检索结果-内蒙古大学图书馆

3D Hierarchical Refinement and Augmentation for Unsupervised Learning of Depth and Pose from Monocular Video

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Wang, Guangming Zhong, Jiquan Zhao, Shijie Wu, Wenhua Liu, Zhe Wang, Hesheng Department of Automation Institute of Medical Robotics Key Laboratory of System Control and Information Processing Ministry of Education Key Laboratory of Marine Intelligent Equipment and System of Ministry of Education Shanghai Engineering Research Center of Intelligent Control and Management Shanghai Jiao Tong University Shanghai200240 China Department of Engineering Mechanics Shanghai Jiao Tong University Shanghai200240 China Department of Computer Science and Technology University of Cambridge CambridgeCB2 1TN United Kingdom

Depth and ego-motion estimations are essential for the localization and navigation of autonomous robots and autonomous driving. Recent studies make it possible to learn the per-pixel depth and ego-motion from the unlabeled monocular video. A novel unsupervised training framework is proposed with 3D hierarchical refinement and augmentation using explicit 3D geometry. In this framework, the depth and pose estimations are hierarchically and mutually coupled to refine the estimated pose layer by layer. The intermediate view image is proposed and synthesized by warping the pixels in an image with the estimated depth and coarse pose. Then, the residual pose transformation can be estimated from the new view image and the image of the adjacent frame to refine the coarse pose. The iterative refinement is implemented in a differentiable manner in this paper, making the whole framework optimized uniformly. Meanwhile, a new image augmentation method is proposed for the pose estimation by synthesizing a new view image, which creatively augments the pose in 3D space but gets a new augmented 2D image. The experiments on KITTI demonstrate that our depth estimation achieves state-of-the-art performance and even surpasses recent approaches that utilize other auxiliary tasks. Our visual odometry outperforms all recent unsupervised monocular learning-based methods and achieves competitive performance to the geometry-based method, ORB-SLAM2 with back-end optimization. Copyright © 2021, The Authors. All rights reserved.

关键词： Robots

Shaping the Metaverse into Reality: A Holistic Multidisciplinary Understanding of Opportunities, Challenges, and Avenues for Future Investigation

学校读者我要写书评

暂无评论

Data Processor for Better Business Education 2023年第3期63卷 735-765页

作者： Alex Koohang Jeretta Horn Nord Keng-Boon Ooi Garry Wei-Han Tan Mostafa Al-Emran Eugene Cheng-Xi Aw Abdullah Mohammed Baabdullah Dimitrios Buhalis Tat-Huei Cham Charles Dennis Vincent Dutot Yogesh K. Dwivedi Laurie Hughes Emmanuel Mogaji Neeraj Pandey Ian Phau Ramakrishnan Raman Anshuman Sharma Marianna Sigala Akiko Ueno Lai-Wan Wong a School of Computing Middle Georgia State University Macon GA USA b Department of Management Science and Information Systems Oklahoma State University Stillwater OK USA c UCSI Graduate Business School UCSI University Kuala Lumpur Malaysiad School of Finance and Economics Nanchang Institute of Technology Nanchang People’s Republic of Chinau Faculty of Business Design and Arts Swinburne University of Technology Sarawak Campus Malaysia c UCSI Graduate Business School UCSI University Kuala Lumpur Malaysiad School of Finance and Economics Nanchang Institute of Technology Nanchang People’s Republic of Chinae School of Economics and Management Yunnan Normal University Kunming People’s Republic of Chinau Faculty of Business Design and Arts Swinburne University of Technology Sarawak Campus Malaysia f Faculty of Engineering & IT The British University in Dubai Dubai United Arab Emiratesg Department of Computer Techniques Engineering Dijlah University College Baghdad Iraq c UCSI Graduate Business School UCSI University Kuala Lumpur Malaysia h Department of Management Information Systems Faculty of Economics and Administration King Abdulaziz University Jeddah Saudi Arabia i Department of Tourism and Hospitality Bournemouth University Poole UK c UCSI Graduate Business School UCSI University Kuala Lumpur Malaysiaj Tashkent State University of Economics Tashkent Uzbekistan k Business School Middlesex University London UK l EM Normandie Business School Métis Lab Clichy France m Digital Futures for Sustainable Business & Society Research Group School of Management Swansea University Swansea UKn Department of Management Symbiosis Institute of Business Management Pune & Symbiosis International (Deemed University) Pune India m Digital Futures for Sustainable Business & Society Research Group School of Management Swansea University Swansea UK o School of Management & Marketing University of Greenwich London UK p National Institute of Industrial Engineering Mum

ABSTRACTABSTRACTThe term metaverse is described as the next iteration of the Internet. Metaverse is a virtual platform that uses extended reality technologies, i.e. augmented reality, virtual reality, mixed reality, 3D graphics, and other emerging technologies to allow real-time interactions and experiences in ways that are not possible in the physical world. Companies have begun to notice the impact of the metaverse and how it may help maximize profits. The purpose of this paper is to offer perspectives on several important areas, i.e. marketing, tourism, manufacturing, operations management, education, the retailing industry, banking services, healthcare, and human resource management that are likely to be impacted by the adoption and use of a metaverse. Each includes an overview, opportunities, challenges, and a potential research agenda.

关键词： Metaverse marketing tourism hospitality manufacturing operations management education the retailing industry banking services healthcare human resource management

NiuEM: A Nested-iterative Unsupervised Learning Model for Single-particle Cryo-EM Image Processing

学校读者我要写书评

暂无评论

NiuEM: A Nested-iterative Unsupervised Learning Model for Si...

2020 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2020

作者： Hu, Rui Cai, Jiaming Zheng, Wangjie Yang, Yang Shen, Hong-Bin Shanghai Jiao Tong University Key Lab. of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering Department of Computer Science and Engineering Shanghai200240 China Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University Key Laboratory of System Control and Information Processing Ministry of Education of China Shanghai200240 China Shanghai Jiao Tong University Department of Bioinformatics and Biostatistics Shanghai200240 China

ISBN: (纸本)9781728162157

Cryo-electron microscopy (cryo-EM) has become a mainstream technology for solving spatial structures of biomacromolecules, while the processing of cryo-EM images is a very challenging task. One of the great challenges is the high noise in the images. A common method is to cluster the images with close projecting angles to get mean images, which are used for 3D reconstruction. However, due to the extremely low signal-to-noise-ratio, common clustering methods often fail to obtain high-quality mean images, leading to poorly reconstructed structures. In this study, we present a new unsupervised learning framework, called NiuEM, to discriminate images captured from different angles and yield cluster-mean images. NiuEM first generates pseudo-labels and then exploits both contrastive loss and cross-entropy loss for training convolutional layers to learn feature representations. Moreover, the pseudo-labels are updated iteratively to enhance the reliability of labels. We assess the performance of NiuEM on four data sets via both visualized and quantitative experiments. Especially, two kinds of metrics are adopted to measure the performance, regarding the clustering quality and the resolution of reconstructed 3D models, respectively. The experimental results show that NiuEM achieves very competitive clustering accuracy in the comparison with the state-of-the-art image clustering methods. Moreover, the cluster mean images yielded by NiuEM lead to better initial 3D models compared with the mainstream reconstruction tools. © 2020 IEEE.

关键词： Image reconstruction

DPNet: Dynamic Pooling Network for Accurate and Efficient Size-Aware Tiny Object Detection

学校读者我要写书评

暂无评论

IEEE Internet of Things Journal 2025年

作者： Gong, Luqi Chen, Haotian Chen, Yikun Yao, Tianliang Li, Chao Zhao, Shuai Han, Guangjie Beijing University of Posts and Telecommunications School of Computer Science Beijing China Zhejiang Lab Research Center for Space Computing System Hangzhou China Southwest Jiaotong University SWJTU-LEEDS Joint School Chengdu China LTD Guangdong Zhiyun City construction Technology Co Zhuhai China Tongji University College of Electronic and Information Engineering Department of Control Science and Engineering Shanghai China Hohai University Key Laboratory of Maritime Intelligent Network Information Technology Ministry of Education China

In unmanned aerial systems, especially in complex environments, accurately detecting tiny objects is crucial. Resizing images is a common strategy to improve detection accuracy, particularly for small objects. However, simply enlarging images significantly increases computational costs and the number of negative samples, severely degrading detection performance and limiting its applicability. This paper proposes a Dynamic Pooling Network (DPNet) for tiny object detection to mitigate these issues. DPNet employs a flexible down-sampling strategy by introducing a factor (df) to relax the fixed down-sampling process of the feature map to an adjustable one. Furthermore, we design a lightweight predictor to predict df for each input image, which will be used to decrease the resolution of feature map in backbone. Thus, we achieve input-aware down-sampling. We design an Adaptive Normalization Module (ANM) to make a unified detector well compatible with different dfs. At the same time, we also design a guidance loss to supervise the predictor’s training. DPNet realizes the dynamic allocation of computing resources to trade off detection accuracy and efficiency through this. Experiments on the TinyCOCO and TinyPerson datasets show that our DPNet can save over 35% and 25% GFLOPs, respectively, while maintaining comparable detection *** code will be made publicly available. © 2014 IEEE.

关键词： Object detection

Resonant beam communications with echo interference elimination

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Xiong, Mingliang Liu, Qingwen Wang, Gang Giannakis, Georgios B. Zhang, Sihai Zhu, Jinkang Huang, Chuan College of Electronics and Information Engineering Tongji University Shanghai201804 China State Key Lab of Intelligent Control and Decision of Complex Systems School of Automation Beijing Institute of Technology Beijing100081 China Department of Electrical and Computer Engineering University of Minnesota MinneapolisMN55455 United States Key Lab of Wireless-Optical Communications University of Science and Technology of China Hefei Anhui230026 China School of Science and Engineering Chinese University of Hong Kong Shenzhen Guangdong518172 Hong Kong

Resonant beam communications (RBCom) is capable of providing wide bandwidth when using light as the carrier. Besides, the RBCom system possesses the characteristics of mobility, high signal-to-noise ratio (SNR), and multiplexing. Nevertheless, the channel of the RBCom system is distinct from other light communication technologies due to the echo interference issue. In this paper, we reveal the mechanism of the echo interference and propose the method to eliminate the interference. Moreover, we present an exemplary design based on frequency shifting and optical filtering, along with its mathematic model and performance analysis. The numerical evaluation shows that the channel capacity is greater than 15 bit/s/Hz. Copyright © 2021, The Authors. All rights reserved.

关键词： Internet of things

DPNet: Dynamic Pooling Network for Tiny Object Detection

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Gong, Luqi Chen, Haotian Chen, Yikun Yao, Tianliang Li, Chao Zhao, Shuai Han, Guangjie School of Computer Science Beijing University of Posts and Telecommunications Beijing China Research Center for Space Computing System Zhejiang Lab Hangzhou China SWJTU-LEEDS Joint School Southwest Jiaotong University Chengdu China Guangdong Zhiyun City construction Technology Co. LTD Zhuhai China Department of Control Science and Engineering College of Electronic and Information Engineering Tongji University Shanghai China Key Laboratory of Maritime Intelligent Network Information Technology Ministry of Education Hohai University China

In unmanned aerial systems, especially in complex environments, accurately detecting tiny objects is crucial. Resizing images is a common strategy to improve detection accuracy, particularly for small objects. However, simply enlarging images significantly increases computational costs and the number of negative samples, severely degrading detection performance and limiting its applicability. This paper proposes a Dynamic Pooling Network (DPNet) for tiny object detection to mitigate these issues. DPNet employs a flexible down-sampling strategy by introducing a factor (df) to relax the fixed downsampling process of the feature map to an adjustable one. Furthermore, we design a lightweight predictor to predict df for each input image, which will be used to decrease the resolution of feature map in backbone. Thus, we achieve input-aware downsampling. We design an Adaptive Normalization Module (ANM) to make a unified detector well compatible with different dfs. At the same time, we also design a guidance loss to supervise the predictor’s training. DPNet realizes the dynamic allocation of computing resources to trade off detection accuracy and efficiency through this. Experiments on the TinyCOCO and TinyPerson datasets show that our DPNet can save over 35% and 25% GFLOPs, respectively, while maintaining comparable detection *** code will be made publicly available. Copyright © 2025, The Authors. All rights reserved.

关键词： Image coding

Opportunities and challenges for monitoring terrestrial biodiversity in the robotics age

学校读者我要写书评

暂无评论

Nature Ecology and Evolution 2025年第6期9卷 1031-1042页

作者： Pringle, Stephen Dallimer, Martin Goddard, Mark A. Le Goff, Léni K. Hart, Emma Langdale, Simon J. Fisher, Jessica C. Abad, Sara-Adela Ancrenaz, Marc Angeoletto, Fabio Auat Cheein, Fernando Austen, Gail E. Bailey, Joseph J. Baldock, Katherine C. R. Banin, Lindsay F. Banks-Leite, Cristina Barau, Aliyu S. Bashyal, Reshu Bates, Adam J. Bicknell, Jake E. Bielby, Jon Bosilj, Petra Bush, Emma R. Butler, Simon J. Carpenter, Dan Clements, Christopher F. Cully, Antoine Davies, Kendi F. Deere, Nicolas J. Dodd, Michael Drinkwater, Rosie Driscoll, Don A. Dutilleux, Guillaume Dyrmann, Mads Edwards, David P. Farhadinia, Mohammad S. Faruk, Aisyah Field, Richard Fletcher, Robert J. Foster, Chris W. Fox, Richard Francksen, Richard M. Franco, Aldina M. A. Gainsbury, Alison M. Gardner, Charlie J. Giorgi, Ioanna Griffiths, Richard A. Hamaza, Salua Hanheide, Marc Hayward, Matt W. Hedblom, Marcus Helgason, Thorunn Heon, Sui P. Hughes, Kevin A. Hunt, Edmund R. Ingram, Daniel J. Jackson-Mills, George Jowett, Kelly Keitt, Timothy H. Kloepper, Laura N. Kramer-Schadt, Stephanie labisko, Jim labrosse, Frédéric Lawson, Jenna Lecomte, Nicolas de Lima, Ricardo F. Littlewood, Nick A. Marshall, Harry H. Masala, Giovanni L. Maskell, Lindsay C. Matechou, Eleni Mazzolai, Barbara McConnell, Alistair Melbourne, Brett A. Miriyev, Aslan Nana, Eric Djomo Ossola, Alessandro Papworth, Sarah Parr, Catherine L. Payo-Payo, Ana Perry, Gad Pettorelli, Nathalie Pillay, Rajeev Potts, Simon G. Prendergast-Miller, Miranda T. Qie, Lan Rolley-Parnell, Persie Rossiter, Stephen J. Rowcliffe, Marcus Rumble, Heather Sadler, Jon P. Sandom, Christopher J. Sanyal, Asiem Schrodt, Franziska Sethi, Sarab S. Shabrani, Adi Siddall, Robert Smith, Simón C. Snep, Robbert P. H. Soulsbury, Carl D. Stanley, Margaret C. Stephens, Philip A. Stephenson, P.J. Struebig, Matthew J. Studley, Matthew Svátek, Martin Tang, Gilbert Taylor, Nicholas K. Umbers, Kate D. L. Ward, Robert J. White, Patrick J. C. Whittingham, Mark J. Wich, Serge Williams, Christopher D. Yakubu, Ibrahim B. Yoh, Natalie Durrell Institute of Conservation and Ecology (DICE) School of Natural Sciences University of Kent Canterbury United Kingdom Centre for Environmental Policy Imperial College London London United Kingdom Department of Geography and Environmental Sciences Northumbria University Newcastle upon Tyne United Kingdom School of Computing Engineering and the Built Environment Edinburgh Napier University Edinburgh United Kingdom Synthotech Ltd Milner Court Hornbeam Square Harrogate United Kingdom Mechanical Engineering Department University College London London United Kingdom HUTAN SWD Kota Kinabalu Malaysia Programa de Pós-Graduação em Gestão e Tecnologia Ambiental da Universidade Federal de Rondonópolis Rondonópolis Brazil Department of Engineering Harper Adams University Newport United Kingdom Applied Ecology Research Group School of Life Sciences Anglia Ruskin University Cambridge United Kingdom Centre for Ecology and Hydrology Penicuik United Kingdom Department of Life Sciences Imperial College London Silwood Park Campus Ascot United Kingdom Department of Urban and Regional Planning Faculty of Earth and Environmental Sciences Bayero University Kano Nigeria Greenhood Nepal Kathmandu Nepal Animal Rural & Environmental Sciences Nottingham Trent University Nottinghamshire United Kingdom School of Biological and Environmental Sciences Liverpool John Moores University Liverpool United Kingdom Lincoln Centre for Autonomous Systems University of Lincoln Lincoln United Kingdom Royal Botanic Garden Edinburgh Edinburgh United Kingdom School of Biological Sciences University of East Anglia Norwich Research Park Norwich United Kingdom Digital Ecology Limited Bristol United Kingdom School of Biological Sciences University of Bristol Bristol United Kingdom Department of Computing Imperial College London London United Kingdom Department of Ecology and Evolutionary Biology University of Colorado Boulder CO United States Faculty of Science Technology Engineering

With biodiversity loss escalating globally, a step change is needed in our capacity to accurately monitor species populations across ecosystems. Robotic and autonomous systems (RAS) offer technological solutions that may substantially advance terrestrial biodiversity monitoring, but this potential is yet to be considered systematically. We used a modified Delphi technique to synthesize knowledge from 98 biodiversity experts and 31 RAS experts, who identified the major methodological barriers that currently hinder monitoring, and explored the opportunities and challenges that RAS offer in overcoming these barriers. Biodiversity experts identified four barrier categories: site access, species and individual identification, data handling and storage, and power and network availability. Robotics experts highlighted technologies that could overcome these barriers and identified the developments needed to facilitate RAS-based autonomous biodiversity monitoring. Some existing RAS could be optimized relatively easily to survey species but would require development to be suitable for monitoring of more ‘difficult’ taxa and robust enough to work under uncontrolled conditions within ecosystems. Other nascent technologies (for instance, new sensors and biodegradable robots) need accelerated research. Overall, it was felt that RAS could lead to major progress in monitoring of terrestrial biodiversity by supplementing rather than supplanting existing methods. Transdisciplinarity needs to be fostered between biodiversity and RAS experts so that future ideas and technologies can be codeveloped effectively. © The Author(s) 2025.

关键词：

Model based reinforcement learning for robot grasping trajectory generation 4

学校读者我要写书评

暂无评论

Model based reinforcement learning for robot grasping trajec...

4th IEEE International Conference on Advanced Robotics and Mechatronics, ICARM 2019

作者： Xue, Hongxiang Wen, Shuhuan Yang, Chao Liu, Huaping Key Lab of Industrial Computer Control Engineering of Hebei Province Yanshan University Qinhuangdao066004 China Department of Computer Science and Technology State Key Laboratory of Intelligent Technology and System Tsinghua National Laboratory for Information Science and Technology Tsinghua University Beijing100084 China

ISBN: (纸本)9781728100647

Trajectory generation is a fundamental problem for successful robotic grasping. However, most of the existing work dealt with this problem using supervised learning with a prescribed model. It prevents the developed grasp strategies from being used for new unknown scenarios. In this paper, inspired by the success of the reinforcement learning method, we developed a new method which optimizes grasp trajectories using a policy search method. An important advantage of the proposed method is that the complicated 3D model could be used to generate the multi-dimensional grasp configurations. The developed algorithm has been validated on some typical robotic hand. The experimental results show that the proposed method achieves excellent performance. © 2019 IEEE.

关键词： Trajectories

AngularGrad: A New Optimization Technique for Angular Convergence of Convolutional Neural Networks

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Roy, Swalpa Kumar Paoletti, Mercedes E. Haut, Juan M. Dubey, Shiv Ram Kar, Purbayan Plaza, Antonio Chaudhuri, Bidyut B. The Computer Science and Engineering Alipurduar Government Engineering and Management College 736206 India The Hyperspectral Computing Laboratory Department of Technology of Computers and Communications University of Extremadura Cáceres10003 Spain The Computer Vision and Biometrics Lab Indian Institute of Information Technology Prayagraj Uttar Pradesh Allahabad211015 India The Media Analysis Group Sony Research India Private Limited Karnataka Bangalore560103 India The Computer Vision and Pattern Recognition Unit Indian Statistical Institute Kolkata700108 India

Convolutional neural networks (CNNs) are trained using stochastic gradient descent (SGD)-based optimizers. Recently, the adaptive moment estimation (Adam) optimizer has become very popular due to its adaptive momentum, which tackles the dying gradient problem of SGD. Nevertheless, existing optimizers are still unable to exploit the optimization curvature information efficiently. This paper proposes a new AngularGrad optimizer that considers the behavior of the direction/angle of consecutive gradients. This is the first attempt in the literature to exploit the gradient angular information apart from its magnitude. The proposed AngularGrad generates a score to control the step size based on the gradient angular information of previous iterations. Thus, the optimization steps become smoother as a more accurate step size of immediate past gradients is captured through the angular information. Two variants of AngularGrad are developed based on the use of Tangent or Cosine functions for computing the gradient angular information. Theoretically, AngularGrad exhibits the same regret bound as Adam for convergence purposes. Nevertheless, extensive experiments conducted on benchmark data sets against state-of-the-art methods reveal a superior performance of AngularGrad. Source code: https://***/mhaut/AngularGrad. Copyright © 2021, The Authors. All rights reserved.

关键词： Optimization