检索结果-内蒙古大学图书馆

International Joint Conference on Neural Networks (IJCNN)

作者： Shang, Suning Wang, Huaimin Zhang, Pengfei Ding, Bo Natl Univ Def Technol Natl Key Lab Parallel & Distributed Proc Changsha 410073 Peoples R China

ISBN: (纸本)9781509060146

We present an unsupervised learning framework for the task of monocular depth and camera motion estimation from video sequences. In common with recent work, we use an unsupervised end-to-end learning method, requiring monocular video sequences for training. What makes the difference is, our approach not only uses image reconstruction as the supervisory signal but also exploits the pose estimation method which was used in traditional SLAM approach to enhance the supervisory signal and add training constraints. In pose estimation, a continuous frame window is set to construct the pose graph. Our method uses single-view depth and multi-view pose networks, with a loss based on reconstructing nearby images to the target using the predicted depth and pose. During training, the networks are thus coupled by the loss but can be applied independently at test time. Our evaluation of experiments on the KITTI dataset proves the effectiveness of our method: 1) monocular depth performs superior to the supervised methods that use ground-truth depth data for training and the existing unsupervised learning method. Our method performs comparably with the supervised methods that use ground-truth pose data for training. 2) pose estimation performs almost the same compared to established SLAM systems under comparable input settings.

关键词： Depth estimation Pose estimation Unsupervised learning Image reconstruction

来源：评论

学校读者我要写书评

暂无评论

Application of audiomagnetotellurics to delineate an iron deposit in the Liguo area, eastern China

引用

SOLID EARTH SCIENCES 2020年第2期5卷 83-92页

作者： He, Guo-li Kang, Hui-min Wang, Guang-jie Xiao, Tiao-jie Yang, Xiao-yong Minmet Explorat & Dev Co Ltd Beijing 100010 Peoples R China Chinese Acad Sci Inst Geol & Geophys Key Lab Shale Gas & Geoengn Beijing 100029 Peoples R China Natl Univ Def Technol Sci & Technol Parallel & Distributed Proc Lab Changsha 410073 Peoples R China Univ Sci & Technol China Sch Earth & Space Sci Hefei 230026 Peoples R China Univ Chinese Acad Sci Beijing 100049 Peoples R China Chinese Acad Sci Inst Earth Sci Beijing 100029 Peoples R China

A detailed study of the geological and electromagnetic structure of the peripheral area of Liguo iron mine, eastern China, is presented to assess the location and amount of iron ore. This study targeted a strong magnetic anomaly in the Lixi area, west of Liguo, which is located in the northern Yuhuai platform-fold-belt between the Sihu-Jing'an depression and the Xuzhou complex anticline. We carried out a series of audiomagnetotellurics (AMT) prospecting lines to delineate the deposit and assess its potential ore reserve. An integrated analysis of the magnetic anomaly, AMT resistivity, and drill core identified the areas with the great potential for iron ore mineralization. Comparation between magnetic anomaly and the resistivities can reduce the variability of singular geophysics method. The results provide important geophysical evidence regarding the potential iron-ore resources of the Liguo area. Copyright (c) 2020, Guangzhou Institute of Geochemistry. Production and hosting by Elsevier B.V.

关键词： Audiomagnetotellurics Liguo iron mine Mineralization mechanism Magnetic anomaly

来源：评论

学校读者我要写书评

暂无评论

Efficiently Handling Scale Variation for Pedestrian Detection 1

引用

9th International Conference on Intelligence Science and Big Data Engineering (IScIDE)

作者： Cheng, Qihua Zhang, Shanshan Nanjing Univ Sci & Technol PCA Lab Key Lab Intelligent Percept & Syst High Dimens In Minist EducSch Comp Sci & Engn Nanjing Peoples R China Nanjing Univ Sci & Technol Jiangsu Key Lab Image & Video Understanding Socia Sch Comp Sci & Engn Nanjing Peoples R China Sci & Technol Parallel & Distributed Proc Lab PDL Changsha Peoples R China

ISBN: (数字)9783030361891

ISBN: (纸本)9783030361891;9783030361884

Pedestrian detection is a popular yet challenging research topic in the computer vision community. Although it has achieved great progress in recent years, it still remains an open question how to handle scale variation, which commonly exists in real world applications. To address this problem, this paper presents a novel pedestrian detector to better classify and regress proposals of different scales given by a region proposal network (RPN). Specifically, we have made the following major modifications to the Adapted FasterRCNN baseline. First, we divide all proposals into small and large pools according to their scales, and deal with each pool in a separate classification network. Also, we employ two auxiliary supervisions to balance the effect of two parts of proposals on the back propagation. It is worth noting that the proposed new detector does not bring extra computational overhead and only introduces very few additional parameters. We have conducted experiments on the CityPersons, Caltech and ETH datasets and achieved significant improvements to the baseline method, especially on the small scale subset. In particular, on the CityPersons and ETH datasets, our method surpasses previous state-of-the-art methods with lower computational costs at test time.

关键词： Pedestrian detection Scale variation Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Towards a Uniform Template-based Architecture for Accelerating 2D and 3D CNNs on FPGA 18

Towards a Uniform Template-based Architecture for Accelerati...

引用

ACM/SIGDA International Symposium on Field-Programmable Gate Arrays (FPGA)

作者： Shen, Junzhong Huang, You Wang, Zelong Qiao, Yuran Wen, Mei Zhang, Chunyuan Natl Univ Def Technol Coll Comp Natl Key Lab Parallel & Distributed Proc Changsha Hunan Peoples R China

ISBN: (纸本)9781450356145

Three-dimensional convolutional neural networks (3D CNNs) are used efficiently in many computer vision applications. Most previous work in this area has concentrated only on designing and optimizing accelerators for 2D CNN, with few attempts made to accelerate 3D CNN on FPGA. We find accelerating 3D CNNs on FPGA to be challenge due to their high computational complexity and storage demands. More importantly, although the computation patterns of 2D and 3D CNNs are analogous, the conventional approaches adopted for accelerating 2D CNNs may be unfit for 3D CNN acceleration. In this paper, in order to accelerate 2D and 3D CNNs using a uniform framework, we propose a uniform template-based architecture that uses templates based on the Winograd algorithm to ensure fast development of 2D and 3D CNN accelerators. Furthermore, we also develop a uniform analytical model to facilitate efficient design space explorations of 2D and 3D CNN accelerators based on our architecture. Finally, we demonstrate the effectiveness of the template-based architecture by implementing accelerators for real-life 2D and 3D CNNs (VGG16 and C3D) on multiple FPGA platforms. On S2C VUS440, we achieve up to 1.13 TOPS and 1.11 TOPS under low resource utilization for VGG16 and C3D, respectively. End-to-end comparisons with CPU and GPU solutions demonstrate that our implementation of C3D achieves gains of up to 13x and 60x in performance and energy relative to a CPU solution, and a 6.4x energy efficiency gain over a GPU solution.

关键词： 3D CNN Winograd Algorithm Uniform Templates

来源：评论

学校读者我要写书评

暂无评论

DENOISING CONVOLUTIONAL AUTOENCODER BASED B-MODE ULTRASOUND TONGUE IMAGE FEATURE EXTRACTION 44

DENOISING CONVOLUTIONAL AUTOENCODER BASED B-MODE ULTRASOUND ...

引用

44th IEEE International Conference on Acoustics, Speech and Signal processing (ICASSP)

作者： Li, Bo Xu, Kele Feng, Dawei Mi, Haibo Wang, Huaimin Zhu, Jian Beijing Univ Posts & Telecommun Automat Dept Beijing 100876 Peoples R China Natl Key Lab Parallel & Distributed Proc Changsha Hunan Peoples R China Natl Univ Def Technol Changsha Hunan Peoples R China Univ Michigan Dept Linguist Ann Arbor MI 48109 USA

ISBN: (纸本)9781479981311

B-mode ultrasound tongue imaging is widely used in the speech production field. However, efficient interpretation is in a great need for the tongue image sequences. Inspired by the recent success of unsupervised deep learning approach, we explore unsupervised convolutional network architecture for the feature extraction in the ultrasound tongue image, which can be helpful for the clinical linguist and phonetics. By quantitative comparison between different unsupervised feature extraction approaches, the denoising convolutional autoencoder (DCAE)-based method outperforms the other feature extraction methods on the reconstruction task and the 2010 silent speech interface challenge. A Word Error Rate of 6.17% is obtained with DCAE, compared to the state-of-the-art value of 6.45% using Discrete cosine transform as the feature extractor. Our codes are available at https://***/DeePBluE666/Source-code1.

关键词： B-mode ultrasound tongue imaging unsupervised learning feature extraction convolutional autoencoder denoising convolutional autoencoder

来源：评论

学校读者我要写书评

暂无评论

Security/Timing-Aware Design Space Exploration of CAN FD for Automotive Cyber-Physical Systems

引用

IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS 2019年第2期15卷 1094-1104页

作者： Xie, Yong Zeng, Gang Kurachi, Ryo Takada, Hiroaki Xie, Guoqi Xiamen Univ Technol Dept Comp & Informat Engn Key Lab Internet Of Things Applicat Fujian Prov Xiamen 361024 Peoples R China Natl Univ Defens Technol Natl Lab Parallel & Distributed Proc Changsha 410073 Hunan Peoples R China Nagoya Univ Grad Sch Engn Nagoya Aichi 4648601 Japan Nagoya Univ Grad Sch Informat Sci Nagoya Aichi 4648601 Japan Hunan Univ Coll Comp Sci & Elect Engn Key Lab Embedded & Network Comp Hunan Prov Changsha 410073 Hunan Peoples R China

The controller area network with flexible data-rate (CAN FD) is the new generation of the CAN technology to meet the daily increasing bandwidth requirement for automotive cyber-physical systems (ACPS). However, ACPS is a security-critical system, an efficient security/timing-aware design space exploration (DSE) method is required to fully utilize CAN FD's high data phase data rate. In this paper, we propose an AUTOSAR-compliant system model that integrates both timing and security constraint, an integrated mixed-integer linear programming formulation (i-MILP) for the optimal DSE of CAN FD, and a divide-and-conquer approach to the i-MILP (dc-MILP) to address its timing complexity problem. The experiment results show that dc-MILP scales well for industrial-size systems and saves 1.94%-4.76% bandwidth utilization and guarantees the schedulability for more signal sets by comparing with the state-of-the-art algorithm.

关键词： Automotive cyber-physical systems (ACPS) controller area network with flexible data-rate (CAN FD) design space exploration (DSE) mixed integer linear programming security

来源：评论

学校读者我要写书评

暂无评论

Cascade Semantic Fusion for Image Captioning

引用

IEEE ACCESS 2019年 7卷 66680-66688页

作者： Wang, Shiwei Lan, Long Zhang, Xiang Dong, Guohua Luo, Zhigang Natl Univ Def Technol Sci & Technol Parallel & Distributed Proc Changsha 410073 Hunan Peoples R China Natl Univ Def Technol Coll Comp Changsha 410073 Hunan Peoples R China Natl Univ Def Technol Inst Quantum Informat Changsha 410073 Hunan Peoples R China Natl Univ Def Technol State Key Lab High Performance Comp Changsha 410073 Hunan Peoples R China

Recent advances in deep visual attention methods accelerate greatly the research of image captioning. However, how to leverage hand-crafted features or deep features for the encoder of image captioning is not fully explored, due to the difficulty in finding a kind of all-purpose features to entail a set of visual semantics. In this paper, we introduce a cascade semantic fusion architecture (CSF) to mine the representative features to encode image content through attention mechanism without bells and whistles. Specifically, the CSF benefits from three types of visual attention semantics, including object-level, image-level, and spatial attention features, in a novel three-stage cascade manner. In the first stage, object-level attention features are extracted to capture the detailed contents of the objects based on the pretrained detector. Then, the middle stage devises a fusion module to merge object-level attention features with spatial features, thereby inducing image-level attention features to enrich the context information around the objects. In the last stage, spatial attention features are learned to unveil the salient region representation as a complement to two previously learned attention features. In a nutshell, we integrate attention mechanism with three types of features to organize context knowledge about images from different aspects. The empirical analysis shows that the CSF can assist image captioning model in selecting the object regions of interest. The experiments of image captioning on MSCOCO dataset show the efficacy of our semantic fusion architecture in depicting image content.

关键词： Attention mechanism feature fusion image captioning

来源：评论

学校读者我要写书评

暂无评论

Stacked Marginal Time Warping for Temporal Alignment

引用

NEURAL procESSING LETTERS 2019年第2期49卷 711-735页

作者： Zhang, Xiang Nie, Liquan Lan, Long Huang, Xuhui Luo, Zhigang Natl Univ Def Technol State Key Lab High Performance Comp Changsha 410073 Hunan Peoples R China Natl Univ Def Technol Sci & Technol Parallel & distributed Proc Changsha 410073 Hunan Peoples R China Natl Univ Def Technol Coll Comp Changsha 410073 Hunan Peoples R China Natl Univ Def Technol Dept Comp Sci Changsha 410073 Hunan Peoples R China

Time warping is the popular technique of temporally aligning two sequences and has successfully applied in temporal alignment tasks such as activity recognition. However, existing time warping methods suffer from limited representation ability because aligning process is performed on either raw sequences or the projected lower-dimensional features. In this paper, we propose a stacked time warping framework (STW) to learn layer-wise representation for temporal alignment in a stacked structure. By using this structure, STW gives higher flexibility than existing methods meanwhile unifies them into a deep architecture. Based on the proposed STW framework, we explore a stacked marginal time warping (SMTW) method by using marginal stacked denoising autoencoder (mSDA) as the regularization term which enables SMTW to marginalize out noises and learn layer-wise non-linear representations with the effective closed-form solution. Benefitting from the incorporation of mSDA, SMTW achieves better alignment performance and keeps comparable time efficiency with regular time warping methods. Experiments on both synthetic data and practical human activity recognition datasets demonstrate that SMTW is superior to the state-of-the-art time warping methods in quantity.

关键词： Time warping Marginal stacked denoising autoencoder (mSDA) Representation learning

来源：评论

学校读者我要写书评

暂无评论

Person re-identification via adaptive verification loss

引用

NEUROCOMPUTING 2019年 359卷 93-101页

作者： Tian, Hui Zhang, Xiang Lan, Long Luo, Zhigang Natl Univ Def Technol Sci & Technol Parallel & Distributed Proc Changsha 410073 Hunan Peoples R China Natl Univ Def Technol Inst Quantum Informat Changsha 410073 Hunan Peoples R China Natl Univ Def Technol State Key Lab High Performance Comp Changsha 410073 Hunan Peoples R China Natl Univ Def Technol Coll Comp Changsha 410073 Hunan Peoples R China

Person re-identification (re-ID) is the task of associating the image patches of the identical person across disjoint cameras, which is a very challenging topic in computer vision. Recently, with the advance of deep learning techniques and the public of huge datasets, person re-ID has achieved a great success. However, an obvious but ignored issue is that the sample pairs are biased, where the negative sample pairs are far more than the positive ones. In this paper, we develop an adaptive verification loss, termed as ADV-Loss to handle the imbalance of sample pairs. Our ADV-Loss empowers the popular verification loss with the flexibility of adaptively addressed the hard triple units. Specifically, we learn a re-weighted strategy from the triplet loss to balance the sample pairs. According to the objective values of triple loss, the hard triple units will be endowed with larger weights, while the less important triple units are de-emphasized or simply dropped. Thus, ADV-Loss, on the basis of the cross-entropy loss, handles the biased issue by only holding the equal number but informative sample pairs. Moreover, theoretical analysis of our designed ADV-Loss is provided to guarantee the important sample pairs not to be despised during the backpropagation. Ablation analysis verifies the potential efficacy of the proposed loss. Moreover, experiments of person re-ID on three large-scale benchmark datasets show that two popular deep networks with the proposed loss achieve satisfactory performance against several well-established person re-ID methods. (C) 2019 Elsevier B.V. All rights reserved.

关键词： Person re-identification Triplet loss Hard pair mining

来源：评论

学校读者我要写书评

暂无评论

Learning to Cooperate in Decentralized Multi-robot Exploration of Dynamic Environments 25th

Learning to Cooperate in Decentralized Multi-robot Explorati...

引用

25th International Conference on Neural Information processing (ICONIP)

作者： Geng, Mingyang Zhou, Xing Ding, Bo Wang, Huaimin Zhang, Lei Natl Univ Def Technol Coll Comp Natl Key Lab Parallel & Distributed Proc Changsha Peoples R China Northeastern Univ Natl Key Lab Integrated Automat Proc Ind Shenyang Peoples R China

ISBN: (纸本)9783030042394;9783030042387

This paper presents an approach to train a decentralized multi-robot system to learn cooperation strategy in the exploration of dynamic environments. The traditional approaches to multi-robot exploration problem are all based on the "pre-designed" cooperation strategy. However, many real-world settings are too complex for humans to "design" effective strategies. Besides, "pre-designed" strategy does not possess the ability to adapt to different task environment features, which also limits its application in real-world practices. Inspired by the superiority of deep reinforcement learning technique on complex individual behavior design, we apply the same technology to the cooperative learning process on the robot collective level. Our approach has been evaluated in a simulated multi-robot Disaster Exploration scenario and the results show that it could be applied in more complicated scenarios in contrast with two traditional "human-designed" methods.

关键词： distributed multi-robot exploration Deep learning Reinforcement learning Dynamic environments Cooperative behavior

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：