检索结果-内蒙古大学图书馆

SSENet: A Multiscale 3-D Convolutional Neural Network for InSAR Shift Estimation

IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING 2023年 61卷 1页

作者： Wu, Yulun Wang, Jili Zhang, Heng Zhao, Fengjun Xiang, Wei Li, Hongxiang Wang, Huaishuai An, Lianshuo Chinese Acad Sci Aerosp Informat Res Inst Dept Space Microwave Remote Sensing Syst Beijing 100094 Peoples R China Univ Chinese Acad Sci Sch Elect Elect & Commun Engn Beijing 100190 Peoples R China

The interferometric synthetic aperture radar (InSAR) image shift measurement technique is of great significance in processing high-precision digital elevation model (DEM) generation and deformation measurements. It can be used in steps such as image fine coregistration, interferometric phase unwrapping (PU), and absolute phase calibration in the InSAR processing flow without an external DEM. However, the shifts estimated by current methods are of low resolution and have high measurement noise, which may have adverse impacts on subsequent applications. In this article, a lightweight, high-resolution, and low-noise interferometric stereo-radargrammetric shift estimation network (SSENet) is proposed to solve the aforementioned problems. It introduces deep-learning technology to the InSAR shift estimation task for the first time. We propose forming multiscale 3-D coherence coefficient cubes (CC-Cubes) by projecting the shift values of the images onto the third dimension and then using a 3-D convolutional network for multiscale fusion and encoding, followed by decoding with linear layers. In addition, a dataset generation and augmentation scheme based on real data is designed for model training and evaluation. Several sets of real SAR images from different regions of the world were used to evaluate SSENet. Compared with the typical coherent cross correlation (CCC) approach, SSENet reduces the mean absolute error (MAE) of the estimated shifts by approximately 79% while improving the resolution by a factor of 4 x 4 , making it possible to restore the absolute interferometric phase. Finally, we demonstrate a stitching strategy for processing large-scale SAR images and discuss the multiple potential uses of SSENet in the InSAR processing chain.

关键词： Estimation Azimuth Radar polarimetry image resolution Deformation Three-dimensional displays Task analysis deep learning image coregistration interferometric synthetic aperture radar (InSAR) processing stereo-radargrammetry

来源：评论

学校读者我要写书评

暂无评论

deep learning based classification of COVID-19 severity using hierarchical deep maxout model

引用

BIOMEDICAL SIGNAL processing AND CONTROL 2024年 88卷

作者： Rao, M. Surya Bhupal Rao, Y. Mallikarjuna Venkataiah, C. Murthy, G. L. N. Dharani, M. Jayamma, Manjula CVR Coll Engn Dept CSE Ibrahimpatnam 501510 Telangana India Santhiram Engn Coll Dept ECE Nandyal 518501 Andhra Pradesh India Rajeev Gandhi Mem Coll Engn & Technol Dept ECE Nandyal 518501 Andhra Pradesh India Lakireddy Bali Reddy Coll Engn Dept ECE Mylavaram 521230 Andhra Pradesh India Mohan Babu Univ Dept ECE Tirupati 517502 Andhra Pradesh India Mylavaram Engn Coll Dept CSE Nandyal 518501 Andhra Pradesh India

COVID-19 is a huge pandemic disease and is gradually growing into our daily lives. It has affected us all in different ways, which the World Health Organization classified as a pandemic in 2020. A number of deep learning methods have been developed to identify COVID-19 in chest X-rays, but very few techniques are used to estimate the severity of COVID-19. Additionally, it is very complex to effectively train the data, resulting in slower convergence and time complexities. Therefore, to overcome these drawbacks, this study proposes a unique model to classify the severity of Covid-19 using a deep learning method. The pre-processing step includes image enhancement and noise reduction using the Amended Bilateral Filter (ABF). The relevant features are then extracted using an ensemble-based deep learning model, which includes the DenseNet-201, ResNet-152 and VGG-19 models, which are integrated via an ensemble average approach. The most optimal features are selected using the Modified Archer Fish Optimization Algorithm (Mod-AFO). The proposed optimization algorithm exhibits high convergence and reduced model complexity. COVID-19 severity classification can be effectively categorized using a Multihead attention-assisted hierarchical deep-maxout network model (Multi-HDM). The proposed classification model is comparatively faster in severity estimation and high in performance. Using Python, the proposed model performances are analyzed and compared with the existing models to demonstrate the superiority of the proposed model. The proposed model achieved 98.2% accuracy, 98.5% sensitivity, 96.4% specificity and 96.4% precision in COVID-19 severity estimation using the COVID-19 Chest X-ray images and Lung masks dataset.

关键词： COVID-19 Severity estimation Amended Bilateral Filter DenseNet-201 Modified Archer Fish Multihead attention Hierarchical deep Maxout network Chest X-ray ResNet-152 VGG-19 deep learning

来源：评论

学校读者我要写书评

暂无评论

Optimizing 3D UNet for real-time Kidney Segmentation in 3D Ultrasound: A Comparative Study of Loss Functions

Optimizing 3D UNet for Real-Time Kidney Segmentation in 3D U...

引用

2025 Conference on Medical Imaging

作者： Valente, S. Torres, H. Morais, P. Fritz, A. Buschle, L. Lima, E. Vilaca, J. L. IPCA 2Ai Appl Artificial Intelligence Lab Sch Technol Barcelos Portugal Univ Minho Life Hlth Sci Res Inst ICVS Sch Med Braga Portugal ICVS Braga Portugal 3Bs PT Govt Associate Lab Guimaraes Portugal LASI Associate Lab Intelligent Syst Guimaraes Portugal Karl Storz SE Co KG Tuttlingen Germany

ISBN: (纸本)9781510685901

In the field of medical imaging, real-time segmentation of 3D ultrasound (US) volumes of the kidney is crucial for a variety of clinical applications. Recently, deep learning methods have been applied with excellent performance, where deeper and more complex networks are being proposed. However, increasing the complexity of the network can lead to challenges in maintaining real-time application. Therefore, loss functions play a very important role in optimizing state-of-the-art networks that not only demonstrate accurate segmentation performance but also remain suitable for real-time segmentation. This study investigates the impact of different loss functions on the training of 3DUNet for the segmentation of 3D US volumes of the kidney. The aim is to evaluate and compare ten loss functions to enhance the performance of the 3DUNet, while preserving its benefits of reduced computational cost and fast inference times. The research was conducted utilizing a dataset of 66 3D US volumes, randomly divided into 46 volumes for training, 10 for validations, and 10 for testing. Despite the different loss functions evaluated having received interesting segmentations results, the study demonstrated that a boundary-based loss was the most effective for this application. It achieves a superior balance of high Dice scores, low average surface distance (AverSSD) and competitive 95th percentile of the Hausdorff distance (HffSSD95), improving both segmentation quality and boundary precision, maintaining the computational resources of the 3DUNet. The overall findings of this study demonstrate that carefully selected loss functions can enhance network optimization, highlighting the feasibility of optimized deep learning networks for real-time kidney segmentation in 3D US imaging for clinical procedures.

关键词： Kidney deep learning Segmentation real-time Loss Optimization

来源：评论

学校读者我要写书评

暂无评论

Transition Is a Process: Pair-to-Video Change Detection Networks for Very High Resolution Remote Sensing images

引用

IEEE TRANSACTIONS ON image processing 2023年 32卷 57-71页

作者： Lin, Manhui Yang, Guangyi Zhang, Hongyan Wuhan Univ State Key Lab Informat Engn Surveying Mapping & Re Wuhan 430072 Peoples R China Wuhan Univ Sch Elect Informat Wuhan 430072 Peoples R China

As an important yet challenging task in Earth observation, change detection (CD) is undergoing a technological revolution, given the broadening application of deep learning. Nevertheless, existing deep learning-based CD methods still suffer from two salient issues: 1) incomplete temporal modeling, and 2) space-time coupling. In view of these issues, we propose a more explicit and sophisticated modeling of time and accordingly establish a pair-to-video change detection (P2V-CD) framework. First, a pseudo transition video that carries rich temporal information is constructed from the input image pair, interpreting CD as a problem of video understanding. Then, two decoupled encoders are utilized to spatially and temporally recognize the type of transition, and the encoders are laterally connected for mutual promotion. Furthermore, the deep supervision technique is applied to accelerate the model training. We illustrate experimentally that the P2V-CD method compares favorably to other state-of-the-art CD approaches in terms of both the visual effect and the evaluation metrics, with a moderate model size and relatively lower computational overhead. Extensive feature map visualization experiments demonstrate how our method works beyond making contrasts between bi-temporal images. Source code is available at https://***/Bobholamovic/CDLab.

关键词： Change detection deep learning convolutional neural network remote sensing very high resolution image

来源：评论

学校读者我要写书评

暂无评论

Priority-Aware Resource Allocation for VNF Deployment in Service Function Chains Based on Graph Reinforcement learning

引用

Computers, Materials & Continua 2025年第5期83卷 1649-1665页

作者： Seyha Ros Seungwoo Kang Taikuong Iv Inseok Song Prohim Tam Seokhoon Kim Department of Software Convergence Soonchunhyang UniversityAsan31538Republic of Korea School of Digital Technologies American University of Phnom PenhPhnom Penh12106Cambodia Department of Computer Software Engineering Soonchunhyang UniversityAsan31538Republic of Korea

Recently,Network Functions Virtualization(NFV)has become a critical resource for optimizing capability utilization in the 5G/B5G *** decomposes the network resource paradigm,demonstrating the efficient utilization of Network Functions(NFs)to enable configurable service priorities and resource *** Service Providers(TSPs)face challenges in network utilization,as the vast amounts of data generated by the Internet of Things(IoT)overwhelm existing *** applications,which generate massive volumes of diverse data and require real-time communication,contribute to bottlenecks and *** this context,Multiaccess Edge Computing(MEC)is employed to support resource and priority-aware IoT applications by implementing Virtual Network Function(VNF)sequences within Service Function Chaining(SFC).This paper proposes the use of deep Reinforcement learning(DRL)combined with Graph Neural Networks(GNN)to enhance network processing,performance,and resource pooling *** facilitates feature extraction through Message-Passing Neural Network(MPNN)*** with DRL,deep Q-Networks(DQN)are utilized to dynamically allocate resources based on IoT network priorities and *** focus is on minimizing delay times for VNF instance execution,ensuring effective resource placement,and allocation in SFC deployments,offering flexibility to adapt to real-time changes in priority and *** results demonstrate that our proposed scheme outperforms reference models in terms of reward,delay,delivery,service drop ratios,and average completion ratios,proving its potential for IoT applications.

关键词： deep reinforcement learning graph neural network multi-access edge computing network functions virtualization software-defined networking

来源：评论

学校读者我要写书评

暂无评论

AI-Driven image Recognition System for Automated Offside and Foul Detection in Football Matches Using Computer Vision

引用

INTERNATIONAL JOURNAL OF ADVANCED COMPUTER SCIENCE AND APPLICATIONS 2025年第1期16卷 1191-1198页

作者： Zhang, Qianwei Yu, Lirong Yan, Wenke Chengdu Sport Univ Chengdu 610041 Sichuan Peoples R China Sichuan Univ Sch Phys Educ Chengdu 610065 Sichuan Peoples R China Sichuan Univ High Sch 12 High Sch Chengdu Chengdu 610061 Sichuan Peoples R China

artificial intelligence (AI) and computer vision in sports analytics has transformed decision-making processes, enhancing fairness and efficiency. This paper proposes a novel AI-driven image recognition system for automatically detecting offside and foul events in football matches. Unlike conventional methods, which rely heavily on manual intervention or traditional image processing techniques, our approach utilizes a hybrid deep learning model that combines advanced object tracking with motion analysis to deliver real-time, precise event detection. The system employs a robust, self-learning algorithm that leverages spatiotemporal features from match footage to track player movements and ball dynamics. By analyzing the continuous flow of video data, the model detects offside positions and identifies foul types such as tackles, handballs, and dangerous play-through a dynamic pattern recognition process. This multitiered approach overcomes traditional methods' limitations by accurately identifying critical events with minimal latency, even in complex, high-speed scenarios. In experiments conducted on diverse datasets of live match footage, the system achieved an overall accuracy of 99.85% for offside detection and 98.56% for foul identification, with precision rates of 98.32% and 97.12%, respectively. The system's recall rates of 97.45% for offside detection and 96.85% for foul recognition demonstrate its reliability in real-world applications. It's clear from these results that the proposed framework can automate and greatly enhance the accuracy of match analysis, making it a useful tool for both referees and broadcasters. The system's low computational overhead and growing ability make connecting to existing match broadcasting infrastructure easy. This establishes an immediate feedback loop for use during live games. This work marks a significant step forward in applying AI and computer vision for sports, introducing a powerful method to enhance the objectivity and

关键词： Artificial intelligence image recognition automation foul detection deep learning computer vision computer vision

来源：评论

学校读者我要写书评

暂无评论

Enhancing Video Surveillance with deep learning-Based real-time Handgun Detection and Tracking 1

引用

8th International Conference on Computer Vision and image processing (CVIP)

作者： Yadav, Pavinder Gupta, Nidhi Sharma, Pawan Kumar Natl Inst Technol Hamirpur Hamirpur 177005 Himachal Prades India Natl Inst Technol Kurukshetra Kurukshetra 136119 Haryana India

ISBN: (数字)9783031581816

ISBN: (纸本)9783031581809;9783031581816

Handguns, pistols, and revolvers are commonly used in today's world for committing criminal acts, requiring the need for effective surveillance and control systems. However, despite the advancement of security systems, human monitoring and involvement are still necessary to effectively combat these crimes. This paper provides a robust automated handgun identification technique for recorded videos and live CCTV footage that may be used for both control and surveillance purposes. Automatic detection of firearms is crucial for improving people's protection and safety, however, it is a challenging task because of the numerous differences in design, size, and appearance of firearms. In recent years, object detectors have improved, yielding better findings and shorter inference times. The authors used cutting-edge object detector YOLOv7 for firearm detection. A varied and demanding dataset of 15,367 images for weapon identification is also proposed, which is carefully annotated for weapon localization and classification. After analysing the data, it is determined that the model achieves an accuracy rate of 96.80% and recall rate of 90.37%.

关键词： deep learning Machine learning Weapon Detection Surveillance System YOLO

来源：评论

学校读者我要写书评

暂无评论

A lightweight optical flow model for particle image velocimetry

引用

FLOW MEASUREMENT AND INSTRUMENTATION 2025年 102卷

作者： Shan, Liang Lou, Xiaoying Xiong, Junzhe Hong, Bo Jian, Juan Kong, Ming China Jiliang Univ Coll Informat Engn Key Lab Electromagnet Wave Informat Technol & Metr Hangzhou 310018 Peoples R China China Jiliang Univ Coll Metrol & Measurement Engn Hangzhou 310018 Peoples R China

Particle image Velocimetry (PIV) is a reliable non-contact velocimetry technique in fluid visualization, in which the method incorporating deep neural networks outperforms the mainstream conventional algorithms in terms of reconstruction accuracy and computation time. However, some optical flow models suffer from large parameter quantities and high complexity. This paper proposes a lightweight network, which combines blueprint convolution and attention mechanism based on the improved LiteFlowNet3 with supervised learning. The network uses blueprint convolution instead of ordinary convolution to compress the number of model parameters. An attention mechanism is introduced to enhance the extraction of particle image features and improve the accuracy of flow inference. It aims to reduce the number of model parameters and computational complexity while maintaining model accuracy. The experimental results show that the improved lightweight model reduces the average endpoint error by 37.71 % compared to the pre-improved model. Finally, a 2D PIV experimental system was constructed to capture particle image pairs. By applying this model to reconstruct the velocity field, the experimental results proved its feasibility and adaptability in processing real-world images.

关键词： Particle image velocimetry Optical flow Blueprint convolution Attention mechanism Average endpoint error

来源：评论

学校读者我要写书评

暂无评论

Current advances and future perspectives of image fusion: A comprehensive review

引用

INFORMATION FUSION 2023年 90卷 185-217页

作者： Karim, Shahid Tong, Geng Li, Jinyang Qadir, Akeel Farooq, Umar Yu, Yiting Northwestern Polytech Univ Shenzhen Res & Dev Inst Shenzhen 518057 Peoples R China Northwestern Polytech Univ Coll Mech Engn Xian 710072 Peoples R China Northwestern Polytech Univ Key Lab Micronano Syst Aerosp Minist Educ Xian 710072 Peoples R China Northwestern Polytech Univ Shaanxi Prov Key Lab Micro & Nano Electromech Syst Xian 701172 Peoples R China Northwestern Polytech Univ Res Ctr Smart Sensing Chips Ningbo Inst Ningbo 315103 Peoples R China Chinese Acad Sci Inst Biomed & Hlth Engn Shenzhen Inst Adv Technol Paul C Lauterbur Res Ctr Biomed Imaging 1068 Xueyuan Ave Shenzhen 518055 Peoples R China

Multiple imaging modalities can be combined to provide more information about the real world than a single modality alone. Infrared images discriminate targets with respect to their thermal radiation differences, and visible images are promising for texture details. On the other hand, polarized images deliver intensity and po-larization information, and multispectral images dispense the spatial, spectral, and temporal information depending upon the environment. Different sensors provide images with different characteristics, such as type of degradation, important features, textural attributes, etc. Several stimulating tasks have been explored in the last decades based on algorithms, performance assessments, processing techniques, and prospective applications. However, most of the reviews and surveys have not properly addressed the issues of additional possibilities of imaging fusion. The primary goal of this paper is to give a thorough overview of image fusion approaches, including associated background and current breakthroughs. We introduce image fusion and categorize the methods based on conventional image processing, deep learning (DL) architectures, and fusion scenarios. Further, we emphasize the recent DL developments in various image fusion scenarios. However, there are still several difficulties to overcome, including developing more advanced algorithms to support more dependable and real-time practical applications, discussed in future perspectives. This study can assist researchers in coping with multiple imaging modalities, recent fusion developments, and future perspectives.

关键词： image fusion Transform domain deep learning Multi-modal Polarization

来源：评论

学校读者我要写书评

暂无评论

real-time vehicular accident prevention system using deep learning architecture

引用

EXPERT SYSTEMS WITH APPLICATIONS 2022年 206卷

作者： Kabir, Md Faysal Roy, Sahadev NIT Arunachal Pradesh Dept Elect & Commun Engn Jote India

Approximately 20 million people die every year of road accidents, mainly caused due to ignorance of road safety norms and traffic rules. Drivers' experience and prudence still have to be relied upon to prevent vehicle acci-dents. Here, an algorithm is developed to prevent T-bone accidents, rear-end and head-on collisions, pedestrians accidents. Using Raspberry Pi 4 Model B with 8 GB RAM as processing unit, importing OpenCV library into python 3;an intensive dynamic real-time algorithm for the effective rear end and adjacent collision avoidance module is developed on the framework of deep learning architecture(DLA) for image categorization. In this work, real-time object identification, distance estimation, and tracking of the instantaneous position are performed in all environmental conditions using a deep learning algorithm without any additional sensors. The Raspberry Pi NoIR Camera Module V2 keeps sending the captured frames in real-time to feed the MobileNet DLA for Single Shot Multibox Detection (SSD). The attributes of the preceding obstructions are utilized in such a way that the live-distance between vehicles and other objects is approximated successfully by providing the instantaneous positional information whereby generating warning signal on exigency with a processing speed of 0.09 s per frame (11 fps). The real-time output data are analyzed by multivariate analysis and a 3x3x9 repeated measures ANOVA. The result shows that Multimodal Early Accident Alarming System is highly effective for distractive drivers.

关键词： Accident prevention deep learning architecture Repeated measures ANOVA real-time object identification Single Shot Multibox Detection

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：