检索结果-内蒙古大学图书馆

Deep Cross-View Reconstruction GAN Based on Correlated Subspace for Multi-View Transformation

IEEE TRANSACTIONS ON IMAGE PROCESSING 2024年 33卷 4614-4626页

作者： Mi, Jian-Xun He, Junchang Li, Weisheng Chongqing Univ Posts & Telecommun Key Lab Big Data Intelligent Comp Chongqing Key Lab Image Cognit Chongqing 400065 Peoples R China Chongqing Univ Posts & Telecommun Coll Comp Sci & Technol Chongqing 400065 Peoples R China

In scenarios where identifying face information in the visible spectrum (VIS) is challenging due to poor lighting conditions, the use of near-infrared (NIR) and thermal (TH) cameras can provide viable alternatives. However, the unique data distribution of images captured by these cameras compared to VIS images presents challenges in matching face identities. To address these challenges, we propose a novel image transformation framework. The framework includes feature extraction from the input image, followed by a transformation network that generates target domain images with perceptual fidelity. Additionally, a reconstruction network preserves original information by reconstructing the original domain image from the extracted features. By considering the correlation between features from both domains, our framework utilizes paired data obtained from the same individual. We apply this framework to two well-established image-to-image transformation models, pix2pix and CycleGAN, known as CRC-pix2pix and CRC-CycleGAN respectively. The versatility of our approach allows extension to other models based on pix2pix or CycleGAN architectures. Our models generate high-quality images while preserving the identity information of the original face. Performance evaluation on TFW and BUAA NIR-VIS datasets demonstrates the superiority of our models in terms of generated image face matching and evaluation metrics such as SSIM, MSE, PSNR, and LPIPS. Moreover, we introduce the CQUPT-VIS-TH dataset, which enriches the paired dataset with thermal-visual face data capturing various angles and expressions.

关键词： Feature extraction Image reconstruction Generative adversarial networks Face recognition data mining Correlation Image transformation Image-to-image transformation generative adversarial network multi-view learning subspace learning canonical correlation analysis

来源：评论

学校读者我要写书评

暂无评论

Dual-branch feature fusion dehazing network via multispectral channel attention

引用

INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS 2024年第7期15卷 2655-2671页

作者： Jian, Huachun Zhang, Yongjun Gao, Weihao Wang, Bufan Wang, Guomei Guizhou Univ Inst Artificial Intelligence Coll Comp Sci & Technol State Key Lab Publ Big Data Guiyang 550025 Guizhou Peoples R China

Image information degradation significantly complicates single-image dehazing. Despite the noteworthy progress of learning-based methods in removing haze, many existing algorithms focus solely on haze removal while neglecting the crucial aspect of image texture recovery, often resulting in the loss of image details and texture blurring. To tackle this issue, we propose a dual-branch feature fusion dehazing network via multi-spectral channel attention (MS-DBFFN). First, we design a multi-spectral image pre-processing module (MS-PPM), maximizing the utilization of correlation features in hazy images by introducing additional frequency components. This process results in preliminary dehazing images. Second, a dual-branch feature fusion module (DBFF) adaptively fuses features from both the hazy image and preliminary dehazing image via an iterative update strategy. During this gradual iterative updating process, the structural features of the preliminary dehazing image become richer as more feature information is assimilated. Third, our backbone network facilitates the sharing of valuable information across multiple scales from the feature map generated by the DBFF module. Finally, an improved contrastive regularization loss (ICR) generates sharper images with enhanced details and textures. Extensive experiments on both real and synthetic datasets demonstrate that the proposed method, MS-DBFFN, achieves state-of-the-art performance on several image dehazing benchmarks.

关键词： Single-image dehazing Multispectral channel attention Dual branch feature fusion Iterative update strategy

来源：评论

学校读者我要写书评

暂无评论

Prototypical Concept Representation

引用

IEEE TRANSACTIONS ON KNOWLEDGE AND data ENGINEERING 2023年第7期35卷 7357-7370页

作者： Wang, Xintao Liang, Jiaqing Xiao, Yanghua Wang, Wei Fudan Univ Sch Comp Sci Shanghai Key Lab Data Sci Shanghai 200437 Peoples R China Fudan Aishu Cognit Intelligence Joint Res Ctr Shanghai 200437 Peoples R China

Concepts are building blocks of human thinking. For machines, concept understanding has also been increasingly important, which makes concept representation a fundamental problem in artificial intelligence. While many concepts have their instances, the massive amount of information carried by instances has long been ignored in current concept representation, which limits the usage of these concepts in applications. In this paper, inspired by prototype theory in cognitive science, we propose prototypical concept representation for machines, which represents each concept with a distributed prototype derived from representations of its instances. For prototypical representation learning, we further introduce a novel model named Prototypical Siamese Network (PSN). PSN is trained under the supervision of isA determination, one of the most important concept-related applications. Results of extensive experiments demonstrate that, our method achieves state-of-the-art performance, thus validating the effectiveness of prototypical concept representation.

关键词： Concept learning distributed representations knowledge graphs machine learning

来源：评论

学校读者我要写书评

暂无评论

A multi-graph representation for event extraction

引用

ARTIFICIAL intelligence 2024年 332卷

作者： Huang, Hui Chen, Yanping Lin, Chuan Huang, Ruizhang Zheng, Qinghua Qin, Yongbin Guizhou Univ Text Comp & Cognit Intelligence Engn Res Ctr State Key Lab Publ Big Data Natl Educ MinistColl Comp Sci & Technol Guiyang 550025 Peoples R China Xi An Jiao Tong Univ Dept Comp Sci & Technol Xian 710049 Peoples R China

Event extraction has a trend in identifying event triggers and arguments in a unified framework, which has the advantage of avoiding the cascading failure in pipeline methods. The main problem is that joint models usually assume a one -to-one relationship between event triggers and arguments. It leads to the argument multiplexing problem, in which an argument mention can serve different roles in an event or shared by different events. To address this problem, we propose a multigraph-based event extraction framework. It allows parallel edges between any nodes, which is effective to represent semantic structures of an event. The framework enables the neural network to map a sentence(s) into a structurized semantic representation, which encodes multi-overlapped events. After evaluated on four public datasets, our method achieves the stateof-the-art performance, outperforming all compared models. Analytical experiments show that the multigraph representation is effective to address the argument multiplexing problem and helpful to advance the discriminability of the neural network for event extraction.

关键词： Event extraction Multigraph Argument multiplexing Event representation

来源：评论

学校读者我要写书评

暂无评论

Multi-granularity interaction and feature recombination network for fine-grained visual classification

引用

PATTERN REcognitION 2025年 166卷

作者： Ke, Xiao Cai, Yuhang Chen, Baitao Liu, Hao Guo, Wenzhong Fuzhou Univ Coll Comp & Data Sci Fujian Key Lab Network Comp & Intelligent Informat Fuzhou 350116 Peoples R China Fuzhou Univ Engn Res Ctr Big Data Intelligence Minist Educ Fuzhou 350116 Peoples R China

Fine-grained visual classification (FGVC) is a highly challenging task that aims to learn subtle differences between visually similar objects. Most existing methods for FGVC rely on deep convolutional neural networks to mine local fine-grained features, which neglect the learning of relationships between global and local semantics. Moreover, the feature encoding stage inevitably constructs complex feature representations, leading to overfitting to specific feature patterns, which is not beneficial for fine-grained visual classification. To address these issues, we propose a Transformer-based FGVC model, called the Multi-Granularity Interaction and Feature Recombination Network(MGIFR-Net), which consists of three modules. Firstly, a self-attention guided localization module is designed to locate and amplify discriminative local regions, enabling the sufficient learning of local detail information. Secondly, to enhance the perception of multi-granularity semantic interaction information, we construct a multi-granularity feature interaction learning module to jointly learn local and global feature representations. Finally, a dynamic feature recombination enhancement method is proposed, which explores diverse feature pattern combinations while retaining invariant features, effectively alleviating the overfitting problem caused by complex feature representations. Our method achieves stateof-the-art performance on four benchmark FGVC datasets (CUB-200-2011, Stanford Cars, FGVC-Aircraft, and NAbirds), and experimental results demonstrate the superiority of our method on different visual classification benchmarks.

关键词： Fine-grained visual classification Multi-granularity feature interaction Feature recombination Vision transformer

来源：评论

学校读者我要写书评

暂无评论

Progressive fusion of local and global image features for cross-modal image aesthetic assessment

引用

MULTIMEDIA SYSTEMS 2025年第2期31卷 1-16页

作者： Niu, Yuzhen Chen, Siling Chen, Shanshan Li, Fusheng Fuzhou Univ Coll Comp & Data Sci Fujian Key Lab Network Comp & Intelligent Informat Fuzhou 350108 Fujian Peoples R China Minist Educ Engn Res Ctr Big Data Intelligence Fuzhou 350108 Fujian Peoples R China

Image aesthetic assessment (IAA) has drawn wide attention in recent years. This task aims to predict the aesthetic quality of images by simulating human aesthetic perception mechanism, thereby assisting users in selecting images with higher aesthetic value. For IAA, the local information and various global semantic information contained in an image, such as composition, theme, and emotion, all play a crucial role. Existing CNN-based methods attempt to use multi-branch strategies to extract local and global semantic information related to IAA from images. However, these methods can only extract limited and specific global semantic information, and requiring additional labeled datasets. Furthermore, some cross-modal IAA methods have been proposed to use both images and user comments, but they often fail to fully explore the valuable information within each modality and the correlations between cross-modal features, affecting cross-modal IAA accuracy. Considering these limitations, in this paper, we propose a cross-modal IAA model that progressively fuses local and global image features. The model consists of a progressive local and global image feature fusion branch, a text feature enhancement branch, and a cross-modal feature fusion module. In the image branch, we introduce an inter-layer feature fusion module (IFFM) and adopt a progressive way to interact and fuse the extracted local and global features to obtain more comprehensive image features. In the text branch, we propose a text feature enhancement module (TFEM) to strengthen the extracted text features, so as to mine more effective textual information. Meanwhile, considering the intrinsic correlation between image and text features, we propose a cross-modal feature fusion module (CFFM) to integrate and fuse image features with text features for aesthetic assessment. Experimental results on the AVA (Aesthetic Visual Analysis) dataset validate the superiority of our method for IAA task.

关键词： Image aesthetic assessment Local and global features Feature fusion Textual information Cross-modality

来源：评论

学校读者我要写书评

暂无评论

Adaptive feature extraction for entity relation extraction

引用

compUTER SPEECH AND LANGUAGE 2025年 89卷

作者： Yang, Weizhe Qin, Yongbin Huang, Ruizhang Chen, Yanping Guizhou Univ State Key Lab Publ Big Data Guiyang 550025 Peoples R China Guizhou Univ Engn Res Ctr Text Comp & Cognit Intelligence Minist Educ Guiyang 550025 Peoples R China Guizhou Univ Coll Comp Sci & Technol Guiyang 550025 Guizhou Peoples R China

Effective capturing of semantic dependencies within sentences is pivotal to support relation extraction. However, challenges such as feature sparsity, and the complexity of identifying the structure of target entity pairs brought by the traditional methods of feature extraction pose significant obstacles for relation extraction. Existing methods that rely on combined features or recurrent networks also face limitations, such as over-reliance on prior knowledge or the gradient vanishing problem. To address these limitations, we propose an Adaptive Feature Extraction (AFE) method, combining neural networks with feature engineering to capture high-order abstract and long-distance semantic dependencies. Our approach extracts atomic features from sentences, maps them into distributed representations, and categorizes these representations into multiple mixed features through adaptive combination, setting it apart from other methods. The proposed AFE-based model uses four different convolutional layers to facilitate feature learning and weighting from the adaptive feature representations, thereby enhancing the discriminative power of deep networks for relation extraction. Experimental results on the English datasets ACE05 English, SciERC, and the Chinese datasets ACE05 Chinese, and CLTC(SanWen) demonstrated the superiority of our method, the F1 scores were improved by 4.16%, 3.99%, 0.82%, and 1.60%, respectively. In summary, our AFE method provides a flexible, and effective solution to some challenges in cross-domain and cross-language relation extraction.

关键词： Relation extraction Adaptive feature extraction Semantic dependencies Entity-related features Deep learning

来源：评论

学校读者我要写书评

暂无评论

Real-time reliable semantic segmentation of thyroid nodules in ultrasound images

引用

PHYSICS IN MEDICINE AND BIOLOGY 2024年第2期69卷 025016-025016页

作者： Xing, Guangxin Wang, Shuaijie Gao, Jie Li, Xuewei Tianjin Univ State Key Lab Precis Measuring Technol & Instrumen Tianjin Peoples R China Tianjin Univ Coll Intelligence & Comp Tianjin Peoples R China Tianjin Key Lab Cognit Comp & Applicat Tianjin Peoples R China Tianjin Key Lab Adv Networking Tianjin Peoples R China

Objective. Low efficiency in medical image segmentation is a common issue that limits computer-aided diagnosis development. Due to the varying positions and sizes of nodules, it is not easy to accurately segment ultrasound images. This study aims to propose a segmentation model that maintains high efficiency while improving accuracy. Approach. We propose a novel layer that integrates the advantages of dense connectivity, dilated convolution, and factorized filters to maintain excellent efficiency while improving accuracy. Dense connectivity optimizes feature reuse, dilated convolution redesigns layers, and factorized convolution improves efficiency. Moreover, we propose a loss function optimization method from a pixel perspective to increase the network's accuracy further. Main results. Experiments on the Thyroid dataset show that our method achieves 81.70% intersection-over-union (IoU), 90.50% true positive rate (TPR), and 0.25% false positive rate (FPR). In terms of accuracy, our method outperforms the state-of-the-art methods, with twice faster inference and nearly 400 times fewer parameters. Meanwhile, in a test on an External Thyroid dataset, our method achieves 77.03% IoU, 82.10% TPR, and 0.16% FPR, demonstrating our proposed model's robustness. Significance. We propose a real-time semantic segmentation architecture for thyroid nodule segmentation in ultrasound images called fully convolution dense dilated network (FCDDN). Our method runs fast with a few parameters and is suitable for medical devices requiring real-time segmentation.

关键词： computer-aided diagnosis ultrasound images thyroid nodules full convolution dense dilated network real-time segmentation

来源：评论

学校读者我要写书评

暂无评论

Deep clustering via dual-supervised multi-kernel mapping

引用

PATTERN REcognitION 2025年 162卷

作者： Ren, Lina Huang, Ruizhang Ma, Shengwei Qin, Yongbin Chen, Yanping Lin, Chuan Guizhou Light Ind Tech Coll Dept Informat Engn Guiyang 550025 Peoples R China Guizhou Univ Coll Comp Sci & Technol Text Comp & Cognit Intelligence Engn Res Ctr Natl Educ MinistState Key Lab Publ Big Data Guiyang 550025 Peoples R China

In this paper, we propose a novel deep clustering framework via dual-supervised multi-kernel mapping, namely DCDMK, to improve clustering performance by learning linearly structural separable data representations. In the DCDMK framework, we introduce a kernel-aid encoder comprising two key components: a semantic representation learner, which captures the essential semantic information for clustering, and a multi-kernel representation learner, which dynamically selects the optimal combination of kernel functions through dual- supervised multi-kernel mapping to learn structurally separable kernel representations. The dual self-supervised mechanism is devised to jointly optimize both kernel representation learning and structural partitioning. Based on this framework, we introduce different fusion strategies to learn the multi-kernel representation of data samples for the clustering task. We derive two variants, namely DCDMK-WL (with layer-level kernel representation learning) and DCDMK-OL (without layer-level kernel representation learning). Extensive experiments on six real-world datasets demonstrate the effectiveness of our DCDMK framework.

关键词： Deep clustering Multiple kernel Kernel representation Semantic representation

来源：评论

学校读者我要写书评

暂无评论

Glioma segmentation based on dense contrastive learning and multimodal features recalibration

引用

PHYSICS IN MEDICINE AND BIOLOGY 2024年第9期69卷 095016-095016页

作者： Hu, Xubin Wang, Lihui Wang, Li Chen, Qijian Zheng, Licheng Zhu, Yuemin Guizhou Univ Coll Comp Sci & Technol Engn Res Ctr Text Comp & Cognit Intelligence State Key Lab Publ Big DataKey Lab Intelligent Me Guiyang 550025 Peoples R China Univ Lyon INSA Lyon CNRS InsermIRP Metislab CREATIS UMR5220U1206 F-69621 Lyon France

Accurate segmentation of different regions of gliomas from multimodal magnetic resonance (MR) images is crucial for glioma grading and precise diagnosis, but many existing segmentation methods are difficult to effectively utilize multimodal MR image information to recognize accurately the lesion regions with small size, low contrast and irregular shape. To address this issue, this work proposes a novel 3D glioma segmentation model DCL-MANet. DCL-MANet has an architecture of multiple encoders and one single decoder. Each encoder is used to extract MR image features of a given modality. To overcome the entangle problems of multimodal semantic features, a dense contrastive learning (DCL) strategy is presented to extract the modality-specific and common features. Following that, feature recalibration block (RFB) based on modality-wise attention is used to recalibrate the semantic features of each modality, enabling the model to focus on the features that are beneficial for glioma segmentation. These recalibrated features are input into the decoder to obtain the segmentation results. To verify the superiority of the proposed method, we compare it with several state-of-the-art (SOTA) methods in terms of Dice, average symmetric surface distance (ASSD), HD95 and volumetric similarity (Vs). The comparison results show that the average Dice, ASSD, HD95 and Vs of DCL-MANet on all tumor regions are improved at least by 0.66%, 3.47%, 8.94% and 1.07% respectively. For small enhance tumor (ET) region, the corresponding improvement can be up to 0.37%, 7.83%, 11.32%, and 1.35%, respectively. In addition, the ablation results demonstrate the effectiveness of the proposed DCL and RFB, and combining them can significantly increase Dice (1.59%) and Vs (1.54%) while decreasing ASSD (40.51%) and HD95 (45.16%) on ET region. The proposed DCL-MANet could disentangle multimodal features and enhance the semantics of modality-dependent features, providing a potential means to accurately segment

关键词： gliomas feature disentangle modality recalibration small region segmentation

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：