检索结果-内蒙古大学图书馆

DFL: cross-view cross-layer discriminative feature learning for fine-grained 3D shape classification

Neural Computing and Applications 2025年 1-22页

作者： Jiang, Jinzhe Bai, Jing Ma, Xiangyu The School of Computer Science and Engineering North Minzu University Yinchuan China The Key Laboratory of Images Processing and Pattern Recognition Laboratory North Minzu University Yinchuan China

Fine-grained 3D shape classification poses challenges in effectively capturing and integrating discriminative features residing in subtle local regions. Previous methods typically extract features independently from individual views of 3D shapes, with a focus on various strategies for fusing these extracted view features. However, this approach neglects interview correlations and potential redundancies among different views. In this study, we introduce $$\hbox {C}^2$$ DFL, which consists of two primary modules: cross-view discriminative feature extraction (CV-DFE) and cross-layer discriminative feature fusion (CL-DFF). CV-DFE integrates discriminative features by merging inputs from multiple views, mitigating limitations associated with isolated feature extraction. CL-DFF dynamically selects key tokens using a transformer model to interactively fuse discriminative features from various levels. Extensive experiments conducted on three categories of the FG3D dataset demonstrate the exceptional efficacy of $$\hbox {C}^2$$ DFL in capturing and integrating discriminative features of 3D shapes. The proposed method achieves state-of-the-art accuracy in fine-grained 3D shape classification (FGSC).

关键词：

来源：评论

学校读者我要写书评

暂无评论

FG3DFormer: Fine-Grained 3D Shape Classification Based on Vision Transformer

FG3DFormer: Fine-Grained 3D Shape Classification Based on Vi...

引用

International Conference on Acoustics, Speech, and Signal processing (ICASSP)

作者： Xiangyu Ma Jing Bai Jinzhe Jiang Bin Peng The School of Computer Science and Engineering North Minzu University The Key Laboratory of Images Processing and Pattern Recognition Laboratory Yinchuan China

ISBN: (数字)9798350368741

ISBN: (纸本)9798350368758

Fine-grained 3D shape classification (FGSC) remains challenging due to the difficulty of adaptively capturing global structure differences and subtle inter-class distinctions. This paper directly extends Vision Transformer (ViT) to FGSC, proposing a pure Transformer network FG3DFormer that fully leverages ViT’s global correlation and local attention abilities. FG3Dformer comprises the Hierarchical Feature Extraction (HFE) and the Hierarchical Feature Refinement (HFR), interconnected through the Adaptive View Region Selection (AVRS). Firstly, the HFE comprehensively evaluates the significance of intra-view patches and views driven by inter-view and intraview attention. Then, the AVRS adaptively selects crucial patch Tokens from different views to serve as sources of subtle local features. Finally, the HFR refines the 3D shape descriptor, capturing more discriminative global and subtle local features by leveraging both the view and selected crucial patch Tokens. Extensive experiments on FG3D and ModelNet40 demonstrate the superiority of FG3Dformer in FGSC and meta-category 3D shape classification tasks.

关键词： Computer vision Visualization Solid modeling Three-dimensional displays Correlation Shape Signal processing Transformers Feature extraction Speech processing

来源：评论

学校读者我要写书评

暂无评论

Visual Prompt Flexible-Modal Face Anti-Spoofing

引用

IEEE Transactions on Dependable and Secure Computing 2025年第3期22卷 2597-2606页

作者： Yu, Zitong Cai, Rizhao Cui, Yawen Liu, Ajian Chen, Changsheng Great Bay University School of Computing and Information Technology Dongguan523000 China Nanyang Technological University ROSE Lab School of EEE 639798 Singapore Hong Kong Polytechnic University Kowloon Hong Kong Chinese Academy of Sciences University of Chinese Academy of Sciences National Laboratory of Pattern Recognition Institute of Automation Beijing100190 China Shenzhen University Guangdong Key Laboratory of Intelligent Information Processing Shenzhen Key Laboratory of Media Security College of Electronics and Information Engineering Shenzhen518060 China

Recently, vision transformer based multimodal learning methods have been proposed to improve the robustness of face anti-spoofing (FAS) systems. However, multimodal face data collected from the real world is often imperfect due to missing modalities from various imaging sensors. Recently, flexible-modal FAS (Yu et al. 2023) has attracted more attention, which aims to develop a unified multimodal FAS model using complete multimodal face data but is insensitive to test-time missing modalities. In this paper, we tackle one main challenge in flexible-modal FAS, i.e., when missing modality occurs either during training or testing in real-world situations. Inspired by the recent success of the prompt learning in language models, we propose Visual Prompt flexible-modal FAS (VP-FAS), which learns the modal-relevant prompts to adapt the frozen pre-trained foundation model to downstream flexible-modal FAS task. Specifically, both vanilla visual prompts and residual contextual prompts are plugged into multimodal transformers to handle general missing-modality cases, while only requiring less than 4% learnable parameters compared to training the entire model. Furthermore, missing-modality regularization is proposed to force models to learn consistent multimodal feature embeddings when missing partial modalities. Extensive experiments conducted on two multimodal FAS benchmark datasets demonstrate the effectiveness of our VP-FAS framework that improves the performance under various missing-modality cases while alleviating the requirement of heavy model re-training. © 2004-2012 IEEE.

关键词： Alignment

来源：评论

学校读者我要写书评

暂无评论

Discovering the nuclear localization signal universe through a deep learning model with interpretable attention units

Patterns

引用

patterns 2025年

作者： Li, Yi-Fan Pan, Xiaoyong Shen, Hong-Bin Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University and Key Laboratory of System Control and Information Processing Ministry of Education of China Shanghai200240 China

We describe NLSExplorer, an interpretable approach for nuclear localization signal (NLS) prediction. By utilizing the extracted information on nuclear-specific sites from the protein language model to assist in NLS detection, NLSExplorer achieves superior performance with greater than 10% improvement in the F1 score compared with existing methods on benchmark datasets and highlights other nuclear transport segments. We applied NLSExplorer to the nucleus-localized proteins in the Swiss-Prot database to extract valuable segments. A comprehensive analysis of these segments revealed a potential NLS landscape and uncovered features of nuclear transport segments across 416 species. This study introduces a powerful tool for exploring the NLS universe and provides a versatile network that can efficiently detect characteristic domains and motifs. © 2025 The Author(s)

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

OCR4HSV: A Multi-task Learning Approach for Handwritten Signature Verification 27th

OCR4HSV: A Multi-task Learning Approach for Handwritten Sig...

引用

27th International Conference on pattern recognition, ICPR 2024

作者： Lin, Chao-Qun Wang, Da-Han Su, Yan-Fei Ge, De-Wu Zhang, Xu-Yao School of Computer and Information Engineering Xiamen University of Technology Xiamen361024 China Fujian Key Laboratory of Pattern Recognition and Image Understanding Xiamen361024 China Xiamen KEYTOP Communication Technology Co. Xiamen361024 China State Key Laboratory of Multimodal Artificial Intelligence Systems Institute of Automation of Chinese Academy of Sciences Beijing100190 China

ISBN: (纸本)9783031781186

Handwritten signature verification (HSV) models are notably recognized for their ability to discern whether a signature is forged in an offline document. Recently, HSV technology has made significant development. However, these methods have primarily focused on the shape features of the text image and overlooked the textual feature information inherent in the text itself, which makes the HSV model overfit. In this paper, we propose a novel network model named OCR4HSV to solve the above shortcomings. The proposed OCR4HSV first attempts to combine OCR and HSV to learn textual features in a multi-task learning manner. The model employs a dual-parameter approach, combining shared parameters and independent parameters. Specifically, Within the shared parameters, the Laplace attention module (LAM) is incorporated for edge information extraction. For independent parameters, CN-Mamba is utilized for sequence feature extraction in OCR, and the multi-scale global fusion block (MGFB) is designed to enhance the distinction between reference and test sample pairs. Leveraging OCR-related information and these architectural enhancements can fully mine the inherent textual feature information and significantly mitigate overfitting in the HSV task, thereby boosting verification accuracy. Our model has achieved state-of-the-art performance on the ChiSig and HanSig datasets. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.

关键词： Multi-task learning

来源：评论

学校读者我要写书评

暂无评论

C2DFL: cross-view cross-layer discriminative feature learning for fine-grained 3D shape classification

引用

Neural Computing and Applications 2025年

作者： Jiang, Jinzhe Bai, Jing Ma, Xiangyu The School of Computer Science and Engineering North Minzu University Yinchuan750021 China The Key Laboratory of Images Processing and Pattern Recognition Laboratory North Minzu University Yinchuan750021 China

Fine-grained 3D shape classification poses challenges in effectively capturing and integrating discriminative features residing in subtle local regions. Previous methods typically extract features independently from individual views of 3D shapes, with a focus on various strategies for fusing these extracted view features. However, this approach neglects interview correlations and potential redundancies among different views. In this study, we introduce C2DFL, which consists of two primary modules: cross-view discriminative feature extraction (CV-DFE) and cross-layer discriminative feature fusion (CL-DFF). CV-DFE integrates discriminative features by merging inputs from multiple views, mitigating limitations associated with isolated feature extraction. CL-DFF dynamically selects key tokens using a transformer model to interactively fuse discriminative features from various levels. Extensive experiments conducted on three categories of the FG3D dataset demonstrate the exceptional efficacy of C2DFL in capturing and integrating discriminative features of 3D shapes. The proposed method achieves state-of-the-art accuracy in fine-grained 3D shape classification (FGSC). © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2025.

关键词： 3D shape classification Cross-layer feature fusion Cross-view attention Fine-grained classification Token selection Transformer

来源：评论

学校读者我要写书评

暂无评论

DLS-HCAN: Duplex Label Smoothing Based Hierarchical Context-Aware Network for Fine-grained 3D Shape Classification

引用

IEEE Transactions on Multimedia 2025年

作者： Bai, Shaojin Zheng, Liang Bai, Jing Ma, Xiangyu North Minzu University School of Computer Science and Engineering Yinchuan750021 China Liupanshan Laboratory Yinchuan750021 China North Minzu University Key Laboratory of Images Processing and Pattern Recognition LaboratoryCommission: IPPRLab Yinchuan750021 China

Fine-grained 3D shape classification (FGSC) has garnered significant attention recently and has made notable advancements. However, due to high inter-class similarity and intra-class diversity, it is still a challenge for existing methods to capture subtle differences between different subcategories for FGSC. On the one hand, one-hot labels in loss function are too hard to describe the above data characteristics, and on the other hand, local details are submerged in the global features extraction process and final network constraints, impacting classification results. In this paper, we propose a duplex label smoothingbased hierarchical context-aware network for fine-grained 3D shape classification, named DLS-HCAN. Specifically, DLS-HCAN firstly employs a hierarchical context-aware network (HCAN), in which the intra-view context attention mechanism (intra-ATT) and the inter-view context multilayer perceptron (inter-MLP) are designed to focus on and discern the beneficial local details. Subsequently, we propose a novel duplex label smoothing (DLS) regularization in which shape-level and view-level smooth labels are separately applied in two improved loss functions, adapting to the fine-grained data characteristics and considering the varying uniqueness of different views. Notably, our approach does not require additional annotation information. Experimental results and comparison with state-of-the-art methods demonstrate the superiority of our proposed DLS-HCAN for FGSC. In addition, our approach also achieves comparable performance for the coarse-grained dataset on ModelNet40. © 2025 IEEE.

关键词： Coarse-grained modeling

来源：评论

学校读者我要写书评

暂无评论

TNIE-SLAM: Neural Implicit Surface Reconstruction for Tracking-Oriented SLAM

引用

Journal of Intelligent & Robotic Systems 2025年第2期111卷 1-22页

作者： Gan, Baolin Zhang, Congxuan Chen, Shuaixin Chen, Zhen He, Chao Lu, Ke Lu, Feng School of Instrument Science and Optoelectronic Engineering Nanchang Hangkong University Nanchang China Jiangxi Provincial Key Laboratory of Image Processing and Pattern Recognition Nanchang Hangkong University Nanchang China School of Information Engineering Nanchang Hangkong University Nanchang China The College of Engineering Science University of Chinese Academy of Sciences Beijing China

Recent studies on simultaneous localization and mapping (SLAM) have tended to employ implicit neural representation, which can improve the efficiency and robustness of SLAM system. However, these methodologies still face challenges, such as tracking failures and low-precision mapping. In this paper, we propose a dense reconstruction visual SLAM system enhanced with closed-loop threading and local map optimization, named TNIE-SLAM. First, we propose a tracking module that utilizes the similarity of ORB feature descriptors and the feature overlap rate of the current frame to model key frames, and then we define a complete and accurate initial map based on full bundle adjustment, which addresses the issue of tracking failure due to undermapped areas. Second, we add the 2D features of the initial map to the spatiotemporal encoding module to obtain the 3D features, enabling real-time prediction and tracking of unknown areas. Finally, considering the low-precision mapping issue arising from the complex geometric shapes of objects within the scene, we propose a local map optimization module that utilizes truncated signed distance fields to model 3D features and update the spatial occupancy of boundary and contour features of objects. We test our method on the synthetic Replica dataset and the real-world ScanNet and TUM RGB-D datasets to compare with some state-of-the-art RGB-D SLAM methods, and the experimental results indicate our method performs well in both tracking and mapping accuracy, surpassing the existing dense neural RGB-D SLAM methods.

关键词： Surface reconstruction

来源：评论

学校读者我要写书评

暂无评论

A Research Mode Based Evolutionary Algorithm for Many-Objective Optimization

引用

Chinese Journal of Electronics 2019年第4期28卷 764-772页

作者： CHEN Guoyu LI Junhua Key Laboratory of Jiangxi Province for Image Processing and Pattern Recognition Nanchang Hangkong University

The development of algorithms to solve Many-objective optimization problems(MaOPs) has attracted significant research interest in recent *** various types of Pareto front(PF) is a daunting challenge for evolutionary algorithm. A Research mode based evolutionary algorithm(RMEA) is proposed for many-objective optimization. The archive in the RMEA is used to store non-dominated solutions that can reflect the shape of the PF to guide the reference vector *** concerning the population is collected, once the number of non-dominated solutions reaches its limit after many generations without exceeding a given threshold, RMEA introduces a research mode that generates more reference vectors to search through the solutions. The proposed algorithm showed competitive performance with four state-of-the-art evolutionary algorithms in a large number of experiments.

关键词： Many-objective optimization Reference vector Research mode Evolutionary algorithm

来源：评论

学校读者我要写书评

暂无评论

Efficient image Super-Resolution With Feature Interaction Weighted Hybrid Network

引用

IEEE Transactions on Multimedia 2025年 27卷 2256-2267页

作者： Li, Wenjie Li, Juncheng Gao, Guangwei Deng, Weihong Yang, Jian Qi, Guo-Jun Lin, Chia-Wen Beijing University of Posts and Telecommunications Pattern Recognition and Intelligent System Laboratory School of Artificial Intelligence Beijing100080 China Shanghai University School of Communication and Information Engineering Shanghai200444 China Nanjing University of Posts and Telecommunications IVIPLab Institute of Advanced Technology Nanjing210046 China Ministry of Education Key Laboratory of Artificial Intelligence Shanghai200240 China Soochow University Provincial Key Laboratory for Computer Information Processing Technology Suzhou215006 China Nanjing University of Science and Technology School of Computer Science and Technology Nanjing210094 China Westlake University Research Center for Industries of the Future School of Engineering Hangzhou310024 China OPPO Research SeattleWA98101 United States National Tsing Hua University Department of Electrical Engineering Institute of Communications Engneering Hsinchu300044 Taiwan

Lightweight image super-resolution aims to reconstruct high-resolution images from low-resolution images using low computational costs. However, existing methods result in the loss of middle-layer features due to activation functions. To minimize the impact of intermediate feature loss on reconstruction quality, we propose a Feature Interaction Weighted Hybrid Network (FIWHN), which comprises a series of Wide-residual Distillation Interaction Block (WDIB) as the backbone. Every third WDIB forms a Feature Shuffle Weighted Group (FSWG) by applying mutual information shuffle and fusion. Moreover, to mitigate the negative effects of intermediate feature loss, we introduce Wide Residual Weighting units within WDIB. These units effectively fuse features of varying levels of detail through a Wide-residual Distillation Connection (WRDC) and a Self-Calibrating Fusion (SCF). To compensate for global feature deficiencies, we incorporate a Transformer and explore a novel architecture to combine CNN and Transformer. We show that our FIWHN achieves a favorable balance between performance and efficiency through extensive experiments on low-level and high-level tasks. © 1999-2012 IEEE.

关键词： Transformers Convolutional neural networks Computational modeling Feature extraction Training Superresolution Adaptation models Computer architecture Telecommunications image reconstruction

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：