检索结果-南通市图书馆

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

基于resnet-18的磁瓦缺陷检测

佳木斯大学学报（自然科学版） 2023年第2期41卷 136-139,143页

作者：周孟然吴长志安徽理工大学电气与信息工程学院安徽淮南232001 安徽理工大学人工智能学院安徽淮南232001

由于磁瓦在实际生产中会产生多种类型的缺陷,针对磁瓦生产的表面检测高质量要求,提出一种基于resnet-18的磁瓦缺陷检测方法。首先通过构造磁瓦缺陷的不同类型数据集,再将数据集通过图像增强等方式扩充数据集进行预处理;引用resnet-18作... 详细信息

由于磁瓦在实际生产中会产生多种类型的缺陷,针对磁瓦生产的表面检测高质量要求,提出一种基于resnet-18的磁瓦缺陷检测方法。首先通过构造磁瓦缺陷的不同类型数据集,再将数据集通过图像增强等方式扩充数据集进行预处理;引用resnet-18作为训练所用的网络,设置好网络模型参数;最后通过网络训练完成对磁瓦缺陷的分类及标注。实验结果表明:扩充后数据集对比于原始数据集和其他方法,采用的方法中模型测试准确率更高,搭建的resnet-18网络具有更好的鲁棒性和泛化性,证明了可以满足磁瓦缺陷检测中的可能性。

关键词：磁瓦缺陷检测图像增强 resnet-18

基于改进resnet-18模型的弧齿锥齿轮安装位置偏差分类识别方法研究

在线全文

学校读者我要写书评

暂无评论

农业装备与车辆工程 2024年第3期62卷 96-100页

作者：刘小凯边骥轩张江勇董岑鑫王卫军广州先进技术研究所机器人与智能装备中心广东广州511458 中国北方车辆研究所北京100072

针对现有的弧齿锥齿轮安装位置偏差分类识别效率低的问题,使用改进的resnet-18神经网络构建轻量级弧齿锥齿轮安装位置偏差分类识别模型,改进的网络修改了resnet-18网络第一卷积层,并引入了通道与空间注意力机制。实验结果表明,改进ResNe... 详细信息

针对现有的弧齿锥齿轮安装位置偏差分类识别效率低的问题,使用改进的resnet-18神经网络构建轻量级弧齿锥齿轮安装位置偏差分类识别模型,改进的网络修改了resnet-18网络第一卷积层,并引入了通道与空间注意力机制。实验结果表明,改进resnet-18较原网络对于齿面接触印痕图像的分类准确率提高了3.23%,损失值降低了0.04。该方法为齿轮安装偏差识别提供了新路径,对弧齿锥齿轮总成的安装调整作业提供了一定的指导。

关键词：弧齿锥齿轮接触印痕安装位置偏差注意力机制 resnet-18

基于resnet-18和集成学习方法的坑洼道路检测与识别

在线全文

学校读者我要写书评

暂无评论

内蒙古科技与经济 2024年第16期 108-112页

作者：李奇冯瑜沈阳理工大学理学院辽宁沈阳110000

随着图像处理技术的不断进步,特别是深度学习在这一领域的广泛运用,图像识别技术已实现了显著的发展。文章探讨了如何利用深度学习和集成学习方法对坑洼道路进行检测和识别。通过resnet-18模型进行图像特征提取,最终输出512个特征。应... 详细信息

随着图像处理技术的不断进步,特别是深度学习在这一领域的广泛运用,图像识别技术已实现了显著的发展。文章探讨了如何利用深度学习和集成学习方法对坑洼道路进行检测和识别。通过resnet-18模型进行图像特征提取,最终输出512个特征。应用主成分分析(PCA)进行降维,设置方差总解释率为80%,从而得到58个关键特征。采用三种集成学习方法(GBDT、AdaBoost和随机森林)进行分类识别。通过特征重要性、测试集上的性能和学习曲线分析得出:AdaBoost和随机森林对坑洼道路检测任务更为适合,特别是在数据量有限的情况下,随机森林表现出了较强的性能,而AdaBoost在数据量增加时能够显著提升性能。

关键词： resnet-18 集成学习坑洼识别道路检测

同方期刊数据库博看期刊评论

在线全文

学校读者我要写书评

暂无评论

基于双重注意力机制的鱼类体表病理识别方法

华中农业大学学报 2025年

作者：王一非袁涛吴鹏飞华中农业大学信息学院农业农村部智慧养殖技术重点实验室

为提高鱼类体表病理识别精确度及识别效率，解决当前识别过程中对人工依赖严重且识别准确性低等问题，根据4种发病率高且对鱼类危害大的鱼病构建鱼类体表病理数据集，基于resnet-18模型进行改进优化，融合空间注意力和SE通道注意力双重... 详细信息

为提高鱼类体表病理识别精确度及识别效率，解决当前识别过程中对人工依赖严重且识别准确性低等问题，根据4种发病率高且对鱼类危害大的鱼病构建鱼类体表病理数据集，基于resnet-18模型进行改进优化，融合空间注意力和SE通道注意力双重注意力机制构建出高识别精度的DBA＿resnet-18模型，并基于该模型开发了鱼病实时智能识别可视化系统。改进后模型将SE通道注意力模块添加在网络中部，在网络尾部引入了空间注意力机制。测试结果显示，DBA＿resnet-18模型对鱼类体表病理分类准确率达到了96.75%，相比于常见的模型resnet-18、resnet-34、resnet-50、resnet-101、Swin Transformer、VGG-16、VGG-19和AlexNet分别高出1.71、2.12、2.37、2.83、2.51、2.23、2.50和3.53百分点。研究结果表明，本研究提出的模型及鱼病智能识别可视化系统能够对不同鱼类体表病理进行快速、精确的分类识别，实现了鱼病识别系统的智能化，可用于实际环境中诊断鱼类体表病理类型。

关键词：鱼类体表病理深度学习注意力机制 resnet-18 鱼病智能识别

Convolutional Bi-LSTM Based Human Gait Recognition Using Video Sequences

在线全文

同方期刊数据库

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2021年第8期68卷 2693-2709页

作者： Javaria Amin Muhammad Almas Anjum Muhammad Sharif Seifedine Kadry Yunyoung Nam ShuiHua Wang University of Wah Wah Cantt47040Pakistan National University of Technology(NUTECH) Islamabad44000Pakistan COMSATS University Islamabad Wah CampusWah CanttPakistan Faculty of Applied Computing and Technology Noroff University CollegeKristiansandNorway Department of Computer Science and Engineering Soonchunhyang UniversityAsan31538Korea Department of Mathematics University of LeicesterLeicesterUK

Recognition of human gait is a difficult assignment,particularly for unobtrusive surveillance in a video and human identification from a large ***,a method is proposed for the classification and recognition of different types of human *** proposed approach is consisting of two *** phase I,the new model is proposed named convolutional bidirectional long short-term memory(Conv-BiLSTM)to classify the video frames of human *** this model,features are derived through convolutional neural network(CNN)named resnet-18 and supplied as an input to the LSTM model that provided more distinguishable temporal *** phase II,the YOLOv2-squeezeNet model is designed,where deep features are extricated using the fireconcat-02 layer and fed/passed to the tinyYOLOv2 model for recognized/localized the human gaits with predicted *** proposed method achieved up to 90%correct prediction scores on CASIA-A,CASIA-B,and the CASIA-C benchmark *** proposed method achieved better/improved prediction scores as compared to the recent existing works.

关键词： Bi-LSTM YOLOv2 open neural network resnet-18 gait squeezeNet

Deep Learning-Based Model for Detection of Brinjal Weed in the Era of Precision Agriculture

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

Computers, Materials & Continua 2023年第10期77卷 1281-1301页

作者： Jigna Patel Anand Ruparelia Sudeep Tanwar Fayez Alqahtani Amr Tolba Ravi Sharma Maria Simona Raboaca Bogdan Constantin Neagu Department of Computer Science and Engineering Institute of TechnologyNirma UniversityAhmedabad382481India Software Engineering Department College of Computer and Information SciencesKing Saud UniversityRiyadh12372Saudi Arabia Computer Science Department Community CollegeKing Saud UniversityRiyadh11437Saudi Arabia Centre for Inter-Disciplinary Research and Innovation University of Petroleum and Energy StudiesDehradun248001India Doctoral School University Politehnica of BucharestBucharest060042Romania National Research and Development Institute for Cryogenic and Isotopic Technologies-ICSI Rm ValceaRamnicu Valcea240050Romania Power Engineering Department Gheorghe Asachi Technical University of IasiIasi700050Romania

The overgrowth of weeds growing along with the primary crop in the fields reduces crop *** solutions like hand weeding are labor-intensive,costly,and time-consuming;farmers have used *** application of herbicide is effective but causes environmental and health ***,Precision Agriculture(PA)suggests the variable spraying of herbicides so that herbicide chemicals do not affect the primary *** by the gap above,we proposed a Deep Learning(DL)based model for detecting Eggplant(Brinjal)weed in this *** key objective of this study is to detect plant and non-plant(weed)parts from crop *** the help of object detection,the precise location of weeds from images can be *** dataset is collected manually from a private farm in Gandhinagar,Gujarat,*** combined approach of classification and object detection is applied in the proposed *** Convolutional Neural Network(CNN)model is used to classify weed and non-weed images;further DL models are applied for object *** have compared DL models based on accuracy,memory usage,and Intersection over Union(IoU).resnet-18,YOLOv3,CenterNet,and Faster RCNN are used in the proposed *** outperforms all other models in terms of accuracy,i.e.,88%.Compared to other models,YOLOv3 is the least memory-intensive,utilizing 4.78 GB to evaluate the data.

关键词： Precision Agriculture Deep Learning brinjal weed detection resnet-18 YOLOv3 CenterNet Faster RCNN

基于大数据迁移学习的灰岩地区排水孔淤堵自动识别技术

维普期刊数据库评论

在线全文

维普期刊数据库

学校读者我要写书评

暂无评论

现代隧道技术 2021年第4期58卷 37-47页

作者：李鹏举郑方坤吕建兵吴维俊刘锋陈贡发中交广连高速公路投资发展有限公司清远511518 中交四航工程研究院有限公司广州510220 广东工业大学土木与交通工程学院广州510006

隧道排水管淤堵或者失效,会危及边坡的稳定和公路的交通安全。目前缺乏对排水孔淤堵分类的研究。排水孔淤堵分类对排水管道的后期养护具有重大意义。为探索排水孔淤堵智能化检测方法,文章依托广连高速高峰隧道等粤北实体工程,研究了一... 详细信息

隧道排水管淤堵或者失效,会危及边坡的稳定和公路的交通安全。目前缺乏对排水孔淤堵分类的研究。排水孔淤堵分类对排水管道的后期养护具有重大意义。为探索排水孔淤堵智能化检测方法,文章依托广连高速高峰隧道等粤北实体工程,研究了一种采用模型迁移的方法,将排水孔图像数据输入预训练卷积神经网络进行训练,以对新图像进行分类。在采集的排水孔图像数据集上进行试验,对比了三种不同网络模型对该数据集的准确率,结果表明,使用resnet-18进行排水孔淤堵分类识别,准确率可达到93%,实现了对排水孔淤堵状态的有效分类,并且随着日后数据集的扩大,识别准确率将会有进一步的提高。

关键词：排水孔淤堵分类卷积神经网络 resnet-18 Softmax

Research on the Monitoring Method of Rice Blast Disease at Canopy Scale Based on UAV Multispectral Imaging

在线全文

学校读者我要写书评

暂无评论

Research on the Monitoring Method of Rice Blast Disease at C...

Sorting and Identification Method of Camellia Seeds Based on Deep Learning

第43届中国控制会议

作者： Haoran Su Yi Pan Linlin Qin Gang Wu Chun Shi YiKun Wang Sheng Hua ShouWei Luo Department of automation University of Science and Technology of China Anhui Xiyangyang Agricultural Technology Co. Ltd

Rice ranks first in both sown area and total production among domestic grain crops, making it the most important food crop in China. Rice blast, also known as rice smut, is one of the most severe diseases in rice production, posing a significant challenge to rice farmers. To address this issue, a method has been developed for detecting the severity of rice blast using unmanned aerial vehicle(UAV) based on low altitude remote sensing technology and multi-spectral imaging technology. It involved collecting multi-spectral images of rice fields using UAV and extracting 18 spectral features(4 spectral bands + 14vegetation indices) and 32 texture features from these images. Pearson correlation analysis and recursive feature elimination algorithm were used to eliminate redundant features and identify sensitive feature combinations. These feature combinations were used as inputs for the RF and SVM algorithms to construct a model for the severity of rice blast disease. Simultaneously,the resnet-18 convolutional neural network was trained using the R, G, B, and NIR bands as inputs. The results showed that the fusion of spectral and texture features can effectively improve the recognition rate of machine learning models. The convolutional neural network model based on resnet-18 achieved the highest classification accuracy(OA=89.31%) for rice blast disease, outperforming the support vector machine classifier(OA=82.46%) and random forest classifier(OA=75.81%). It has been demonstrated that UAV multispectral imaging technology can accurately detect rice blast disease, while also offering the advantages of speed and non-destructiveness.

关键词： UAV multispectral images rice blast machine learning deep learning resnet-18

来源： cnki会议评论

在线全文

cnki会议

学校读者我要写书评

暂无评论

Sorting and Identification Method of Camellia Seeds Based on...