检索结果-内蒙古大学图书馆

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

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

Application of Classification Framework Based on CDR and CNN in Ophthalmic Prediagnosis 1st

Application of Classification Framework Based on CDR and CNN...

引用

1st International Conference on pattern Analysis and Machine Intelligence, ICPAMI 2024

作者： Xu, Yunjiao Yang, Jie Guo, Tianjiao Scientific Research Office Shanghai Normal University Tianhua College Shanghai China Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University Shanghai China

ISBN: (纸本)9789819633487

The seven-field classification plays a key role in ophthalmology, especially in the diagnosis and treatment of diabetic retinopathy. In this paper, we designed a framework that can classify 7 fields of each eye simultaneously, and applied it to the medical-assisted interactive system of 7 fields of classification。F1 and F2 field of view are detected through a combination of algorithms based on decision rules and convolutional neural networks. The total accuracy on the internal test set is 98.7%(95%CI, 98.4%–99.1%). In external tests, the total accuracy of the algorithm in this paper on the LONGITUDINAL data set is 98.6%(95%CI,97.9%–99.3%). The total accuracy on the Drishti-GS data set is 100%. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Family Doctor Model Training Based on Large Language Model Tuning 1st

Family Doctor Model Training Based on Large Language Model T...

引用

1st International Conference on pattern Analysis and Machine Intelligence, ICPAMI 2024

作者： Zhu, Yu Li, Wangtao Ling, Xufeng Yang, Jie Shanghai China School of AI Shanghai Normal University Tianhua College Shanghai China Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University Shanghai China

ISBN: (纸本)9789819633487

As society increasingly focuses on health issues, the importance of the role of family doctors becomes more apparent. The question of how to use cutting-edge technology to improve the accessibility and efficiency of medical services is a critical topic. Generative large language models, with their excellent capabilities in natural language processing, offer new possibilities for the implementation of intelligent family doctor systems. This study focuses on the application of generative large language models in family doctor systems, selecting the ChatGLM2-6B model as the base model for this research. Through a custom-written data collection program, a dataset containing 170,000 high-quality medical question-and-answer entries was constructed. The LoRA method was used to tune the base model, and the performance of the original and tuned models in medical question-and-answer tasks was examined. The practical results of this study show that the generatively tuned large language model has achieved a 70% improvement in ROUGE scores compared to the untuned model, validating the effectiveness of large language model tuning in the family doctor field. Furthermore, this study has open-sourced the collected dataset and program, providing a technical reference for deploying efficient and accurate question-and-answer systems in similar vertical fields in the future. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Natural language processing systems

来源：评论

学校读者我要写书评

暂无评论

Reflecting topology consistency and abnormality via learnable attentions for airway labeling

引用

International Journal of Computer Assisted Radiology and Surgery 2025年 1-9页

作者： Li, Chenyu Zhang, Minghui Zhang, Chuyan Gu, Yun Institute of Medical Robotics Shanghai Jiao Tong University Shanghai China Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University Shanghai China Shanghai Key Laboratory of Flexible Medical Robotics Tongren Hospital Shanghai Jiao Tong University Shanghai China

Purpose: Accurate airway anatomical labeling is crucial for clinicians to identify and navigate complex bronchial structures during bronchoscopy. Automatic airway labeling is challenging due to significant anatomical variations. Previous methods are prone to generate inconsistent predictions, hindering preoperative planning and intraoperative navigation. This paper aims to enhance topological consistency and improve the detection of abnormal airway branches. Methods: We propose a transformer-based framework incorporating two modules: the soft subtree consistency (SSC) and the abnormal branch saliency (ABS). The SSC module constructs a soft subtree to capture clinically relevant topological relationships, allowing for flexible feature aggregation within and across subtrees. The ABS module facilitates interaction between node features and prototypes to distinguish abnormal branches, preventing the erroneous features aggregation between normal and abnormal nodes. Results: Evaluated on a challenging dataset characterized by severe airway deformities, our method achieves superior performance compared to state-of-the-art approaches. Specifically, it attains an 83.7% subsegmental accuracy, along with a 3.1% increase in segmental subtree consistency, a 45.2% increase in abnormal branch recall. Notably, the method demonstrates robust performance in cases with airway deformities, ensuring consistent and accurate labeling. Conclusion: The enhanced topological consistency and robust identification of abnormal branches provided by our method offer an accurate and robust solution for airway labeling, with potential to improve the precision and safety of bronchoscopy procedures. © CARS 2025.

关键词： Airway anatomical labeling Anomaly detection Structural prior 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

DeepRAG: Thinking to Retrieval Step by Step for Large Language Models

arXiv

引用

arXiv 2025年

作者： Guan, Xinyan Zeng, Jiali Meng, Fandong Xin, Chunlei Lu, Yaojie Lin, Hongyu Han, Xianpei Sun, Le Zhou, Jie Chinese Information Processing Laboratory Institute of Software Chinese Academy of Sciences China University of Chinese Academy of Sciences China Pattern Recognition Center WeChat AI Tencent Inc China

Large Language Models (LLMs) have shown remarkable potential in reasoning while they still suffer from severe factual hallucinations due to timeliness, accuracy, and coverage of parametric knowledge. Meanwhile, integrating reasoning with retrieval-augmented generation (RAG) remains challenging due to ineffective task decomposition and redundant retrieval, which can introduce noise and degrade response quality. In this paper, we propose DeepRAG, a framework that models retrieval-augmented reasoning as a Markov Decision Process (MDP), enabling strategic and adaptive retrieval. By iteratively decomposing queries, DeepRAG dynamically determines whether to retrieve external knowledge or rely on parametric reasoning at each step. Experiments show that DeepRAG improves retrieval efficiency while improving answer accuracy by 21.99%, demonstrating its effectiveness in optimizing retrieval-augmented reasoning. © 2025, CC BY.

关键词：

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：

来源：评论

学校读者我要写书评

暂无评论

CATransUnetLBP: Accurate Prediction of Protein-Ligand Binding Pockets Using a Hybrid Network

IEEE Transactions on Computational Biology and Bioinformatic...

引用

IEEE Transactions on Computational Biology and Bioinformatics 2025年第1期22卷 355-367页

作者： Cheng Cai Zhaohong Deng Andong Li Yun Zuo Haoran Chen Zhisheng Wei Lei Wang Xiaoyong Pan Hong-Bin Shen Dong-Jun Yu School of Artificial Intelligence and Computer Science Jiangnan University Wuxi China Shanghai Key Lab of Intell. Info. Processing School of CS Fudan University Shanghai China School of Biotechnology Jiangnan University Wuxi China Institute of Image Processing and Pattern Recognition Shanghai Jiao Tong University Shanghai China School of Computer Science and Engineering Nanjing University of Science and Technology Nanjing China

The development of intelligent methods capable of predicting protein-ligand binding sites has become a popular research field. Recently, deep learning based methods have been proposed as a promising solution for this task. However, some limitations still exist. For example, the network structure is not optimized for predicting protein binding pockets, which limits the model's capabilities. To address the aforementioned challenges, a novel method called CATransUnetLPB is proposed, in which a new network structure named CATransUnet is designed. The proposed CATransUnet combines CNN and Transformer models to accurately segment binding pocket regions from protein 3D structures. It outperforms existing representative methods on three test sets, demonstrating the effectiveness of optimizing the deep network model for detecting protein ligand binding pockets. Furthermore, we conduct thorough analysis on applying data augmentation to protein data structure and confirm that such technique can enhance the model's generalization ability, thereby ensuring good performance on new protein structures. Moreover, experiments show that the predicted binding pockets from our model can complement the results obtained from other methods. This suggests that integrating our method with existing approaches could further improve the prediction of protein-ligand binding pockets.

关键词： Proteins Three-dimensional displays Probes Vectors image segmentation Deep learning Atoms Surface treatment Prediction algorithms Training

来源：评论

学校读者我要写书评

暂无评论

Contour detection network simulating the primary visual pathway

引用

Digital Signal processing 2025年 165卷

作者： Ke Chen Yingle Fan Tao Fang Laboratory of Pattern Recognition and Image Processing Hangzhou Dianzi University Hangzhou 310018 Zhejiang China

Biological vision exhibits exceptional contour perception capabilities. In view of this, research on contour detection guided by biological vision is gradually gaining attention. Inspired by the transmission and processing of visual signals in the primary visual pathway, this study proposes a lightweight contour detection network called the Primary-Visual-Pathway UNet (PVP-UNet), comprising an encoder and a decoder. Primarily, inspiration from the binocular vision mechanism, we transmit the original image through dual deformable convolution modules that simulate the receptive fields of the left and right retinas in the encoder. Utilizing the characteristic of the optic chiasm, the output features of the retinal layers are split, swapped, and merged for further processing. Subsequently, dilated convolution modules and normal convolution modules are involved to simulate the magnocellular (M) and parvocellular (P) pathways of the lateral geniculate nucleus (LGN), respectively. An inhibition module was designed based on the suppression mechanism of classical/non-classical receptive fields (CRF/NCRF) in the primary visual cortex (V1). Additionally, the interconnection pattern of the inhibition modules is deployed by leveraging the aggregation characteristics of simple cells to complex cells in the V1 layer. Inspired by the feedback mechanism of visual information, the feature fusion module is introduced in the decoder to integrate features from different encoding layers in the reverse direction of signal transmission along the primary visual pathway. Our experiments show that the respective Optimal Dataset Scale (ODS) scores achieve 0.811, 0.756, and 0.896 on BSDS500, NYUD, and BIPED datasets. The experimental results demonstrate that the proposed network effectively suppresses background noise and highlights primary contours, exhibiting excellent detection performance on the tested datasets. The code is available at https://***/k3chencoco/PVP-UNet .

关键词：

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

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：