检索结果-内蒙古大学图书馆

Pedestrian and Vehicle Detection Based on Pruning YOLOv4 with Cloud-Edge Collaboration

computer Modeling in engineering & Sciences 2023年第11期137卷 2025-2047页

作者： Huabin Wang Ruichao Mo Yuping Chen Weiwei Lin Minxian Xu Bo Liu School of Computer Science and Engineering Huizhou UniversityHuizhou516007China School of Computer Science and Engineering South China University of TechnologyGuangzhou510006China School of Computer Science and Technology South China Normal UniversityGuangzhou510631China Department of New Network Technologies Peng Cheng LaboratoryShenzhen518066China Shenzhen Institute of Advanced Technology Chinese Academy of SciencesShenzhen518055China

Nowadays,the rapid development of edge computing has driven an increasing number of deep learning applications deployed at the edge of the network,such as pedestrian and vehicle detection,to provide efficient intelligent services to mobile ***,as the accuracy requirements continue to increase,the components of deep learning models for pedestrian and vehicle detection,such as YOLOv4,become more sophisticated and the computing resources required for model training are increasing dramatically,which in turn leads to significant challenges in achieving effective deployment on resource-constrained edge devices while ensuring the high accuracy *** addressing this challenge,a cloud-edge collaboration-based pedestrian and vehicle detection framework is proposed in this paper,which enables sufficient training of models by utilizing the abundant computing resources in the cloud,and then deploying the well-trained models on edge devices,thus reducing the computing resource requirements for model training on edge ***,to reduce the size of the model deployed on edge devices,an automatic pruning method combines the convolution layer and BN layer is proposed to compress the pedestrian and vehicle detection model *** results show that the framework proposed in this paper is able to deploy the pruned model on a real edge device,Jetson TX2,with 6.72 times higher ***,the channel pruning reduces the volume and the number of parameters to 96.77%for the model,and the computing amount is reduced to 81.37%.

关键词： Pedestrian and vehicle detection YOLOv4 channel pruning cloud-edge collaboration

来源：评论

学校读者我要写书评

暂无评论

Few-Shot Partial Multi-View Learning

引用

IEEE Transactions on Pattern Analysis and Machine Intelligence 2023年第10期45卷 11824-11841页

作者： Zhou, Yuan Guo, Yanrong Hao, Shijie Hong, Richang Luo, Jiebo Ministry of Education Hefei University of Technology The Key Laboratory of Knowledge Engineering with Big Data Hefei230009 China Hefei University of Technology The School of Computer Science and Information Engineering Hefei230009 China University of Rochester The Department of Computer Science RochesterNY14627 United States

It is often the case that data are with multiple views in real-world applications. Fully exploring the information of each view is significant for making data more representative. However, due to various limitations and failures in data collection and pre-processing, it is inevitable for real data to suffer from view missing and data scarcity. The coexistence of these two issues makes it more challenging to achieve the pattern classification task. Currently, to our best knowledge, few appropriate methods can well-handle these two issues simultaneously. Aiming to draw more attention from the community to this challenge, we propose a new task in this paper, called few-shot partial multi-view learning, which focuses on overcoming the negative impact of the view-missing issue in the low-data regime. The challenges of this task are twofold: (i) it is difficult to overcome the impact of data scarcity under the interference of missing views;(ii) the limited number of data exacerbates information scarcity, thus making it harder to address the view-missing issue in turn. To address these challenges, we propose a new unified Gaussian dense-anchoring method. The unified dense anchors are learned for the limited partial multi-view data, thereby anchoring them into a unified dense representation space where the influence of data scarcity and view missing can be alleviated. We conduct extensive experiments to evaluate our method. The results on Cub-googlenet-doc2vec, Handwritten, Caltech102, Scene15, Animal, ORL, tieredImagenet, and Birds-200-2011 datasets validate its effectiveness. The codes will be released at https://***/zhouyuan888888/UGDA. © 1979-2012 IEEE.

关键词： Neural networks

来源：评论

学校读者我要写书评

暂无评论

A Real-time Approach for Multi-physics Field Reconstruction of Active Phased Array Antenna

引用

IEEE Transactions on Antennas and Propagation 2025年第6期73卷 3585-3592页

作者： Feng, Shizhe Chen, Qihang Li, Zhixiong Sotelo, Miguel Angel The State Key Laboratory of Reliability and Intelligence of Electrical Equipment Hebei University of Technology Tianjin300130 China The Faculty of Mechanical Engineering Opole University of Technology Opole45-758 Poland The Department of Computer Engineering University of Alcalá Madrid28801 Spain

This study proposes an effective approach to reconstruct the multi-physics field of the active phased array antenna (APAA) through limited information of temperature and strain. First, the multi-physics coupling model of the APAA is constructed to analyze the thermal-mechanical-electromagnetic responses;the model calculation accuracy is verified by experimental testing. Then, a large number of samples containing the temperature and deformation information are generated through the coupling model. Subsequently, the multi-physics field reconstruction model is established based on the extreme gradient boosting (XGBoost) method, in which the parallelization computing is implemented to fully improve the prediction efficiency. Lastly, the testing samples are used to evaluate the performance of the reconstruction model. The obtained results indicate that the developed model is able to reconstruct the multi-physics field of the APAA in real time more precisely through comparing with existing machine learning methods. This approach can provide reliable temperature and displacement reconstruction information for the compensation of the APAA electromagnetic performance. © 1963-2012 IEEE.

关键词： Antenna phased arrays

来源：评论

学校读者我要写书评

暂无评论

Enhancing WRSN Sustainability Through On-Demand Directional Wireless Charging With Multiple Green-Powered Mobile Vehicles

引用

IEEE Internet of Things Journal 2025年第11期12卷 18019-18030页

作者： Ma, Xiongbo Liu, Xilong Ansari, Nirwan Ding, Jun Yunnan University School of Information Science and Engineering Yunnan Kunming650500 China New Jersey Institute of Technology Advanced Networking Laboratory Helen and John C. Hartmann Department of Electrical and Computer Engineering NewarkNJ07102 United States

Current research on scheduling mobile charging vehicles (MCVs) generally focuses on periodic and omnidirectional charging of sensor nodes (SNs). However, this approach leads to significant energy wastage, especially when relying on fossil energy sources. In this work, we propose to schedule multiple green energy-powered MCVs equipped with directional antennas to efficiently charge SNs. We first develop an on-demand and directional charging model based on far-field wireless charging. Then, we formulate the SN survival maximization problem to efficiently prolong the lifetime of wireless rechargeable sensor networks (WRSNs). Given the NP-hard nature of this optimization problem, we develop the three-Step directIonal wireless charGiNg (SIGN) algorithm to efficiently solve this problem. SIGN strategically determines the MCVs’ anchor points (APs), traveling paths and wireless energy emitting directions. Finally, we conduct extensive simulation experiments to validate the superiority of our proposed algorithm in optimizing energy usage and prolonging network lifetime. © 2014 IEEE.

关键词： Wireless charging

来源：评论

学校读者我要写书评

暂无评论

Assessing radiographic findings on finger X-rays using an enhanced deep learning approach

引用

International Journal of Information Technology (Singapore) 2024年第7期16卷 4279-4288页

作者： Kumar, Rahul K, Suganya Devi Mohapatra, Devi Prasad Medical Imaging Laboratory Department of Computer Science and Engineering National Institute of Technology Assam Silchar 788010 India Department of Plastic Surgery JIPMER Puducherry 605006 India

The human upper extremity, which includes the shoulder, humerus, elbow, forearm, wrist, hand, and fingers, demonstrates incredible biological complexity, enabling us to make crucial and intricate movements in our daily lives. Due to its delicate construction, it is prone to a variety of fractures and injuries. Radiologists utilize a combination of visual perception, memory, pattern recognition, and cognitive reasoning when interpreting radiographic studies. However, a variety of circumstances, including clinical staff shortages, a lack of skilled radiologists, heavy caseloads, and continual distractions, can all hurt their performance, resulting in higher workloads, fatigue, wrong detection rates, and weariness. Addressing these challenges, our study introduced a novel approach employing an advanced deep learning model to enhance the accuracy and efficiency of diagnosing finger radiograph abnormalities. In this work, a modified DenseNet hybrid architecture was proposed for classifying abnormalities in finger radiographs. These models were trained using the finger dataset, a subset of the extensive Musculoskeletal Radiographs (MURA) dataset. The improved performance indicators for our proposed model over the current benchmark models demonstrated the model's effectiveness. The proposed models exhibited superior performance in Cohen's kappa metrics for classifying abnormalities in finger radiographs on the test dataset, outperforming other benchmark models. The advanced deep learning model based on DenseNet architecture showcased the promising potential for automating the classification of abnormalities in finger radiographs. © Bharati Vidyapeeth's Institute of computer Applications and Management 2024.

关键词： Convolutional neural network Deep learning DenseNet Finger radiographs Fracture detection Medical imaging

来源：评论

学校读者我要写书评

暂无评论

Pure Detail Feature Extraction Network for Visible-Infrared Re-Identification

引用

Intelligent Automation & Soft Computing 2023年第8期37卷 2263-2277页

作者： Jiaao Cui Sixian Chan Pan Mu Tinglong Tang Xiaolong Zhou Zhejiang University of Technology Hangzhou310023China Hubei Key Laboratory of Intelligent Vision Based Monitoring for Hydroelectric Engineering The College of Computer and InformationChina Three Gorges UniversityYichang443002China The College of Electrical and Information Engineering Quzhou UniversityQuzhou324000China

Cross-modality pedestrian re-identification has important appli-cations in the field of *** to variations in posture,camera per-spective,and camera modality,some salient pedestrian features are difficult to provide effective retrieval ***,it becomes a challenge to design an effective strategy to extract more discriminative pedestrian *** many effective methods for detailed feature extraction are proposed,there are still some shortcomings in filtering background and modality *** further purify the features,a pure detail feature extraction network(PDFENet)is proposed for *** includes three modules,adaptive detail mask generation module(ADMG),inter-detail interaction module(IDI)and cross-modality cross-entropy(CMCE).ADMG and IDI use human joints and their semantic associations to suppress background noise in *** guides the model to ignore modality noise by generating modality-shared feature ***,ADMG generates masks for pedestrian details based on pose *** are used to suppress background information and enhance pedestrian detail ***,IDI mines the semantic relations among details to further refine the ***,CMCE cross-combines classifiers and features to generate modality-shared feature labels to guide model *** ablation experiments as well as visualization results have demonstrated the effectiveness of PDFENet in eliminating background and modality *** addition,comparison experi-ments in two publicly available datasets also show the competitiveness of our approach.

关键词： Person re-identification multimedia image retrieval

来源：评论

学校读者我要写书评

暂无评论

Snowball Effect in Federated Learning: An Approach of Exponentially Expanding Structures for Optimizing the Training Efficiency

引用

IEEE Transactions on Cognitive Communications and Networking 2024年第3期11卷 1803-1817页

作者： Cheng, Guoliang Li, Peichun Tan, Beihai Yu, Rong Wu, Yuan Pan, Miao Guangdong University of Technology School of Automation Guangzhou China Guangdong Key Laboratory of IoT Information Technology Guangzhou China University of Macau State Key Laboratory of Internet of Things for Smart City China University of Macau Department of Computer and Information Science China Guangdong University of Technology School of Integrated Circuits Guangzhou China University of Houston Department of Electrical and Computer Engineering Houston United States

Efficient federated learning (FL) in mobile edge networks faces challenges due to energy-consuming on-device training and wireless transmission. Optimizing the neural network structures is an effective approach to achieving energy savings. In this paper, we present a Snowball FL training with expanding neural network structure, which starts with a small-sized submodel and gradually progresses to a full-sized model. To achieve this, we first design the submodel and embedding extraction schemes for fine-grained model structure expansion. We then investigate the joint minimization problem of the global training loss and system-wise energy consumption. After that, we decompose the optimization problem into a long-term model structure expansion subproblem and a single-round local resource allocation subproblem. Specifically, the former subproblem is transformed into a variational calculus problem by leveraging theoretical analysis of the convergence bound. The Euler-Lagrange method is used to derive the solution, where the optimal evolution strategy for the model structure exponentially increases with the global round (i.e., the Snowball effect). Meanwhile, the latter subproblem is solved by convex optimization to acquire the optimal computing frequency and transmission power. Experiments indicate that the proposed framework can save about 50% of energy consumption to achieve on-par accuracy against state-of-the-art algorithms. © 2015 IEEE.

关键词： Resource allocation

来源：评论

学校读者我要写书评

暂无评论

XDT-CXR: Investigating Cross-Disease Transferability in Zero-Shot Binary Classification of Chest X-Rays 9

XDT-CXR: Investigating Cross-Disease Transferability in Zero...

引用

9th Machine Learning for Healthcare Conference, MLHC 2024

作者： Rahman, Umaima Basu, Abhishek Khattak, Muhammad Uzair Rahman, Aniq Ur Department of Computer Vision Mohamed Bin Zayed University of Artificial Intelligence Abu Dhabi Masdar City United Arab Emirates Department of Engineering Science University of Oxford Oxford United Kingdom

This study explores the concept of cross-disease transferability (XDT) in medical imaging, focusing on the potential of binary classifiers trained on one disease to perform zero-shot classification on another disease affecting the same organ. Utilizing chest X-rays (CXR) as the primary modality, we investigate whether a model trained on one pulmonary disease can make predictions about another novel pulmonary disease, a scenario with significant implications for medical settings with limited data on emerging diseases. The XDT framework leverages the embedding space of a vision encoder, which, through kernel transformation, aids in distinguishing between diseased and non-diseased classes in the latent space. This capability is especially beneficial in resource-limited environments or in regions with low prevalence of certain diseases, where conventional diagnostic practices may fail. However, the XDT framework is currently limited to binary classification, determining only the presence or absence of a disease rather than differentiating among multiple diseases. This limitation underscores the supplementary role of XDT to traditional diagnostic tests in clinical settings. Furthermore, results show that XDT-CXR as a framework is able to make better predictions compared to other zero-shot learning (ZSL) baselines. Our code and pre-trained models are available at ***/rumaima/xdt cxr. © 2024 U. Rahman, A. Basu, M.U. Khattak & A.U. Rahman.

关键词： Medical imaging

来源：评论

学校读者我要写书评

暂无评论

Improved Dijkstra Algorithm for Mobile Robot Path Planning and Obstacle Avoidance

引用

computers, Materials & Continua 2022年第9期72卷 5939-5954页

作者： Shaher Alshammrei Sahbi Boubaker Lioua Kolsi Department of Mechanical Engineering College of EngineeringUniversity of Ha’ilHa’il2440Saudi Arabia Department of Computer and Network Engineering College of Computer Science and EngineeringUniversity of JeddahJeddah21959Saudi Arabia Research Laboratory of Metrology and Energy Systems National Engineering SchoolEnergy Engineering DepartmentUniversity of MonastirMonastir5000Tunisia

Optimal path planning avoiding obstacles is among the most attractive applications of mobile robots(MRs)in both research and *** this paper,an optimal collision-free algorithm is designed and implemented practically based on an improved Dijkstra *** achieve this research objectives,first,the MR obstacle-free environment is modeled as a diagraph including nodes,edges and ***,Dijkstra algorithm is used offline to generate the shortest path driving the MR from a starting point to a target *** its movement,the robot should follow the previously obtained path and stop at each node to test if there is an obstacle between the current node and the immediately following *** this aim,the MR was equipped with an ultrasonic sensor used as obstacle *** an obstacle is found,the MR updates its diagraph by excluding the corresponding ***,Dijkstra algorithm runs on the modified *** procedure is repeated until reaching the target *** verify the efficiency of the proposed approach,a simulation was carried out on a hand-made MR and an environment including 9 nodes,19 edges and 2 *** obtained optimal path avoiding obstacles has been transferred into motion control and implemented practically using line tracking *** study has shown that the improved Dijkstra algorithm can efficiently solve optimal path planning in environments including obstacles and that STEAM-based MRs are efficient cost-effective tools to practically implement the designed algorithm.

关键词： Mobile robot(MR) STEAM path planning obstacle avoidance improved dijkstra algorithm

来源：评论

学校读者我要写书评

暂无评论

DeepGCN based on variable multi‐graph and multimodal data for ASD diagnosis

引用

CAAI Transactions on Intelligence Technology 2024年第4期9卷 879-893页

作者： Shuaiqi Liu Siqi Wang Chaolei Sun Bing Li Shuihua Wang Fei Li College of Electronic and Information Engineering Hebei UniversityBaodingChina Machine Vision Technology Innovation Center of Hebei Province BaodingChina The State Key Laboratory of Multimodal Artificial Intelligence Systems Institute of AutomationChinese Academy of SciencesBeijingChina School of Computer Science and Technology Henan Polytechnic UniversityJiaozuoChina

Diagnosing individuals with autism spectrum disorder(ASD)accurately faces great chal-lenges in clinical practice,primarily due to the data's high heterogeneity and limited sample *** tackle this issue,the authors constructed a deep graph convolutional network(GCN)based on variable multi‐graph and multimodal data(VMM‐DGCN)for ASD ***,the functional connectivity matrix was constructed to extract primary ***,the authors constructed a variable multi‐graph construction strategy to capture the multi‐scale feature representations of each subject by utilising convolutional filters with varying kernel ***,the authors brought the non‐imaging in-formation into the feature representation at each scale and constructed multiple population graphs based on multimodal data by fully considering the correlation between *** extracting the deeper features of population graphs using the deep GCN(DeepGCN),the authors fused the node features of multiple subgraphs to perform node classification tasks for typical control and ASD *** proposed algorithm was evaluated on the Autism Brain Imaging Data Exchange I(ABIDE I)dataset,achieving an accuracy of 91.62%and an area under the curve value of 95.74%.These results demon-strated its outstanding performance compared to other ASD diagnostic algorithms.

关键词： machine learning medical image processing medical signal processing

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：