidoptera: Noctuidae), the oriental armyworm, Mythimna separata (Walker;Lepidoptera: Noctuidae), and the tobacco cutworm, Spodoptera litura (Fabricius;Lepidoptera: Noctuidae). Employing 5 distinct convolutional neural network architectures-Convnext, Densenet121, Efficientnetv2, Mobilenet, and Resnet-we aimed to automate the process of identifying these larval developmental stages. Each model underwent fine-tuning using 2 different optimizers: stochastic gradient descent with momentum and adaptive moment estimation (Adam). Among the array of models tested, Densenet121, coupled with the Adam optimizer, exhibited the highest classification accuracy, achieving an impressive 96.65%. The configuration performed well in identifying the larval development stages of all 4 pests, with precision, recall, and F1 score evaluation indicators reaching 98.71%, 98.66%, and 98.66%, respectively. Notably, the model was ultimately tested in a natural field environment, demonstrating that Adam_Densenet121 model achieved an accuracy of 90% in identifying the 23 instars of the 4 pests. The application of transfer learning methodology showcased its effectiveness in automating the identification of larval developmental stages, underscoring promising implications for precision-integrated pest management strategies in agriculture.

关键词： convolutional neural network model transfer learning larval development Lepidoptera

来源：评论

学校读者我要写书评

暂无评论

Using optimal transport theory to optimize a deep convolutional neural network microscopic cell counting method

引用

MEDICAL & BIOLOGICAL ENGINEERING & COMPUTING 2023年第11期61卷 2939-2950页

作者： Ding, Yuanyuan Zheng, Yuanjie Han, Zeyu Yang, Xinbo Shandong Normal Univ Sch Informat Sci & Engn Jinan 250358 Shandong Peoples R China Shandong Univ Weihai Sch Math & Stat Weihai 264209 Shandong Peoples R China

Medical image processing has become increasingly important in recent years, particularly in the field of microscopic cell imaging. However, accurately counting the number of cells in an image can be a challenging task due to the significant variations in cell size and shape. To tackle this problem, many existing methods rely on deep learning techniques, such as convolutional neural networks (CNNs), to count cells in an image or use regression counting methods to learn the similarities between an input image and a predicted cell image density map. In this paper, we propose a novel approach to monitor the cell counting process by optimizing the loss function using the optimal transport method, a rigorous measure to calculate the difference between the predicted count map and the dot annotation map generated by the CNN. We evaluated our algorithm on three publicly available cell count benchmarks: the synthetic fluorescence microscopy (VGG) dataset, the modified bone marrow (MBM) dataset, and the human subcutaneous adipose tissue (ADI) dataset. Our method outperforms other state-of-the-art methods, achieving a mean absolute error (MAE) of 2.3, 4.8, and 13.1 on the VGG, MBM, and ADI datasets, respectively, with smaller standard deviations. By using the optimal transport method, our approach provides a more accurate and reliable cell counting method for medical image processing.

关键词： Cell counting Density regression Optimal transport convolutional neural network model

来源：评论

学校读者我要写书评

暂无评论

Fast H.266/VVC intra-coding by mode inheritance

引用

MULTIMEDIA TOOLS AND APPLICATIONS 2023年第23期82卷 36041-36065页

作者： Chen, Jiann-Jone Su, Jiann-Ann Natl Taiwan Univ Sci & Technol Dept Elect Engn Taipei 10607 Taiwan

The fifth-generation (5G) mobile networks pave a highway path for ultra-high-definition video communications and the newest versatile video coding standard, VVC/H.266, supporting 8K video coding, is best suited to offer media streaming applications over 5G networks, such as remote desktop, online streaming, cloud gaming, and other interactive media services. For a platform to provide better media-consuming experiences, it had to control quality and response in real-time. The VVC/H.266 adopts a quadtree plus multitype tree (QTMT) coding structure that requires exhaustive search operations, such that its time complexity is 18 times of the previous HEVC/H.265. To make practical application feasible, we proposed to quickly determine whether one coding unit (CU) resides on static regions, based on which the VVC coding controller can decide to inherit the co-located QTMT coding mode of a previously coded frame or not to reduce encoding time complexity. A subjective similarity measure, MS-SSIM, is used to determine CU static. In addition, a learned optical flow motion estimation (OFME) model is developed to measure motion activity to screen out false-positive results, such that BDBR can be kept small. By quickly locating static CUs and precisely screening out false-positive ones, the VVC encoding time complexity can be largely reduced while maintaining good quality. Experiments showed that the proposed method can save 42.34% of encoding time with 1.49% of BDBR increment, as compared with the default VTM 11.0 intra-coding. The percentage of static region blocks is found to be 61.32% on average from test video sequences.

关键词： Versatile video coding convolutional neural network model Learning optical flow VTM

来源：评论

学校读者我要写书评

暂无评论

Classification of mango disease using ensemble convolutional neural network

引用

SMART AGRICULTURAL TECHNOLOGY 2024年 8卷

作者： Bezabh, Yohannes Agegnehu Ayalew, Aleka Melese Abuhayi, Biniyam Mulugeta Demlie, Tensay Nigussie Awoke, Eshete Ayenew Mengistu, Taye Endeshaw Univ Gondar Dept Informat Technol Gondar Ethiopia Univ Gondar Gondar Ethiopia Jigjiga Univ Dept Informat Technol Jigjiga Ethiopia Univ Gondar Informat Tecnol Merawi Ethiopia

Mango is a highly significant fruit crop that thrives in a variety of agro-ecologies around the world. Mangoes are rich in vitamins and minerals. However, its yield is currently severely constrained due to disease and pest infestations. Thus, in order to improve mango fruit quality and productivity, illnesses and insect pests must be detected early on. In this study, we conceived and constructed a mango leaf disease detection mechanism utilizing an ensemble convolutional neural network approach. Healthy and diseased mango leaf images were manually obtained from main producing locations in Amhara Region for Merawi fruit and vegetable research identification. To improve the datasets, several pre-processing procedures (such as image resizing, noise reduction, and image augmentation) were used. To improve classification performance and meet the study 's purpose, various segmentation approaches such as k means and Mask R-CNN were applied. Furthermore, following pre-processing and segmentation, features of mango leaf images were retrieved using CNN to obtain important features. The classification model was then constructed using fully-connected layer classifiers on the retrieved features of mango leaf images. The ensemble proposed GoogLeNet and VGG16 based CNN model in the study encompasses various operations, including dataset collection, image preprocessing, noise removal, segmentation, data augmentation, feature extraction, and classification. Upon testing, the model demonstrated impressive performance with 99.87 % training classification accuracy, 99.72 % validation accuracy, and 99.21 % testing accuracy. This indicates the effectiveness of the ensemble approach in achieving high accuracy in image classification tasks.

关键词： Classification mango disease Digital imaging convolutional neural network model

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：