检索结果-内蒙古大学图书馆

Analysis of Anchor-Based and Anchor-Free Object Detection Methods Based on Deep Learning

学校读者我要写书评

暂无评论

Analysis of Anchor-Based and Anchor-Free Object Detection Me...

IEEE International Conference on Mechatronics and Automation

作者： Shujian Liu Haibo Zhou Chenming Li Shuo Wang Tianjin Key Laboratory for Advanced Mechatronic System Design and Intelligent Control School of Mechanical Engineering Tianjin University of Technology Tianjin China National Demonstration Center for Experimental Mechanical and Electrical Engineering Education Tianjin University of Technology Tianjin China

ISBN: (数字)9781728164168

ISBN: (纸本)9781728164175

As one of the core mission of computer vision, object detection has been widely applied with the rapid development of computer technology, especially in the fields of face detection, behavior detection, auto driving, and intelligent monitoring. Aiming at the shortcomings of the traditional object detection, such as low detection accuracy, low efficiency and poor robustness, this article summarizes deep learning-based detectors of two kinds of modules: Anchor-Based and Anchor-Free. The performance of each detector is compared and analyzed in this paper as well. In addition, we summarize the development of object detection's key technologies from the aspect of improvement of Backbone, the optimization of NMS, imbalance of positive and negative samples' solution etc. Finally, the development trend of object detection is discussed from the prospect of lightweight detection model, weakly supervision detection and small object detection etc.

关键词：

Influence of Micro-texture Parameters on Bone Drilling Force and Torque: A Finite Element Model 4

学校读者我要写书评

暂无评论

Influence of Micro-texture Parameters on Bone Drilling Force...

2018 4th International Conference on Mechanical Engineering and Automation Science, ICMEAS 2018

作者： Hu, Yahui Li, Xiangjun Chen, Chuncheng Su, Lili Zheng, Qingchun Tianjin Key Laboratory for Advanced Mechatronic System Design and Intelligent Control School of Mechanical Engineering Tianjin University of Technology Tianjin300384 China Tianjin University of Technology 391 Binshui West Road Tianjin Xiqing District300384 China

The micro-texture on the tool surface is an effective way to improve the quality of the drilling. While high-performance micro-texture plays an important role in improving the quality of working surface and prolonging the life of tool. In this study, a three-dimensional finite element model was established to study the relationship between drilling thrust, torque and microstructure parameters (micro-pit depth, micro-pit diameter and micro-pit spacing) during drilling of cortical bone. The model was validated by cortical bone drilling using a micro-structured tool. On basis of this, the thrust and torque prediction models of the micro-texture parameters were obtained. The results indicate that the thrust and torque increased with the increase of micro-pit diameter;while the increase of the micro-pit spacing caused the thrust and torque to decrease. When the depth of the micro-pit increased, the thrust and torque decreased first and then increased, and we found that the effect of micro-pit spacing and micro-pit depth on force and torque is insignificant. © Published under licence by IOP Publishing Ltd.

关键词： Torque

Visual Servoing of Magnetic Swimming Robot Based on Mean Shift and Fast Template Matching Algorithm

学校读者我要写书评

暂无评论

Visual Servoing of Magnetic Swimming Robot Based on Mean Shi...

IEEE International Conference on Robotics and Biomimetics

作者： Hongbiao Xiang Tilei Zhang Mengwei Li Shoujun Wang Xiuping Yang Tianjin Key Laboratory for Advanced Mechatronic System Design and Intelligent Control Tianjin University of Technology Tianjin China

The actuation and control of mini-scale swimming robots have exhibit promising potential in limited and constrained environments, such as biomedical, micro-assembly and microfluidic applications. However, closed-loop control is still a challenging for the precision and repeatability in those applications. In this paper, a new type of soft-micro swimming robot based on magnetoelastic composite material is designed, and the swimmer can move freely on the water surface controlled by a rotating magnetic field with visual servoing. The visual feedback control strategy is proposed based on Mean Shift Algorithm and Fast Template Matching. The algorithm includes three parts: prediction, template matching, target positioning and template updating. To understand the characteristics of the swimmer, the relationship between swimming speed and magnetic field strength and frequency is verified by the algorithm in experiments. Furthermore, the proposed path following method with visual servoing applied in this control system. The experimental results show that the algorithm runs fast, has high accuracy and good real-time performance, and can provide great potential in mini-scale robot applications.

关键词：

Methodology for optimization of preload in a bolted-flange connection based on Markov theory

学校读者我要写书评

暂无评论

Journal of Physics: Conference Series 2022年第1期2174卷 012072页

作者： Yuezhen Wang Zhifeng Liu Ying Li Congbin Yang Institute of Advanced Manufacturing and Intelligent Technology Beijing University of Technology Beijing Beijing 100124 China Mechanical Industry Key Laboratory of Heavy Machine Tool Digital Design and Testing Beijing University of Technology Beijing Beijing 100124 China Key Laboratory of CNC Equipment Reliability Ministry of Education Jilin University Changchun Jilin 130012 China

The bolted-flange connection is one of the most popular detachable mechanical joints in mechanical flange connection. It is well known that bolted-flange joints need to be tightened to meet the assembly performance in practical engineering applications. However, the dispersion of the pre-tightening force caused by the elastic interaction between the bolts has a serious impact on the sealing and safety of the product, and therefore the optimization of bolt preload has always been a hot spot in the assembly research field. Combining mathematical theory and experimental data, this paper proposes an optimization model for uniformly distributed preload based on the Markov theory. The torque method was utilized to fasten the bolt-flange joint. The bolt pre-tightening force in the final state was collected and used as sample data. A state transition model was constructed to determine the state transition matrix. Ultimately, an iterative model of the preload was deduced, and the rationality of the model was verified through experimentation. The proposed optimization method of bolt preload can be applied to determine the initial preload of each bolt. This method improves the assembly efficiency and accuracy, which provides a theoretical approach to effectively guide the tightening work of bolted-flange connections.

关键词：

Constraining ultralight vector dark matter with the Parkes Pulsar Timing Array second data release

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Wu, Yu-Mei Chen, Zu-Cheng Huang, Qing-Guo Zhu, Xingjiang Bhat, N.D. Ramesh Feng, Yi Hobbs, George Manchester, Richard N. Russell, Christopher J. Shannon, R.M. School of Fundamental Physics and Mathematical Sciences Hangzhou Institute for Advanced Study UCAS Hangzhou310024 China School of Physical Sciences University of Chinese Academy of Sciences No. 19A Yuquan Road Beijing100049 China CAS Key Laboratory of Theoretical Physics Institute of Theoretical Physics Chinese Academy of Sciences Beijing100190 China Department of Astronomy Beijing Normal University Beijing100875 China Advanced Institute of Natural Sciences Beijing Normal University Zhuhai519087 China International Centre for Radio Astronomy Research Curtin University BentleyWA6102 Australia Research Center for Intelligent Computing Platforms Zhejiang Laboratory Hangzhou311100 China Australia Telescope National Facility CSIRO Astronomy and Space Science P.O. Box 76 EppingNSW1710 Australia CSIRO Scientific Computing Australian Technology Park Locked Bag 9013 AlexandriaNSW1435 Australia Centre for Astrophysics and Supercomputing Swinburne University of Technology HawthornVIC3122 Australia Australia

Composed of ultralight bosons, fuzzy dark matter provides an intriguing solution to challenges that the standard cold dark matter model encounters on sub-galactic scales. The ultralight dark matter with mass m ∼ 10−23eV will induce a periodic oscillation in gravitational potentials with a frequency in the nanohertz band, leading to observable effects in the arrival times of radio pulses from pulsars. Unlike scalar dark matter, pulsar timing signals induced by the vector dark matter are dependent on the oscillation direction of the vector fields. In this work, we search for ultralight vector dark matter in the mass range of [2 × 10−24, 2 × 10−22 ]eV through its gravitational effect in the Parkes Pulsar Timing Array (PPTA) second data release. Since no statistically significant detection is made, we place 95% upper limits on the local dark matter density as ρVF . 5 GeV/cm3 for m . 10−23 eV. As no preferred direction is found for the vector dark matter, these constraints are comparable to those given by the scalar dark matter search with an earlier 12-year data set of PPTA. Copyright © 2022, The Authors. All rights reserved.

关键词： Vectors

Knowledge-routed visual question reasoning: Challenges for deep representation embedding

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Cao, Qingxing Li, Bailin Liang, Xiaodan Wang, Keze Lin, Liang School of Intelligent Systems Engineering Sun Yat-sen University China DarkMatter AI Research China Department of statistics University of California Los AngelesCA90095 United States School of Computer Science and Engineering Sun Yat-sen University Key Laboratory of Machine Intelligence and Advanced Computing Ministry of Education Engineering Research Center for Advanced Computing Engineering Software of Ministry of Education China

Though beneficial for encouraging the Visual Question Answering (VQA) models to discover the underlying knowledge by exploiting the input-output correlation beyond image and text contexts, the existing knowledge VQA datasets are mostly annotated in a crowdsource way, e.g., collecting questions and external reasons from different users via the internet. In addition to the challenge of knowledge reasoning, how to deal with the annotator bias also remains unsolved, which often leads to superficial over-fitted correlations between questions and answers. To address this issue, we propose a novel dataset named Knowledge-Routed Visual Question Reasoning for VQA model evaluation. Considering that a desirable VQA model should correctly perceive the image context, understand the question, and incorporate its learned knowledge, our proposed dataset aims to cutoff the shortcut learning exploited by the current deep embedding models and push the research boundary of the knowledge-based visual question reasoning. Specifically, we generate the question-answer pair based on both the Visual Genome scene graph and an external knowledge base with controlled programs to disentangle the knowledge from other biases. The programs can select one or two triplets from the scene graph or knowledge base to push multi-step reasoning, avoid answer ambiguity, and balanced the answer distribution. In contrast to the existing VQA datasets, we further imply the following two major constraints on the programs to incorporate knowledge reasoning: i) multiple knowledge triplets can be related to the question, but only one knowledge relates to the image object. This can enforce the VQA model to correctly perceive the image instead of guessing the knowledge based on the given question solely;ii) all questions are based on different knowledge, but the candidate answers are the same for both the training and test sets. We make the testing knowledge unused during training to evaluate whether a model can unde

关键词： Knowledge based systems

Towards Graph Self-Supervised Learning with Contrastive Adjusted Zooming

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Zheng, Yizhen Jin, Ming Pan, Shirui Li, Yuan-Fang Peng, Hao Li, Ming Li, Zhao Department of Data Science and AI Faculty of IT Monash University ClaytonVIC3800 Australia School of Information and Communication Technology Griffith University Australia SouthportQLD4222 Australia Beijing Advanced Innovation Center for Big Data and Brain Computing Beihang University Beijing100191 China Key Laboratory of Intelligent Education Technology Application of Zhejiang Province Zhejiang Normal University Jinhua321004 China Alibaba Group Hangzhou311100 China

Graph representation learning (GRL) is critical for graph-structured data analysis. However, most of the existing graph neural networks (GNNs) heavily rely on labeling information, which is normally expensive to obtain in the real world. Although some existing works aim to effectively learn graph representations in an unsupervised manner, they suffer from certain limitations, such as the heavy reliance on monotone contrastiveness and limited scalability. To overcome the aforementioned problems, in light of the recent advancements in graph contrastive learning, we introduce a novel self-supervised graph representation learning algorithm via Graph Contrastive Adjusted Zooming, namely G-Zoom, to learn node representations by leveraging the proposed adjusted zooming scheme. Specifically, this mechanism enables G-Zoom to explore and extract self-supervision signals from a graph from multiple scales: micro (i.e., node-level), meso (i.e., neighborhood-level), and macro (i.e., subgraph-level). Firstly, we generate two augmented views of the input graph via two different graph augmentations. Then, we establish three different contrastiveness on the above three scales progressively, from node, neighboring, to subgraph level, where we maximize the agreement between graph representations across scales. While we can extract valuable clues from a given graph on the micro and macro perspectives, the neighboring-level contrastiveness offers G-Zoom the capability of a customizable option based on our adjusted zooming scheme to manually choose an optimal viewpoint that lies between the micro and macro perspectives to better understand the graph data. Additionally, to make our model scalable to large graphs, we employ a parallel graph diffusion approach to decouple model training from the graph size. We have conducted extensive experiments on real-world datasets, and the results demonstrate that our proposed model outperforms state-of-the-art methods consistently. © 2021, CC BY.

关键词： Graph neural networks

Convolutional neural network-assisted recognition of nanoscale L12 ordered structures in face-centred cubic alloys

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Li, Yue Zhou, Xuyang Colnaghi, Timoteo Wei, Ye Marek, Andreas Li, Hongxiang Bauer, Stefan Rampp, Markus Stephenson, Leigh T. Max-Planck Institut für Eisenforschung GmbH Max-Planck-Straße 1 Düsseldorf40237 Germany Max Planck Computing and Data Facility Gießenbachstraße 2 Garching85748 Germany State Key Laboratory for Advanced Metals and Materials University of Science and Technology Beijing Beijing100083 China Max Planck Institute for Intelligent Systems Max-Planck-Ring 4 Tübingen72076 Germany

Nanoscale L12-type ordered structures are widely used in face-centred cubic (FCC) alloys to exploit their hardening capacity and thereby improve mechanical properties. These fine-scale particles are typically fully coherent with matrix with the same atomic configuration disregarding chemical species, which makes them challenging to be characterized. Spatial distribution maps (SDMs) are used to probe local order by interrogating the three-dimensional (3D) distribution of atoms within reconstructed atom probe tomography (APT) data. However, it is almost impossible to manually analyse the complete point cloud (>10 million) in search for the partial crystallographic information retained within the data. Here, we proposed an intelligent L12-ordered structure recognition method based on convolutional neural networks (CNNs). The SDMs of a simulated L12-ordered structure and the FCC matrix were firstly generated. These simulated images combined with a small amount of experimental data were used to train a CNN-based L12-ordered structure recognition model. Finally, the approach was successfully applied to reveal the 3D distribution of L12–type d¢–Al3(LiMg) nanoparticles with an average radius of 2.54 nm in a FCC Al-Li-Mg system. The minimum radius of detectable nanodomain is even down to 5 Å. The proposed CNN-APT method is promising to be extended to recognize other nanoscale ordered structures and even more-challenging short-range ordered phenomena in the near future. Copyright © 2021, The Authors. All rights reserved.

关键词： Machine learning

Swimming characteristics of soft robot with magnetoelastic material

学校读者我要写书评

暂无评论

Swimming characteristics of soft robot with magnetoelastic m...

2019 IEEE International Conference on Robotics and Biomimetics, ROBIO 2019

作者： Xiang, Hongbiao Li, Mengwei Zhang, Tilei Wang, Shoujun Yang, Xiuping Tianjin University of Technology Tianjin Key Laboratory for Advanced Mechatronic System Design and Intelligent Control School of Mechanical Engineering Tianjin300384 China National Demonstration Center for Experimental Mechanical and Electrical Engineering Education Tianjin University of Technology Tianjin300384 China

ISBN: (纸本)9781728163215

Cable-less micro-robots have exhibit great capabilities for remote applications in small and constrained environments, especially in biotechnology and healthcare. This paper presents a swimmer propelled by its undulatory deformation using magnetoelastic composite material. The robot can move freely on water surface driven through a function of magnetic field, such as forward, backward and steering with different velocity. The magnetic field is generated by 3D Helmholtz coils, and the locomotion of the micro swimming robot is controlled by different applied magnetic field. To understand the characteristics of the swimmer, the force balance equation is established, and the morphologies of swimmer with different magnetic field are analyzed and described through finite element analysis method. Furthermore, we discuss and analyze relationship between the parameters of magnetic field and the speed of the swimmer. The experimental results verified the speed model of the swimming robot, which is influenced by the frequency and strength of magnetic field. © 2019 IEEE.

关键词： Magnetic fields