检索结果-内蒙古大学图书馆

Conference on computer vision and pattern recognition (CVPR)

作者： Shaoxin Li Junliang Xing Zhiheng Niu Shiguang Shan Shuicheng Yan Department of Electrical and Computer Engineering National University of Singapore Singapore Key Lab of Intelligent Information Processing of Chinese Academy of Sciences (CAS) Institute of Computing Technology CAS Beijing China National Laboratory of Pattern Recognition Institute of Automation CAS Beijing China

ISBN: (纸本)9781467369657

One key challenge of facial trait recognition is the large non-rigid appearance variations due to some irrelevant real world factors, such as viewpoint and expression changes. In this paper, we explore how the shape information, i.e. facial landmark positions, can be explicitly deployed into the popular Convolutional Neural Network (CNN) architecture to disentangle such irrelevant non-rigid appearance variations. First, instead of using fixed kernels, we propose a kernel adaptation method to dynamically determine the convolutional kernels according to the spatial distribution of facial landmarks, which helps learning more robust features. Second, motivated by the intuition that different local facial regions may demand different adaptation functions, we further propose a tree-structured convolutional architecture to hierarchically fuse multiple local adaptive CNN subnetworks. Comprehensive experiments on WebFace, Morph II and MultiPIE databases well validate the effectiveness of the proposed kernel adaptation method and tree-structured convolutional architecture for facial trait recognition tasks, including identity, age and gender recognition. For all the tasks, the proposed architecture consistently achieves the state-of-the-art performances.

关键词： Kernel Shape Face Face recognition Robustness Neural networks computer architecture

来源：评论

学校读者我要写书评

暂无评论

A Novel Ordinal Regression Method with Minimum Class Variance Support Vector Machine

A Novel Ordinal Regression Method with Minimum Class Varianc...

引用

2015 International Conference on Materials Engineering and Information Technology Applications(MEITA 2015)

作者： Jinrong Hu Xiaoming Wang Zengxi Huang School of Computer and Soft Engineering Xihua University Key Laboratory of Pattern Recognition and Intelligent Information Processing Chengdu University

ISBN: (纸本)9781510812055

In the paper, we propose a novel ordinal regression method called minimum class variance support vector ordinal regression(MCVSVOR). MCVSVOR is derived from minimum class variance support vector machine(MCVSVM) which is a variant of SVM, and so inherits the latter's characteristics such as taking the distribution of the categories into consideration and good generalization performance. Finally, the experimental results validate the effectiveness of MCVSVOR and indicate its superior generalization performance over SVOR.

关键词： Machine learning Ordinal regression Support vector machine Support vector ordinal regression

来源：评论

学校读者我要写书评

暂无评论

Sampling and surface reconstruction of large scale point cloud 13

Sampling and surface reconstruction of large scale point clo...

引用

13th ACM SIGGRAPH International Conference on Virtual-Reality Continuum and Its Applications in Industry, VRCAI 2014

作者： Li, Er Zhang, Xiaopeng Chen, Yanyun State Key Laboratory of Computer Science Institute of Software CAS Beijing China National Laboratory of Pattern Recognition Institute of Automation CAS Beijing China

ISBN: (纸本)9781450332545

In this paper, we propose a new approach on sampling and surface reconstruction of large-scale point cloud data. The sampling method is for huge point cloud data using spatial curve order and the surface reconstruction approach being based on witness complex theory. The approach first reorders the point cloud according to the spatial curve order and then sequential samples the ordered data. The technique preserves the spatial characteristic of the point cloud data well, and it is also suitable for out-of-core implementation. After the sampling, we use witness complex theory to reconstruct a manifold triangle surface from sampling data under the constraint of original data. Experiments demonstrate that the proposed method improves the topological consistency of the reconstruction result. Copyright © ACM.

关键词： Surface reconstruction

来源：评论

学校读者我要写书评

暂无评论

Road Segmentation via Iterative Deep Analysis

Road Segmentation via Iterative Deep Analysis

引用

IEEE International Conference on Robotics and Biomimetics

作者： Xiang Chen Yu Qiao Student at Shenzhen Key Laboratory of Computer Vision and Pattern Recognition Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences Address: 1068 Xueyuan Avenue Shenzhen University Town Shenzhen P.R.China Researcher at Shenzhen Key Laboratory of Computer Vision and Pattern Recognition Shenzhen Institutes of Advanced Technology Chinese Academy of Sciences Address: 1068 Xueyuan Avenue Shenzhen University Town Shenzhen P.R.China

ISBN: (纸本)9781467396769

Nowadays, people are increasingly concerned about the safety of traffic systems. Road segmentation and recognition is a fundamental problem in perceiving traffic environments and serve as the basis for self-driving cars. In this paper, inspired by an iterative deep analysis thinking, we propose a novel method which is able to learning powerful features step by step, and solve the optimal precision by balancing local and global information to conduct pixel-level classification for road segmentation. Firstly, we introduce an iterative deep analysis thinking which shows that how to design a strong and robustness deep model from failure experience. Secondly, we choose a powerful global features learning network as basis to create a novel framework for our task. Meanwhile, we employ the patch and multi-scale pyramid as input to enhance local features learning. We conduct experiments on three datasets from KITTI vision Benchmark, namely UU, UM, UMM. The experimental results demonstrate that our proposed method obtains comparable performance with state-of-the-art methods on these datasets.

关键词： recognition fundamental datasets

来源：评论

学校读者我要写书评

暂无评论

Safety and health monitoring of bridge construction based on wireless sensor network

International Journal of Multimedia and Ubiquitous Engineeri...

引用

International Journal of Multimedia and Ubiquitous Engineering 2014年第3期9卷 257-268页

作者： Li, Xiaoling Yuan, Jimin School of Computer Science and Technology Chengdu University Chengdu 610106 China University Key Laboratory of Pattern Recognition and Intelligent information Processing Chengdu 610106 China The Department Computer Engineering Chengdu Aeronautic Vocational and Technical College Chengdu 610000 China

The existing safety and health monitoring methods for bridge construction are mainly manual monitoring and wired monitoring with many disadvantages, such as low efficiency, poor accuracy, great implementation difficulty, short bridge age, etc. To solve such problems, the paper provides a design scheme for the bridge construction safety and health monitoring System based on wireless sensor networks. In the system, various sensors are mainly used to collect data about the wind, temperature, traffic load, humidity, static position, static displacement and external load response, and then send to the host by route and gateway nodes. After data analysis and processing, the safety of the bridge is evaluated to help the personnel make correct bridge safety precautions. With such advantages as low power demand, wireless transmission of monitoring data, etc., the system is suitable for large range of promotion. © 2014 SERSC.

关键词： Data handling

来源：评论

学校读者我要写书评

暂无评论

When Face recognition Meets with Deep Learning: An Evaluation of Convolutional Neural Networks for Face recognition

When Face Recognition Meets with Deep Learning: An Evaluatio...

引用

International Conference on computer vision Workshops (ICCV Workshops)

作者： Guosheng Hu Yongxin Yang Dong Yi Josef Kittler William Christmas Stan Z. Li Timothy Hospedales Centre for Vision Speech and Signal Processing University of Surrey UK Indicates equal contribution LEAR team Inria Grenoble Rhone-Alpes Montbonnot France Electronic Engineering and Computer Science Queen Mary University of London UK Chinese Academy of Sciences Center for Biometrics and Security Research & National Laboratory of Pattern Recognition China

Deep learning, in particular Convolutional Neural Network (CNN), has achieved promising results in face recognition recently. However, it remains an open question: why CNNs work well and how to design a 'good' architecture. The existing works tend to focus on reporting CNN architectures that work well for face recognition rather than investigate the reason. In this work, we conduct an extensive evaluation of CNN-based face recognition systems (CNN-FRS) on a common ground to make our work easily reproducible. Specifically, we use public database LFW (Labeled Faces in the Wild) to train CNNs, unlike most existing CNNs trained on private databases. We propose three CNN architectures which are the first reported architectures trained using LFW data. This paper quantitatively compares the architectures of CNNs and evaluates the effect of different implementation choices. We identify several useful properties of CNN-FRS. For instance, the dimensionality of the learned features can be significantly reduced without adverse effect on face recognition accuracy. In addition, a traditional metric learning method exploiting CNN-learned features is evaluated. Experiments show two crucial factors to good CNN-FRS performance are the fusion of multiple CNNs and metric learning. To make our work reproducible, source code and models will be made publicly available.

关键词： Face recognition Face Databases Measurement Training Object recognition Convolutional codes

来源：评论

学校读者我要写书评

暂无评论

Global investments in pandemic preparedness and COVID-19: development assistance and domestic spending on health between 1990 and 2026

引用

The Lancet Global Health 2023年第3期11卷 e385-e413页

作者： Micah, Angela E. Bhangdia, Kayleigh Cogswell, Ian E. Lasher, Dylan Lidral-Porter, Brendan Maddison, Emilie R. Nguyen, Trang Nhu Ngoc Patel, Nishali Pedroza, Paola Solorio, Juan Stutzman, Hayley Tsakalos, Golsum Wang, Yifeng Warriner, Wesley Zhao, Yingxi Zlavog, Bianca S. Abbafati, Cristiana Abbas, Jaffar Abbasi-Kangevari, Mohsen Abbasi-Kangevari, Zeinab Abdelmasseh, Michael Abdulah, Deldar Morad Abedi, Aidin Abegaz, Kedir Hussein Abhilash, E.S. Aboagye, Richard Gyan Abolhassani, Hassan Abrigo, Michael R.M. Abubaker Ali, Hiwa Abu-Gharbieh, Eman Adem, Mohammed Hussien Afzal, Muhammad Sohail Ahmadi, Ali Ahmed, Haroon Ahmed Rashid, Tarik Aji, Budi Akbarialiabad, Hossein Akelew, Yibeltal Al Hamad, Hanadi Alam, Khurshid Alanezi, Fahad Mashhour Alanzi, Turki M. Al-Hanawi, Mohammed Khaled Alhassan, Robert Kaba Aljunid, Syed Mohamed Almustanyir, Sami Al-Raddadi, Rajaa M. Alvis-Guzman, Nelson Alvis-Zakzuk, Nelson J. Amare, Azmeraw T. Ameyaw, Edward Kwabena Amini-Rarani, Mostafa Amu, Hubert Ancuceanu, Robert Andrei, Tudorel Anwar, Sumadi Lukman Appiah, Francis Aqeel, Muhammad Arabloo, Jalal Arab-Zozani, Morteza Aravkin, Aleksandr Y. Aremu, Olatunde Aruleba, Raphael Taiwo Athari, Seyyed Shamsadin Avila-Burgos, Leticia Ayanore, Martin Amogre Azari, Samad Baig, Atif Amin Bantie, Abere Tilahun Barrow, Amadou Baskaran, Pritish Basu, Sanjay Batiha, Abdul-Monim Mohammad Baune, Bernhard T. Berezvai, Zombor Bhardwaj, Nikha Bhardwaj, Pankaj Bhaskar, Sonu Boachie, Micheal Kofi Bodolica, Virginia Botelho, João Silva Botelho Braithwaite, Dejana Breitborde, Nicholas J.K. Busse, Reinhard Cahuana-Hurtado, Lucero Catalá-López, Ferrán Chansa, Collins Charan, Jaykaran Chattu, Vijay Kumar Chen, Simiao Chukwu, Isaac Sunday Dadras, Omid Dandona, Lalit Dandona, Rakhi Dargahi, Abdollah Debela, Sisay Abebe Denova-Gutiérrez, Edgar Desye, Belay Dharmaratne, Samath Dhamminda Diao, Nancy Doan, Linh Phuong Dodangeh, Milad dos Santos, Wendel Mombaque Doshmangir, Leila Dube, John Eini, Ebrahim El Sayed Zaki, Maysaa El Tantawi, Maha Enyew, Daniel Berhanie Institute for Health Metrics and Evaluation University of Washington Seattle WA United States Department of Applied Mathematics University of Washington Seattle WA United States Department of Health Metrics Sciences School of Medicine University of Washington Seattle WA United States Nuffield Department of Medicine University of Oxford Oxford United Kingdom Department of Juridical and Economic Studies La Sapienza University Rome Italy Antai College of Economics Shanghai Jiao Tong University Shanghai China Non-communicable Diseases Research Center Tehran University of Medical Sciences Tehran Iran Research Center for Immunodeficiencies Tehran University of Medical Sciences Tehran Iran Multiple Sclerosis Research Center Tehran University of Medical Sciences Tehran Iran Children’s Medical Center Tehran University of Medical Sciences Tehran Iran Department of Epidemiology and Biostatistics Tehran University of Medical Sciences Tehran Iran Tehran Heart Center Tehran University of Medical Sciences Tehran Iran Faculty of Medicine Tehran University of Medical Sciences Tehran Iran National Institute for Health Research Tehran University of Medical Sciences Tehran Iran Department of Pharmacology Tehran University of Medical Sciences Tehran Iran Department of Cardiology Tehran University of Medical Sciences Tehran Iran Sina Trauma and Surgery Research Center Tehran University of Medical Sciences Tehran Iran Digestive Diseases Research Institute Tehran University of Medical Sciences Tehran Iran Department of Neurology Tehran University of Medical Sciences Tehran Iran Social Determinants of Health Research Center Shahid Beheshti University of Medical Sciences Tehran Iran Department of Epidemiology Shahid Beheshti University of Medical Sciences Tehran Iran Department of Health Economics and Statistics Shahid Beheshti University of Medical Sciences Tehran Iran Department of Neurosurgery Shahid Beheshti University of Medical Sciences Tehran Iran Ophthalmic R

Background: The COVID-19 pandemic highlighted gaps in health surveillance systems, disease prevention, and treatment globally. Among the many factors that might have led to these gaps is the issue of the financing of national health systems, especially in low-income and middle-income countries (LMICs), as well as a robust global system for pandemic preparedness. We aimed to provide a comparative assessment of global health spending at the onset of the pandemic;characterise the amount of development assistance for pandemic preparedness and response disbursed in the first 2 years of the COVID-19 pandemic;and examine expectations for future health spending and put into context the expected need for investment in pandemic preparedness. Methods: In this analysis of global health spending between 1990 and 2021, and prediction from 2021 to 2026, we estimated four sources of health spending: development assistance for health (DAH), government spending, out-of-pocket spending, and prepaid private spending across 204 countries and territories. We used the Organisation for Economic Co-operation and Development (OECD)'s Creditor Reporting System (CRS) and the WHO Global Health Expenditure Database (GHED) to estimate spending. We estimated development assistance for general health, COVID-19 response, and pandemic preparedness and response using a keyword search. Health spending estimates were combined with estimates of resources needed for pandemic prevention and preparedness to analyse future health spending patterns, relative to need. Findings: In 2019, at the onset of the COVID-19 pandemic, US$9·2 trillion (95% uncertainty interval [UI] 9·1–9·3) was spent on health worldwide. We found great disparities in the amount of resources devoted to health, with high-income countries spending $7·3 trillion (95% UI 7·2–7·4) in 2019;293·7 times the $24·8 billion (95% UI 24·3–25·3) spent by low-income countries in 2019. That same year, $43·1 billion in development assistance was provided

关键词：

来源：评论

学校读者我要写书评

暂无评论

A new network traffic prediction algorithm based on discrete queue

引用

Journal of Computational Information Systems 2015年第12期11卷 4255-4261页

作者： Zhang, Hong Hu, Jun Shen, Yuncheng Key Laboratory of Pattern Recognition and Intelligent Information Processing Chengdu China College of Information Science and Technology Chengdu University Chengdu China College of Computer Science Sichuan University Chengdu China School of Traffic and Transportation Beijing Jiaotong University Beijing China Faculty of Computer Science Chengdu Normal University Chengdu China College of Information Science and Technology Zhaotong University Zhaotong China

In order to verify the network traffic decline because by node breakdown, this paper proposes a new type of prediction algorithm (Prediction algorithm based on Discrete-Queue for FARIMA model, PDF). At first, the mathematic formula for queuing situation and average queue length in steady state is derived with queuing theory in this algorithm based on discrete time, and the prediction method is established by FARIMA Model. Then, a simulation was conducted to research on the relationships between average queue length and service rate. The result shows that it has good adaptability, and the standard deviation between prediction traffic and original traffic is 10.18. ©, 2015, Binary Information Press. All right reserved.

关键词： Forecasting

来源：评论

学校读者我要写书评

暂无评论

A multi-dimensional perception of adaptive routing algorithm based OLSR

引用

Journal of Information and Computational Science 2015年第8期12卷 3165-3172页

作者： Zhang, Hong Shen, Yuncheng Li, Zhongwen Hu, Jun Key Laboratory of Pattern Recognition and Intelligent Information Processing Chengdu China College of Information Science and Technology Chengdu University Chengdu China College of Computer Science Sichuan University Chengdu China College of Information Science and Technology Zhaotong University Zhaotong China School of Traffic and Transportation Beijing Jiaotong University Beijing China Faculty of Computer Science Chengdu Normal University Chengdu China

Center nodes have a bigger load and burden with lots of routing in an Ad Hoc Network Model. Congestion of the nodes' packets has a great impact on network performance, especially in wireless networks. This paper proposes a multi-dimensional perception of adaptive routing algorithms based on the loading of MAC, in particular its utility. Through calculating the probabilities of packet arrival, distribution and intensity of interference, conicting nodes and adjusting the contracting strength of packets would optimize network performance. Our simulated experiments show that this algorithm improves the success rate of data transmission, reduces the delay of packet transmission and improves network throughputs. Copyright © 2015 Binary Information Press.

关键词： Network performance

来源：评论

学校读者我要写书评

暂无评论

Comparison of appearance-based and geometry-based bubble detectors

Lecture Notes in Computer Science (including subseries Lectu...

引用

Lecture Notes in computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 2014年 8671卷 610-617页

作者： Strokina, Nataliya Juránek, Roman Eerola, Tuomas Lensu, Lasse Zemčik, Pavel Kälviäinen, Heikki Tampere University of Technology Department of Signal Processing P.O. Box 527 Tampere33101 Finland Brno University of Technology Department of Computer Graphics and Multimedia Brno Czech Republic Lappeenranta University of Technology Machine Vision and Pattern Recognition Laboratory P.O. Box 20 Lappeenranta53851 Finland

Bubble detection is a complicated tasks since varying lighting conditions changes considerably the appearance of bubbles in liquid. The two common techniques to detect circular objects such as bubbles, the geometry-based and appearance-based approaches, have their advantages and weaknesses. The geometry-based methods often fail to detect small blob-like bubbles that do not match the used geometrical model, and appearance-based approaches are vulnerable to appearance changes caused by, e.g., illumination. In this paper, we compare a geometry-based concentric circular arrangements (CCA) and appearance-based sliding window methods as well as their combinations in terms of bubble detection, gas volume computation, and size distribution estimation. The best bubble detection performance was achieved with the sliding window method whereas the most precise volume estimate was produced by the CCA method. The combination of the two approaches gave only a minor advantage compared to the base methods. © Springer International Publishing Switzerland 2014.

关键词： Geometry

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：