检索结果-内蒙古大学图书馆

Efficient parallel TLD on CPU-GPU Platform for Real-Time Tracking

KSII TRANSACTIONS ON INTERNET AND INFORMATION SYSTEMS 2020年第1期14卷 201-220页

作者： Chen, Zhaoyun Huang, Dafei Luo, Lei Wen, Mei Zhang, Chunyuan Natl Univ Def Technol Coll Comp Changsha 410073 Peoples R China Natl Key Lab Parallel & Distributed Proc Changsha 410073 Peoples R China Southwest Elect & Telecommun Technol Res Inst Chengdu 610000 Peoples R China

Trackers, especially long-term (LT) trackers, now have a more complex structure and more intensive computation for nowadays' endless pursuit of high accuracy and robustness. However, computing efficiency of LT trackers cannot meet the real-time requirement in various real application scenarios. Considering heterogeneous CPU-GPU platforms have been more popular than ever, it is a challenge to exploit the computing capacity of heterogeneous platform to improve the efficiency of LT trackers for real-time requirement. This paper focuses on TLD, which is the first LT tracking framework, and proposes an efficient parallel implementation based on OpenCL. In this paper, we firstly make an analysis of the TLD tracker and then optimize the computing intensive kernels, including Fern Feature Extraction, Fern Classification, NCC Calculation, Overlaps Calculation, Positive and Negative Samples Extraction. Experimental results demonstrate that our efficient parallel TLD tracker outperforms the original TLD, achieving the 3.92 speedup on CPU and GPU. Moreover, the parallel TLD tracker can run 52.9 frames per second and meet the real-time requirement.

关键词： TLD tracker Real-Time Heterogeneous Platform OpenCL parallel Optimizations

来源：评论

学校读者我要写书评

暂无评论

NTIRE 2022 Challenge on Perceptual Image Quality Assessment

NTIRE 2022 Challenge on Perceptual Image Quality Assessment

引用

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

作者： Gu, Jinjin Cai, Haoming Dong, Chao Ren, Jimmy S. Timofte, Radu Gong, Yuan Lao, Shanshan Shi, Shuwei Wang, Jiahao Yang, Sidi Wu, Tianhe Xia, Weihao Yang, Yujiu Cao, Mingdeng Heng, Cong Fu, Lingzhi Zhang, Rongyu Zhang, Yusheng Wang, Hao Song, Hongjian Wang, Jing Fan, Haotian Hou, Xiaoxia Sun, Ming Li, Mading Zhao, Kai Yuan, Kun Kong, Zishang Wu, Mingda Zheng, Chuanchuan Conde, Marcos, V Burchi, Maxime Feng, Longtao Zhang, Tao Li, Yang Xu, Jingwen Wang, Haiqiang Liao, Yiting Li, Junlin Xu, Kele Sun, Tao Xiong, Yunsheng Keshari, Abhisek Komal, Komal Thakur, Sadbhawana Jakhetiya, Vinit Subudhi, Badri N. Yang, Hao-Hsiang Chang, Hua-En Huang, Zhi-Kai Chen, Wei-Ting Kuo, Sy-Yen Dutta, Saikat Das, Sourya Dipta Shah, Nisarg A. Tiwari, Anil Kumar Shanghai AI Lab Shanghai Peoples R China Univ Sydney Sch Elect & Informat Engn Sydney NSW Australia Chinese Acad Sci Shenzhen Inst Adv Technol Beijing Peoples R China SenseTime Res Beijing Peoples R China Univ Wurzburg Wurzburg Germany Swiss Fed Inst Technol Zurich Switzerland Tsinghua Univ Tsinghua Shenzhen Int Grad Sch Beijing Peoples R China Tsinghua Univ Dept Automat Beijing Peoples R China UCL London England Netease Hangzhou Peoples R China Kuaishou Beijing Peoples R China Univ Wurzburg Comp Vis Lab Wurzburg Germany ByteDance Beijing Peoples R China Key Lab Parallel & Distributed Proc Beijing Peoples R China Indian Inst Technol Jammu Jagti Jammu & Kashmir India Indian Inst Technol Madras Chennai Tamil Nadu India Jadavpur Univ Kolkata W Bengal India Indian Inst Technol Jodhpur Jodhpur Rajasthan India Natl Taiwan Univ New Taipei Taiwan

ISBN: (数字)9781665487399

ISBN: (纸本)9781665487399

This paper reports on the NTIRE 2022 challenge on perceptual image quality assessment (IQA), held in conjunction with the New Trends in Image Restoration and Enhancement workshop (NTIRE) workshop at CVPR 2022. This challenge is held to address the emerging challenge of IQA by perceptual image processing algorithms. The output images of these algorithms have completely different characteristics from traditional distortions and are included in the PIPAL dataset used in this challenge. This challenge is divided into two tracks, a full-reference IQA track similar to the previous NTIRE IQA challenge and a new track that focuses on the no-reference IQA methods. The challenge has 192 and 179 registered participants for two tracks. In the final testing stage, 7 and 8 participating teams submitted their models and fact sheets. Almost all of them have achieved better results than existing IQA methods, and the winning method can demonstrate state-of-the-art performance.

关键词： Image quality Computer vision Conferences Computational modeling Market research Distortion Pattern recognition

来源：评论

学校读者我要写书评

暂无评论

Towards a Uniform Architecture for the Efficient Implementation of 2D and 3D Deconvolutional Neural Networks on FPGAs

Towards a Uniform Architecture for the Efficient Implementat...

引用

IEEE International Symposium on Circuits and Systems (IEEE ISCAS)

作者： Wang, Deguang Shen, Junzhong Wen, Mei Zhang, Chunyuan Natl Univ Def Technol Coll Comp Changsha 410073 Hunan Peoples R China Natl Univ Def Technol Natl Key Lab Parallel & Distributed Proc Changsha 410073 Hunan Peoples R China

ISBN: (纸本)9781728103976

Three-dimensional deconvolution is widely used in many computer vision applications. However, most previous works have only focused on accelerating 2D deconvolutional neural networks (DCNNs) on FPGAs, while the acceleration of 3D DCNNs has not been studied in depth as they have higher computational complexity and sparsity than 2D DCNNs. In this paper, we focus on the acceleration of both 2D and 3D DCNNs on FPGAs by proposing efficient schemes for mapping 2D and 3D DCNNs on a uniform architecture. By implementing our design on the Xilinx VC709 platform for four real-life 2D and 3D DCNNs, we can achieve up to 3.0 TOPS with high hardware efficiency. Comparisons with CPU and GPU solutions demonstrate that we can achieve an improvement of up to 63.3X in throughput relative to a CPU solution and an improvement of up to 8.3X in energy efficiency compared to a GPU solution.

关键词： Deconvolution Acceleration Computer architecture Field programmable gate arrays Two dimensional displays Three-dimensional displays Benchmark testing

来源：评论

学校读者我要写书评

暂无评论

A distributed Topic Model for Large-Scale Streaming Text 12th

A Distributed Topic Model for Large-Scale Streaming Text

引用

12th International Conference on Knowledge Science, Engineering and Management (KSEM)

作者： Li, Yicong Feng, Dawei Lu, Menglong Li, Dongsheng Natl Univ Def Technol Coll Comp Sci & Technol Changsha 410073 Peoples R China Natl Univ Def Technol Natl Key Lab Parallel & Distributed Proc Changsha 410073 Peoples R China

ISBN: (纸本)9783030295639;9783030295622

Learning topic information from large-scale unstructured text has attracted extensive attention from both the academia and industry. Topic models, such as LDA and its variants, are a popular machine learning technique to discover such latent structure. Among them, online variational hierarchical Dirichlet process (onlineHDP) is a promising candidate for dynamically processing streaming text. Instead of a static assignment in advance, the number of topics in onlineHDP is inferred from the corpus as the training process proceeds. However, when dealing with large scale streaming data it still suffers from the limited model capacity problem. To this end, we proposed a distributed version of the onlineHDP algorithm (named as DistHDP) in this paper, the training task is split into many sub-batch tasks and distributed across multiple worker nodes, such that the whole training process is accelerated. The model convergence is guaranteed through a distributed variation inference algorithm. Extensive experiments conducted on several real-world datasets demonstrate the usability and scalability of the proposed algorithm.

关键词： distributed topic model Hierarchical Dirichlet process Variational inference

来源：评论

学校读者我要写书评

暂无评论

TA-Update: An Adaptive Update Scheme with Tree-Structured Transmission in Erasure-Coded Storage Systems

引用

IEEE TRANSACTIONS ON parallel AND distributed SYSTEMS 2018年第8期29卷 1893-1906页

作者： Wang, Yijie Pei, Xiaoqiang Ma, Xingkong Xu, Fangliang Natl Univ Def Technol Coll Comp Natl Key Lab Parallel & Distributed Proc Changsha 410073 Hunan Peoples R China

Erasure coding has received considerable attentions due to the better tradeoff between the space efficiency and reliability. The frequent update of the stored data in the distributed storage systems has posed a new challenge for erasure codes: how to update the erasure-coded data in a general, efficient and adaptive way. However, existing update schemes of erasure codes are inadequate to meet these requirements, since their code-related update manners lead to a low generality, their star-structured data transmission manners lead to a low update efficiency, and their redo manners when encountering the node failure lead to a low adaptivity. In this paper, we propose an adaptive update scheme with the tree-structured transmission, called TA-Update, which consists of a code-independent update framework and three algorithms: the rack-aware tree construction algorithm, the top-down data processing algorithm and the rollback-based failure processing algorithm. For generality, we propose a code-independent update framework with the tree structure to support the MDS code with any coding parameter. For efficiency, a rack-aware tree construction algorithm is proposed to achieve the high available bandwidth, which organizes the data node and parity nodes as an update tree. Moreover, a top-down data processing algorithm is proposed to achieve the high transmission and computation efficiency, which pipelines the data transmission along the update tree and distributes the encoding computations among all the participating nodes. For adaptivity, we propose a rollback-based failure processing algorithm to achieve high adaptivity, which handles the node failure during update with the existing update tree in a rollback manner. To evaluate the performance of TA-Update, we conduct experiments on HDFS-RAID under various parameter settings on both 30 physical and 200 virtual machines. Extensive experiments confirm that TA-Update could support the various erasure codes with any parameter, i

关键词： distributed storage system data update update tree erasure codes

来源：评论

学校读者我要写书评

暂无评论

TMDA: Task-Specific Multi-Source Domain Adaptation via Clustering Embedded Adversarial Training 19

TMDA: Task-Specific Multi-Source Domain Adaptation via Clust...

引用

19th IEEE International Conference on Data Mining (ICDM)

作者： Wang, Haotian Yang, Wenjing Lin, Zhipeng Yu, Yue Natl Univ Def Technol State Key Lab High Performance Comp HPCL Changsha Peoples R China Natl Univ Def Technol Natl Lab Parallel & Distributed Proc Changsha Peoples R China Peng Cheng Lab Shenzhen Peoples R China

ISBN: (纸本)9781728146034

Beyond classical domain-specific adversarial training, a recently proposed task-specific framework has achieved a great success in single source domain adaptation by utilizing task-specific decision boundaries. However, compared to single-source-single-target setting, multi-source domain adaptation (MDA) shows more powerful capability to handle with most real-life cases. To align target domain with diverse multi-source domains using task-specific decision boundaries, we provide a deep insight of task-specific framework on MDA for the first time. Accordingly, we propose a novel task-specific multi-source domain adaptation method (TMDA) with a clustering embedded adversarial training process. Specifically, the proposed TMDA detects and refines less discriminative target representations through a max-min optimization over two adversarial task-specific classifiers. Moreover, our analysis implies that scattered multi-source representations disturb the adversarial training under the task-specific framework. To tight up the dispersed source representations, we embeds a relationship-based domain clustering into TMDA. Empirical results demonstrate that our TMDA outperforms state-of-the-art methods on toy dataset, sentiment analysis and digit classification.

关键词： Multi-source domain adaptation Adversarial Training Task-specific Domain Clustering

来源：评论

学校读者我要写书评

暂无评论

Ultrasound-Based Silent Speech Interface using Sequential Convolutional Auto-encoder 19

Ultrasound-Based Silent Speech Interface using Sequential Co...

引用

27th ACM International Conference on Multimedia (MM)

作者： Xu, Kele Wu, Yuxiang Gao, Zhifeng Natl Univ Def Technol Natl Key Lab Parallel & Distributed Proc Changsha Peoples R China Natl Univ Def Technol Sch Comp Changsha Peoples R China Peking Univ Beijing Peoples R China

ISBN: (纸本)9781450368896

"Silent Speech Interfaces" (SSI) refers to a system which uses non-audible signals recorded during speech production to perform speech recognition and synthesis tasks. Different approaches have been proposed for the SSI systems. In this paper, we focus on an ultrasound-based SSI. The performance of ultrasound-based SSI system heavily relies on the feature extraction approach. However, most of the previous attempts are often limited to individual frame analysis, and the context information of the image sequence cannot be taken into account. Inspired by the recent success of the recurrent neural network and convolutional auto-encoder, we explore a novel sequential feature extraction approach for SSI system. The architecture can extract spatial and temporal feature from the image sequence, which can be further deployed for the speech recognition and synthetic tasks. By quantitative comparison between different unsupervised feature extraction approaches, the new approach outperforms other methods on the 2010 SSI challenge.

关键词： sequential convolutional auto-encoder silent speech interface

来源：评论

学校读者我要写书评

暂无评论

A Mobile Application for Sound Event Detection 28

A Mobile Application for Sound Event Detection

引用

28th International Joint Conference on Artificial Intelligence

作者： Fu, Yingwei Xu, Kele Mi, Haibo Wang, Huaimin Wang, Dezhi Zhu, Boqing Natl Key Lab Parallel & Distributed Proc Changsha Peoples R China Natl Univ Def Technol Coll Comp Changsha Peoples R China Natl Univ Def Technol Coll Meteorol & Oceanog Changsha Peoples R China

ISBN: (纸本)9780999241141

Sound event detection is intended to analyze and recognize the sound events in audio streams and it has widespread applications in real life. Recently, deep neural networks such as convolutional recurrent neural networks have shown state-of-the-art performance in this task. However, the previous methods were designed and implemented on devices with rich computing resources, and there are few applications on mobile devices. This paper focuses on the solution on the mobile platform for sound event detection. The architecture of the solution includes offline training and online detection. During offline training process, multi model-based distillation method is used to compress model to enable real-time detection. The online detection process includes acquisition of sensor data, processing of audio signals, and detecting and recording of sound events. Finally, we implement an application on the mobile device that can detect sound events in near real time.

关键词：

来源：评论

学校读者我要写书评

暂无评论

CPMTD: Cyber-physical moving target defense for hardening the security of power system against false data injected attack

引用

COMPUTERS & SECURITY 2021年 111卷 102465-102465页

作者： Hu, Yifan Zhu, Peidong Xun, Peng Liu, Bo Kang, Wenjie Xiong, Yinqiao Shi, Weiheng Natl Univ Def Technol Coll Comp Changsha 410073 Peoples R China Natl Univ Def Technol Coll Comp Natl Key Lab Parallel & Distributed Proc Changsha 410073 Peoples R China Changsha Univ Dept Elect Informat & Elect Engn Changsha 410022 Peoples R China Hunan Police Acad Hunan Prov Key Lab Network Invest Technol Changsha 410138 Peoples R China Natl Univ Def Technol Coll Syst Engn Changsha 410073 Peoples R China Minist Publ Secur Key Lab Police Internet Things Applicat Beijing 100089 Peoples R China Natl Univ Def Technol Coll Meteorol & Oceanog Nanjing 211101 Peoples R China

Cyber-physical system (CPS) like smart grids deeply integrated with communication networks are often subjected to sophisticated cyber-attacks, such as false data injection attack (FDIA) with a strong capability of strategic reconnaissance required to learn the environment, where the static characteristics of the system enable an easier profiling of the critical infrastructure resources by the adversary. In this paper, we propose a cyber-physical moving target defense (CPMTD) technique that focuses on both attack prevention and detection to mitigate such static vulnerabilities and provide a combination of defense strategies for power system. For attack prevention, we design the Cyber-MTD strategy to mislead and disrupt attack preparation by randomizing the data acquisition with controlled change across multiple system dimensions based on the network programmability of protocol oblivious forwarding (POF). For attack detection, we design the Physical-MTD strategy to improve the detection probability of FDIA by periodically changing the measurement matrix of state estimation based on the D-FACTS devices' capability of perturbing the transmission line susceptance. Simulations on IEEE 14 bus and 57 bus systems demonstrate the effectiveness of CPMTD against FDIA with small overhead. The probability of cyber-attacks in two cases can be reduced by more than 90%;FDIA introduces little operation cost as most of them are detected. Network throughput barely changes and network latency increases by less than 9%. (c) 2021 Elsevier Ltd. All rights reserved.

关键词： Cyber-physical system Moving target defense Attack prevetion and detection Bi-level framework False data injection attack

来源：评论

学校读者我要写书评

暂无评论

Multi Model-Based Distillation for Sound Event Detection

引用

IEICE TRANSACTIONS ON INFORMATION AND SYSTEMS 2019年第10期E102D卷 2055-2058页

作者： Fu, Yingwei Xu, Kele Mi, Haibo Kong, Qiuqiang Wang, Dezhi Wang, Huaimin Hong, Tie Natl Key Lab Parallel & Distributed Proc Changsha 410073 Hunan Peoples R China Natl Univ Def Technol Coll Comp Changsha 410073 Hunan Peoples R China Univ Surrey Ctr Vis Speech & Signal Proc Guildford GU2 7XH Surrey England Natl Univ Def Technol Coll Meteorol & Oceanog Changsha 410073 Hunan Peoples R China

Sound event detection is intended to identify the sound events in audio recordings, which has widespread applications in real life. Recently, convolutional recurrent neural network (CRNN) models have achieved state-of-the-art performance in this task due to their capabilities in learning the representative features. However, the CRNN models are of high complexities with millions of parameters to be trained, which limits their usage for the mobile and embedded devices with limited computation resource. Model distillation is effective to distill the knowledge of a complex model to a smaller one, which can be deployed on the devices with limited computational power. In this letter, we propose a novel multi model-based distillation approach for sound event detection by making use of the knowledge from models of multiple teachers which are complementary in detecting sound events. Extensive experimental results demonstrated that our approach achieves a compression ratio about 50 times. In addition, better performance is obtained for the sound event detection task.

关键词： sound event detection model distillation model compression convolutional recurrent neural network

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：