检索结果-内蒙古大学图书馆

2025 IEEE International Conference on Acoustics, Speech, and Signal processing, ICASSP 2025

作者： Li, Guohao Yang, Hongyu Men, Yifang Huang, Di Li, Weixin Yang, Ruijie Wang, Yunhong School of Computer Science and Engineering Beihang University Beijing China School of Artificial Intelligence Beihang University Beijing China Shanghai Artificial Intelligence Laboratory Shanghai China

ISBN: (纸本)9798350368741

Generating animatable and editable 3D head avatars is essential for various applications in computer vision and graphics. Traditional 3D-aware generative adversarial networks (GANs), often using implicit fields like neural Radiance Fields (NeRF), achieve photorealistic and view-consistent 3D head synthesis. However, these methods face limitations in deformation flexibility and editability, hindering the creation of lifelike and easily modifiable 3D heads. We propose a novel approach that enhances the editability and animation control of 3D head avatars by incorporating 3D Gaussian Splatting (3DGS) as an explicit 3D representation. This method enables easier illumination control and improved editability. Central to our approach is the Editable Gaussian Head (EG-Head) model, which combines a 3D Morphable Model (3DMM) with texture maps, allowing precise expression control and flexible texture editing for accurate animation while preserving identity. To capture complex non-facial geometries like hair, we use an auxiliary set of 3DGS and tri-plane features. Extensive experiments demonstrate that our approach delivers high-quality 3D-aware synthesis with state-of-the-art controllability. Our code and models are available at https://***/liguohao96/EGG3D. © 2025 IEEE.

关键词： 3D-aware GAN image video synthesis

来源：评论

学校读者我要写书评

暂无评论

Realistic and Visually-Pleasing 3D Generation of Indoor Scenes from a Single image 7th

Realistic and Visually-Pleasing 3D Generation of Indoor Scen...

引用

7th Chinese Conference on Pattern Recognition and Computer Vision

作者： Li, Jie Wang, Lei Chen, Gongbin Li, Ang Qiu, Yuhao Wu, Jiaji Cheng, Jun Chinese Acad Sci Shenzhen Inst Adv Technol Shenzhen 518055 Peoples R China Univ Chinese Acad Sci CAS Beijing Peoples R China Shenzhen MSU BIT Univ Shenzhen Peoples R China Xidian Univ Sch Elect Engn Xian 710071 Peoples R China

ISBN: (纸本)9789819785070;9789819785087

artificial Intelligence Generated Content (AIGC) has experienced significant advancements, particularly in the areas of natural language processing and 2D image generation. However, the generation of three-dimensional (3D) content from a single image still poses challenges, particularly when the input image contains complex backgrounds. This limitation hinders the potential applications of AIGC in areas such as human-machine interaction, virtual reality (VR), and architectural design. Despite the progress made so far, existing methods face difficulties when dealing with single images that have intricate backgrounds. Their reconstructed 3D shapes tend to be incomplete, noisy, or lack of partial geometric structures. In this paper, we introduce a 3D generation framework for indoor scenes from a single image to generate realistic and visually-pleasing 3D geometry shapes, without the requirement of point clouds, multi-view images, depth or masks as input. The main idea of our method is clustering-based 3D shape learning and prediction, followed by a shape deformation. Since more than one objects tend to be existing in indoor scenes, our framework will simultaneously generate multi-objects and predict the layout with a camera pose, as well as 3D object bounding boxes for holistic 3D scene understanding. We have evaluated the proposed framework on benchmark datasets including ShapeNet, SUN RGB-D and Pix3D, and state-of-the-art performance has been achieved. We have also given examples to illustrate immediate applications in virtual reality.

关键词： 3D mesh Reconstruction Deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Character Recognition Using Convolutional neural networks Based on CUDA

Character Recognition Using Convolutional Neural Networks Ba...

引用

International Youth Conference on Radio Electronics, Electrical and Power Engineering (REEPE)

作者： Set Paing Tun Applied Mathematics and Artificial Intelligence Department National Research University “MPEI” Moscow Russian Federation

ISBN: (数字)9798331531836

ISBN: (纸本)9798331531843

The article explores character recognition using convolutional neural networks (CNNs) optimized with the CUDA platform to enhance computational efficiency. It outlines the CNN architecture, methods for leveraging GPU-based parallel data processing, and presents experimental results derived from the MNIST dataset. The study highlights that implementing CUDA drastically reduces processing time while maintaining a high level of predictive accuracy. The findings emphasize the potential of GPU acceleration in handling intensive computational tasks, making it a promising approach for real-time applications in image recognition and machine learning.

关键词： Deep learning Training Energy consumption Graphics processing units Parallel processing Real-time systems Energy efficiency Convolutional neural networks Character recognition Optimization

来源：评论

学校读者我要写书评

暂无评论

Bridging the Machine-Human Gap in Blurred-image Classification via Entropy Maximisation

Bridging the Machine-Human Gap in Blurred-Image Classificati...

引用

International image processing, applications and Systems Conference (IPAS)

作者： Emilio Sansano-Sansano Marina Martínez-García Javier Portilla INIT Universitat Jaume I Castellón de la Plana Spain IMAC Universitat Jaume I Castellón de la Plana Spain Instituto de Óptica CSIC Madrid Spain

ISBN: (数字)9798331506520

ISBN: (纸本)9798331506537

Recent studies point to an accuracy gap between humans and artificial neural Network (ANN) models when classifying blurred images, with humans outperforming ANNs. To bridge this gap, we introduce a spectral channel-based range-constrained entropy merit function, from which we devise a zero-phase, circular symmetric blind deblurring method. We apply it as a pre-processing step for image classification and test it using pre-trained classification models and images blurred by Gaussian kernels. We compare our method to state-of-the-art restoration methods, showing its superiority, effectively bridging the machine-human gap for most models and blur levels. Our results also rank higher than the competitors in no-reference and full-reference image quality metrics. Notwithstanding the limitation to zero-phase blur, this work shows that, for image pre-processing aimed at visual tasks, it may be advantageous to use merit functions based on vision science and information theory, rather than on the expected error to the latent image.

关键词： Measurement image quality Visualization Computer vision artificial neural networks Entropy Kernel Optimization Information theory image classification

来源：评论

学校读者我要写书评

暂无评论

Comparison of Vision Transformer with Convolutional neural networks for Brain Cancer Classification

Comparison of Vision Transformer with Convolutional Neural N...

引用

International Conference on Computer Research and Development (ICCRD)

作者： Doğu Manalı Hasan Demirel Electrical and Electronic Engineering Eastern Mediterranean University Famagusta Northern Cyprus

ISBN: (数字)9798331531881

ISBN: (纸本)9798331531898

Brain cancer is one of the most deadly illnesses. It causes abnormal cells to grow in the brain. Planning for treatment and the prognosis of patients with brain tumors depend greatly on early diagnosis. Brain tumors can have different characteristics, treatments, and forms. Consequently, the process of manually detecting brain tumors is difficult, labor-intensive, and error-prone. Doctors use magnetic resonance imaging to detect those abnormal cells in the brain. With the growth of artificial intelligence, it is possible to diagnose the brain tumor from MIR images. For instance, convolutional neural networks and transformers could be used. The self-attention mechanism is implemented by transformers, which are models that give each input data component a distinct weight. Transformers have limited applications in image classification tasks because they were originally designed for use in natural language processing applications. Thus far, the majority of image classification research has employed convolutional neural networks. In this paper, six different pretrained convolutional neural networks and a vision transformer are used to classify four distinct brain tumor classes. The models include ResNet50, AlexNet, VGG16, InceptionV3, MobileNetV2, FractalNet, and the Vision Transformer. The goal of this study is to compare the performance of these pretrained convolutional neural network models with that of the vision transformer, demonstrating that transformers can also be effectively applied to image classification tasks. The performance of a vision transformer model shows 84.39% accuracy in the classification problem, which is better than the other six architectures.

关键词： Computer vision Accuracy Magnetic resonance imaging Computer architecture Transformers Brain modeling Convolutional neural networks image classification Biomedical imaging Residual neural networks

来源：评论

学校读者我要写书评

暂无评论

International Conference on Innovative Methods and Practical applications for Cognizant Transportation Systems, IMPACTS 2023

International Conference on Innovative Methods and Practical...

引用

International Conference on Innovative Methods and Practical applications for Cognizant Transportation Systems, IMPACTS 2023

ISBN: (纸本)9789819772995

The proceedings contain 48 papers. The special focus in this conference is on Innovative Methods and Practical applications for Cognizant Transportation Systems. The topics include: Scientific Crash Investigation: A Case Study of the Bus Crash at Vadakkencherry, India;modelling of Lateral Positioning of Vehicles on Undivided Urban Roads in Mixed Traffic Conditions;application of Gap Acceptance Concept to Investigate Behaviour of Drivers at Roundabouts Under Mixed Traffic Conditions;intersection Traffic Data Extractor—A Tool for Extracting Traffic Volume and Speed;influence of Geometric and Traffic Characteristics on Priority Violations at Uncontrolled Intersections;road Safety Audit of Kollam Bypass and Traffic Management Measures for Safety Improvement;impact of Pedestrian Flow on Midblock Traffic Characteristics—A Comprehensive Review;road Safety Evaluation of Stretches Along Safe Corridor Demonstration Project in Kerala, India;modification of Crumb Rubber Modified Binders Using Fly Ash-Based Geopolymers;pavement Distress Detection Using image processing from Unmanned Aerial Vehicle Data;formulation of Design Catalogue for Low-Volume Interlock Cement Block Pavements;performance Analysis of Coir Reinforced Pavement For Rural Roads;pavement Deterioration Prediction Models for Non-urban Road networks in Kerala;rice Husk Ash Modified Asphalt: A High-Performing Bio-Binder for Asphalt Pavements;development of a 3-D Simulation Model of Flexible Pavement Responses Reinforced with Glass Fiber Geogrid Under Moving Load;effect of Soil Stabilization on Pavement Design and Its Cost-Effectiveness Using Cement in Soil Subgrade;prediction of California Bearing Ratio from Soil Index Properties Using artificial neural Network;influence of Different Warm Mix Chemical Additives on Characteristics of Warm Mix Asphalt.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Improving accuracy in Text Extraction From images using Region-Based Convolutional neural networks algorithm compared to Convolutional neural Network algorithm

Improving accuracy in Text Extraction From images using Regi...

引用

artificial Intelligence and Data Engineering (AIDE), International Conference on

作者： Navaprakash. N S. Vikram Reddy S. Dakshinesh Department of Electronics and Communication Engineering Saveetha School of Engineering SIMATS Chennai Department of Embedded System Saveetha School of Engineering SIMATS Chennai

ISBN: (数字)9798331527518

ISBN: (纸本)9798331527525

Text extraction is a critical task in data processing and analysis, requiring high accuracy for effective applications. This research employed a publicly accessible Kaggle dataset to assess the performance of two neural network architectures that are Region-Based Convolutional neural networks (R-CNN) with that of standard Convolutional neural networks (CNN). A total sample of 40 test images were taken and segregated into groups, with 20 samples allocated to each method. Implementation was conducted using Python, and statistical analysis was performed using ***. The study set a statistical power of 85%, with alpha level of 0.05 and 0.2 beta level set to ensure a thorough analysis. Performance was assessed based on accuracy, with R-CNN achieving 96.461%, significantly outperforming CNN, which recorded an accuracy of 83.583%. The statistical comparison yielded a two-tailed significance value of p < 0.001, highlighting the superior efficiency of R-CNN in text extraction tasks. These findings underscore the potential of R-CNN as a preferred approach for achieving higher accuracy in text extraction applications.

关键词： image segmentation Accuracy Text recognition Statistical analysis Optical character recognition Education neural networks Symbols Convolutional neural networks Standards

来源：评论

学校读者我要写书评

暂无评论

LoG-VMamba: Local-Global Vision Mamba for Medical image Segmentation 17th

LoG-VMamba: Local-Global Vision Mamba for Medical Image Seg...

引用

17th Asian Conference on Computer Vision, ACCV 2024

作者： Dang, Trung DQ. Nguyen, Huy Hoang Tiulpin, Aleksei University of Oulu Oulu Finland

ISBN: (纸本)9789819609000

Mamba, a State Space Model (SSM), has recently shown competitive performance to Convolutional neural networks (CNNs) and Transformers in Natural Language processing and general sequence modeling. Various attempts have been made to adapt Mamba to Computer Vision tasks, including medical image segmentation (MIS). Vision Mamba (VM)-based networks are particularly attractive due to their ability to achieve global receptive fields, similar to Vision Transformers, while also maintaining linear complexity in the number of tokens. However, the existing VM models still struggle to maintain both spatially local and global dependencies of tokens in high dimensional arrays due to their sequential nature. Employing multiple and/or complicated scanning strategies is computationally costly, which hinders applications of SSMs to high-dimensional 2D and 3D images that are common in MIS problems. In this work, we propose Local-Global Vision Mamba, LoG-VMamba, that explicitly enforces spatially adjacent tokens to remain nearby on the channel axis, and retains the global context in a compressed form. Our method allows the SSMs to access the local and global contexts even before reaching the last token while requiring only a simple scanning strategy. Our segmentation models are computationally efficient and substantially outperform both CNN and Transformers-based baselines on a diverse set of 2D and 3D MIS tasks. The implementation of LoG-VMamba is available at https://***/Oulu-IMEDS/LoG-VMamba. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

Robust occlusion-aware orbital angular momentum feature extraction via all-optical diffractive processing systems

引用

Optics Express 2025年第11期33卷 23053-23064页

作者： Keyao Li Yuetian Jia Min Gu Xinyuan Fang School of Artificial Intelligence Science and Technology University of Shanghai for Science and Technology Shanghai 200093 China Institute of Photonic Chips University of Shanghai for Science and Technology Shanghai 200093 China yuetianjia@*** xinyuan.fang@***

As a brain-inspired optical computing architectures, diffractive optical neural networks (DONN) harness light’s wave nature for high-speed, energy efficient and parallel information processing, enabling applications such as image classification and wavefront shaping. However, conventional spatially encoded DONNs struggle with robustness in complex and unpredictable environments, where occlusions and distortions degrade processing accuracy. To address these challenges, we propose a robust all-optical feature extraction framework based on orbital angular momentum (OAM). This approach converts optical information into target OAM modes using a diffractive processing framework trained via deep learning, enabling stable and efficient information representation in the OAM domain. Unlike conventional DONNs, our method maintains high performance across diverse and irregular occlusions without requiring network retraining. This self-adaptive occlusion immune operates with zero additional training samples, effectively enhancing optical computing tasks under dynamic and uncertain conditions. By fully utilizing the helical wavefront and orthogonality of OAM, our approach improves the robustness and scalability of DONNs, demonstrating superior performance in challenging optical environments. Our work paves the way for next-generation optical computing systems that can operate reliably in unpredictable and occlusion-rich environment, unlocking what we believe to be new possibilities for robust, real-time processing in a variety of applications.

关键词： Deep learning Machine vision Mode conversion Optical computing Optical neural systems Optical systems

来源：评论

学校读者我要写书评

暂无评论

TinySpiking: a lightweight and efficient python framework for unsupervised learning spiking neural networks

ENGINEERING RESEARCH EXPRESS

引用

ENGINEERING RESEARCH EXPRESS 2025年第1期7卷

作者： Liu, Xin Mo, Lingfei Tang, Mengting Southeast Univ Sch Instrument Sci & Engn FutureX LAB Nanjing Peoples R China

neural computation frameworks are essential for advancing computational neuroscience and artificial intelligence, offering a robust platform for simulating intricate brain-like processes and fostering the growth of intelligent systems. This study presents TinySpiking, a novel, lightweight, and energy-efficient Python framework designed for the simulation and learning of Spiking neural networks (SNNs). Unlike traditional frameworks, TinySpiking does not depend on third-party libraries, thereby reducing computational overhead and energy consumption. The framework's innovation is that it implements unsupervised learning through Spike-Time Dependent Plasticity (STDP), a biologically inspired learning rule that adjusts synaptic weights based on the precise timing of neuronal spikes. This mechanism is crucial for constructing complex neural architectures and effectively processing spatio-temporal data, which is vital for the dynamic and intricate demands of real-world data analysis. Demonstrating the practical utility of TinySpiking, we applied it to image reconstruction tasks using the MNIST and landmark datasets. The results not only validate the network's ability to autonomously identify and enhance key visual features without external supervision but also highlight its efficiency in learning from data, mirroring the adaptability of biological neural systems. In conclusion, TinySpiking's innovative, lightweight design and its proven effectiveness in unsupervised learning tasks make it a standout computational tool for the fields of computational neuroscience and machine learning. Its low time consumption, biological plausibility, and independence from third-party libraries position it as a compelling platform for future research and applications, promising to drive advancements in unsupervised learning and intelligent system development.

关键词： spiking neural networks STDP unsupervised learning image reconstruction computational neuroscience

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：