检索结果-内蒙古大学图书馆

Relation-wise transformer network and reinforcement learning for visual navigation

Neural Computing and Applications 2024年第21期36卷 13205-13221页

作者： He, Yu Zhou, Kang School of Computer Science and Artificial Intelligence Huanghuai University Zhumadian463000 China Henan Key Laboratory of Smart Lighting Zhumadian463000 China Henan International Joint Laboratory of Behavior Optimization Control for Smart Robots Henan463000 China School of Computer Science City University of Hong Kong 999077 Hong Kong School of Computer Science Wuhan University Wuhan430072 China

The task of object goal navigation is to drive an embodied agent to find the location of a given target only using visual observation. The mapping from visual perception of observation determines the navigation actions. Heterogeneous relationships in the observation are the essential part of the scene graph, which can guide the agent to find the target more easily. In this work, we propose a novel Heterogeneous Zone Graph Visual Transformer formulation for graph representation and visual perception. It consists of two key ideas: (1) Heterogeneous Zone Graph (HZG) that explores the heterogeneous target-related zones graph and spatial information. It allows the agent to navigate efficiently. (2) Relation-wise Transformer Network (RTNet) that transforms the relationship between previously observed objects and navigation actions. RTNet extracts rich nodes and edges features as pays more attention to the target-related zone. We model self-attention on the node-to-node encoder and cross-attention on the edge-to-node decoder. We evaluate our methods on the AI2THOR dataset and show superior navigation performance. Code and datasets can be found inhttps://***/zhoukang12321/RTNet_VN_2023. © The Author(s) 2024.

关键词： Navigation

来源：评论

学校读者我要写书评

暂无评论

Novel Visual Odometry Method for Water Autonomous Navigation 2

Novel Visual Odometry Method for Water Autonomous Navigation

引用

2nd IEEE international Conference on Computation, Communication and Engineering, ICCCE 2022

作者： Zou, Xiong Gao, Jinfeng Li, Huyong Zhang, Yu Liu, Fang Wu, Haitao Yao, Ruxian College of Computer and Artificial Intelligence Huanghuai University Zhumadian China Henan International Joint Laboratory of Behavior Optimization Control for Smart Robots Henan Provincial Key Laboratory of Smart Lighting Zhumadian China

ISBN: (纸本)9781665482486

A special VO algorithm designed for the inland river environment is proposed to improve the performance. Firstly, the statistical properties of the feature of a typical maritime scene are collected and analyzed for the survival time and reprojection error at different areas in the image. Then, a water-shoreline detection (WSD) algorithm is adopted to divide the scene into two parts, i.e., water area and non-water area, discarding unstable features of the water surface. Finally, the Visual Odometry (VO) method is customized to generate navigation trajectories. The results demonstrate that this method is robust, and WSD improves the accuracy of trajectory estimation. © 2022 IEEE.

关键词： Vision

来源：评论

学校读者我要写书评

暂无评论

An End-to-end Foreground Category Attention Boosting Network 12

An End-to-end Foreground Category Attention Boosting Network

引用

12th international Conference on Information Technology in Medicine and Education, ITME 2022

作者： Zhao, Guangwei Zhang, Yu Sun, Li Gao, Jinfeng Liu, Fang Zhang, Renzhi Huanghuai University College of Computer and Artificial Intelligence Zhumadian China Henan Provincial Key Laboratory of Smart Lighting Zhumadian China Henan International Joint Laboratory of Behavior Optimization Control for Smart Robots Zhumadian China

ISBN: (纸本)9798350310153

In this paper, we introduce a foreground category attention boosting (CAB) framework for coarse-to-fine image segmentation, which makes full use of the coarse category information in low resolution layers. Our contributions can be summarized as: Firstly, we introduce the category attention mechanism to build the CAB module, which realizes a new stronger learner by aggregating the coarse category information and the feature maps of the current layer. Secondly, we propose an end-to-end foreground category attention boosting (FAB) neural network, which fully utilizes the complementary effects among different resolution feature maps and enhances the contribution from low resolution layers to fine segmentation with low computational cost. Thirdly, Extensive experiments on public dataset show that our approach can achieve superior performance over the state-of-the-art methods. © 2022 IEEE.

关键词： Image segmentation

来源：评论

学校读者我要写书评

暂无评论

Personalized Custom Virtual Fitting Display Method 12

Personalized Custom Virtual Fitting Display Method

引用

12th international Conference on Information Technology in Medicine and Education, ITME 2022

作者： Zhu, Yuxiang Wu, Haitao Li, Gangqiang Gao, Jinfeng Liu, Shuan Zhang, Yu Zhang, Junming Huanghuai University College of Computer and Artificial Intelligence Zhumadian China Huanghuai University Henan Provincial Key Laboratory of Smart Lighting Zhumadian China Huanghuai University Henan International Joint Laboratory of Behavior Optimization Control for Smart Robots Zhumadian China

ISBN: (纸本)9798350310153

With the maturity of 3D and AR technologies, many e-commerce platforms have launched the function of virtual fitting rooms. Although it attracted a lot of attention, the real experience is not good. At present, the main problems of many virtual fitting platforms are inaccurate in clothing version and size, and the dressing effect is quite different. Therefore, the virtual fitting platform to customize personalized clothing has potential application value. In this study, we consider trying on different clothing on a detailed 3D mannequin to increase the experience. First, a human body database is established, and then the fine faces are obtained according to the 3D face scanning technology. According to the synthesis technology of human body and human face, a virtual human body with fine facial features is obtained. Finally, the different fabrics clothes can try on the virtual body to show the effect. The experimental results show that this method can quickly try on clothes of different fabrics to complete personalized customization. © 2022 IEEE.

关键词： Virtual reality

来源：评论

学校读者我要写书评

暂无评论

Dynamic Adjustment of Packet Based on Fair Bandwidth Allocation in IEEE 802.11 Wireless Networks 12

Dynamic Adjustment of Packet Based on Fair Bandwidth Allocat...

引用

12th international Conference on Information Technology in Medicine and Education, ITME 2022

作者： Zou, Xiong Gao, Jinfeng Zhang, Yu Li, Huyong Liu, Fang Wu, Haitao Lu, Shejie Hu, Siping Huanghuai University College of Computer and Artificial Intelligence International Joint Laboratory of Behavior Optimization Control for Smart Robots Zhumadian China Hubei University of Science and Technology Institute of Engineering and Technology Xianning China

ISBN: (纸本)9798350310153

In view of the imbalance of upstream and downstream flows in the access of local manufacturing resources, this paper proposes a dynamic adjustment algorithm based on bandwidth fair allocation (DAPFBA, Dynamic Adjustment of Packet based on Fair Bandwidth Allocation), which dynamically adjusts the packet size of nodes according to the number of downflows and realizes the downstream throughput balance, so as to realize the fair distribution of upstream and downstream bandwidth. According to the actual situation of the network, the algorithm can be implemented in the AP without modifying the MAC of other wireless nodes, which saves the workload of the implementation. Especially in the case of more and scattered wireless nodes, it only needs to modify the AP without modifying other nodes, eliminating a lot of work to configure the decentralized nodes. © 2022 IEEE.

关键词： Bandwidth

来源：评论

学校读者我要写书评

暂无评论

Performance Improvement Strategy for Differentiated Services in Multi-rate IEEE 802.11 Wireless Networks 12

Performance Improvement Strategy for Differentiated Services...

引用

12th international Conference on Information Technology in Medicine and Education, ITME 2022

作者： Xu, Haihong Lu, Shejie Hu, Siping Gao, Jinfeng Liu, Fang Wu, Haitao Hubei University of Science and Technology Information Center Institute of Engineering and Technology Xianning437100 China Huanghuai University International Joint Laboratory of Behavior Optimization Control for Smart Robots College of Computer and Artificial Intelligence Zhumadian463000 China

ISBN: (纸本)9798350310153

With the wide application of wireless LAN, the real-time and differentiated services of IEEE 802.11 MAC are becoming more and more important. IEEE 802.11 uses CSMA/CA mechanism in MAC layer to coordinate access, but it does not meet the different requirements of different users for QoS. Considering the different requirements of different priority users for QoS, a solution is proposed for this problem. The core idea of this strategy is: in a multi rate network, according to the priority of nodes in the network, through adaptive optimization of relevant parameters, the probability of high priority nodes accessing the channel can be improved, and its throughput can be improved accordingly;The channel access probability of nodes with lower priority is reduced accordingly, and the throughput is reduced accordingly. Therefore, this policy can expand the throughput of each priority and realize the service differentiation function. © 2022 IEEE.

关键词： IEEE Standards

来源：评论

学校读者我要写书评

暂无评论

Secure and Efficient Gossip Learning with Adversarial Nodes

Secure and Efficient Gossip Learning with Adversarial Nodes

引用

IEEE international Conference on Computation, Communication and Engineering (ICCCE)

作者： Gangqiang Li Haitao Wu Huyong Li Yuxiang Zhu Junming Zhang Yu Zhang Jinfeng Gao Henan International Joint Laboratory of Behavior Optimization Control for Smart Robots & Henan Provincial Key Laboratory of Smart Lighting College of Computer and Artificial Intelligence Huanghuai University Zhumadian China

ISBN: (纸本)9781665482493

Gossip learning is one of the most advanced approaches in decentralized techniques and performs almost as well as centralized training. However, gossip learning is inherently vulnerable to malicious attacks from adversarial nodes due to its flat architecture, which leads to performance degradation of collaborative training nodes. To reduce the negative impact of adversarial nodes, we propose an effective strategy to protect the distributed training process. The results show that the proposed strategy is robust to malicious attacks.

关键词： Training Degradation Focusing Collaboration Interference Computer architecture

来源：评论

学校读者我要写书评

暂无评论

A Dynamic Time Slot Algorithm Based on Throughput optimization in IEEE 802.11 Wireless Networks 12

A Dynamic Time Slot Algorithm Based on Throughput Optimizati...

引用

12th international Conference on Information Technology in Medicine and Education, ITME 2022

作者： Jin, Guonian Lu, Shejie Hu, Siping Xu, Haihong Gao, Jinfeng Liu, Fang Wu, Haitao Hubei University of Science and Technology School of Computer Science and Technology Institute of Engineering and Technology Xianning437100 China Huanghuai University International Joint Laboratory of Behavior Optimization Control for Smart Robots College of Computer and Artificial Intelligence Zhumadian463000 China

ISBN: (纸本)9798350310153

In IEEE 802.11 wireless LAN, as the number of nodes increases, collisions will increase correspondingly, channel utilization will decline, and the total throughput will decline instead of increasing, leading to system performance degradation. Through in-depth analysis of the root cause of this problem, this paper designs a simple and effective algorithm, which uses ARMA filtering algorithm to measure the conditional collision probability and then calculate the number of nodes in the network. The AP tells all nodes in the network in the form of broadcast when the number of nodes in the network changes, after all nodes receive the broadcast sent by the AP, they dynamically adjust the network parameters accordingly. Finally, NS2 simulation software is used to conduct simulation experiments in a variety of network scenarios, and the experimental results verify that the algorithm in this paper is simple and effective, with less changes to IEEE 802.11, and is suitable for use and promotion in wireless networks. It can optimize the network system performance based on the number of nodes, improve the system throughput, and significantly improve the network performance. According to the number and density of nodes, the algorithm reduces the collision probability by dynamically adjusting the timeslot, accordingly increases the total network throughput, and achieves the purpose of effectively improving the network performance. © 2022 IEEE.

关键词： IEEE Standards

来源：评论

学校读者我要写书评

暂无评论

Contention Window Adaptive Adjustment Strategy for Fairness in Multi-rate IEEE 802.11 Network 6

Contention Window Adaptive Adjustment Strategy for Fairness ...

引用

6th international Conference on Communication and Information Systems, ICCIS 2022

作者： Lu, Shejie Hu, Siping Duan, Yuntao Gao, Jinfeng Liu, Fang Hubei University of Science and Technology Institute of Engineering and Technology Xianning437100 China Xi'An Technological University School of Mechatronic Engineering Xi'an710000 China Huanghuai University International Joint Laboratory of Behavior Optimization Control for Smart Robots Zhumadian463000 China

ISBN: (数字)9781665463850

ISBN: (纸本)9781665463850

The wireless network technology represented by IEEE 802.11 technology plays an increasingly important role in practice. However, in the multi-node and multi-rate network scenario, this technology will produce abnormal performance, resulting in high-speed and medium-speed nodes. The throughput drops to the same as the throughput of the low-speed node, which reduces the total throughput of the system and degrades the performance. Aiming at this problem, a solution (TRCWAA) is designed. Through the minimum contention window, the size of the minimum contention window of each node is inversely proportional to its transmission rate, so as to realize the reasonable allocation of channel resources. Theoretical analysis and simulation experiments show that the TRCWAA strategy can well solve the problem of abnormal performance and unfairness, greatly improve the network fairness, and improve the total throughput of the system. In addition, the significant advantage of this method is that it only needs to make minor modifications on the basis of 802.11, that is, to adjust the minimum contention window size according to the transmission rate. Its physical implementation is very simple and convenient. No modification to the hardware, so it is easy to implement and generalize in real networks. © 2022 IEEE.

关键词： IEEE Standards

来源：评论

学校读者我要写书评

暂无评论

A Framework for simulation control Systems Based on Software Product Line 12

A Framework for simulation control Systems Based on Software...

引用

12th international Conference on Information Technology in Medicine and Education, ITME 2022

作者： Huijuan, Xu Xu, Li Xiaoning, Sun Huanghuai University Dept. Computer and Artificial Intelligence ZhuMaDian China Henan International Joint Laboratory of Behavior Optimization Control for Smart Robots ZhuMaDian China Huanghuai University Dept. Intelligent Manufacturing ZhuMaDian China ZhuMaDian Vocational and Technical College Dept. Economic Management ZhuMaDian China

ISBN: (纸本)9798350310153

At present, the fast-changing functional requirements of simulation control system in Aerospace and navigation, where various hardware boards are used, have caused the high complexity and long development cycle of control systems. Although such systems have a certain degree of similarity, they are not well reused. In order to solve these problems, this paper puts forward the idea of using the software product line to model the control system of spacecraft and designs a framework. Using this framework, a simulation wave data acquisition software is developed. The application of this framework can achieve maximum granularity of software reuse, and improve the development efficiency. © 2022 IEEE.

关键词： control systems

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：