检索结果-内蒙古大学图书馆

intelligent SYSTEMS WITH APPLICATIONS 2025年 25卷

作者： Gao, Kai Zhou, Jiahao Chi, Yunxian Wen, Yimin Hebei Univ Sci & Technol Sch Informat Sci & Engn Shijiazhuang 050018 Peoples R China Guilin Tourism Univ Guangxi Key Lab Culture & Tourism Smart Technol Guilin 541006 Peoples R China Guilin Univ Elect Technol Guangxi Key Lab Image & Graph Intelligent Proc Guilin 541004 Peoples R China

Tourism named entity recognition is indispensable in tourism information extraction, and plays a crucial role in constructing tourism knowledge map and enhancing tourism knowledge quiz system. The difficulty of tourism named entity recognition lies in its complex nested structure, and the lengthy entity naming length. To address these existing problems, we propose a tourism named entity recognition model that jointly predicts entity boundaries, adopting a training strategy of data preprocessing to enhance the model's ability for tourism named entity boundary recognition, while our model introduces a pre-trained Bert model as well as BiLSTM coding to enhance the representation of the model's contexts, and uses a combined predictor of Biaffine and MLP to enhance the model's recognition performance for boundaries, as well as introducing label smoothing cross entropy to smooth the target labels during the training process. Experiments are conducted on three datasets with different granularities. From the analysis of the experimental results, it can be seen that the named entity recognition method achieves higher accuracy and F1 value compared with the optimal baseline model, and also proves the effectiveness and generality of the modeling method proposed in this paper.

关键词： Natural Language processing Tourism Named Entity Recognition Entity boundary recognition Joint prediction

来源：评论

学校读者我要写书评

暂无评论

Deepfake Video Detection Using Facial Feature Points and Ch-Transformer

引用

ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS 2025年第2期21卷 1-22页

作者： Yang, Rui Lan, Rushi Deng, Zhenrong Luo, Xiaonan Sun, Xiyan Guilin Univ Elect Technol Guangxi Key Lab Image & G Intelligent Proc Guilin Peoples R China Guilin Univ Elect Technol Nanning Res Inst Nanning Peoples R China

With the development of Metaverse technology, the avatar in Metaverse has faced serious security and privacy concerns. Analyzing facial features to distinguish between genuine and manipulated facial videos holds significant research importance for ensuring the authenticity of characters in the virtual world and for mitigating discrimination as well as preventing malicious use of facial data. To address this issue, the Facial Feature Points and Class-head-Transformer (FFP-ChT) deepfake video detection model is designed based on the clues of different FFPs distribution in real and fake videos and different displacement distances of real and fake FFPs between frames. The face video input is first detected by the BlazeFace model, and the face detection results are fed into the FaceMesh model to extract 468 FFPs. Then, the Lucas-Kanade (LK) optical flow method is used to track the points of the face, the face calibration algorithm is introduced to re-calibrate the FFPs, and the jitter displacement is calculated by tracking the FFPs between frames. Finally, the Ch is designed in the transformer, and the FFPs and FFP displacement are jointly classified through the ChT model. In this way, the designed ChT classifier is able to accurately and effectively identify deepfake videos. Experiments on open datasets clearly demonstrate the effectiveness and generalization capabilities of our approach. CCS Concepts: center dot Security and privacy - Privacy protections;Social network security and privacy;center dot Theory of computation - Structured prediction

关键词： Transformer deepfake detection facial feature FFP-ChT facial feature points

来源：评论

学校读者我要写书评

暂无评论

SHGNN: A Substructure-Aware and High Expressive graph Neural Network for graph Classification 5th

SHGNN: A Substructure-Aware and High Expressive Graph Neural...

引用

5th International Conference on Neural Computing for Advanced Applications (NCAA)

作者： Sun, Jinyong Dong, Zhiwei Sun, Zhigang Wen, Yimin Zhao, Xiang Guilin Univ Elect Technol Guangxi Key Lab Trusted Software Guilin 541004 Peoples R China Guangxi Normal Univ Key Lab Educ Blockchain & Intelligent Technol Minist Educ Guilin 541004 Peoples R China Guilin Univ Elect Technol Guangxi Key Lab Image & Graph Intelligent Proc Guilin 541004 Peoples R China Guilin Univ Elect Technol Key Lab Cognit Radio & Informat Proc Guilin 541004 Peoples R China

ISBN: (纸本)9789819770007;9789819770014

graph Neural Networks (GNNs), with their remarkable capabilities in learning graph features, have achieved success in graph classification tasks. However, the Message Passing GNNs (MP-GNNs), sharing the feature aggregation mechanism with the 1-Weisfeiler-Leman (1-WL) graph isomorphism test, suffer from limited expressiveness in downstream tasks, such as graph classification. To enhance the expressive power of GNNs, we propose a substructure-aware and high expressive graph neural network model. Firstly, Ego substructures are extracted from a graph and fed into a MP-GNN to obtain substructure-level features. Then, a count-sensitive neighborhood hashing method is introduced to encode the structural characteristics of Ego substructures, which are then injected into the corresponding substructure-level features to enhance the model's expressiveness. Next, an Ego-substructure-aware Transformer is utilized to extract global features from the graph, which use a token to fuse the interaction information among graphs' nodes, thereby generating graph-level feature vector. Finally, we apply a Multi-Layer Perceptron (MLP) followed by a softmax layer to classify graphs. Theoretical analysis demonstrates that the proposed SHGNN ismore powerful than 1-WL test. Experimental results show that, compared to the state-of-the-art graph deep learning methods, our model not only outperforms 1-WL in expressiveness, but also achieves higher classification accuracy.

关键词： graph classification graph neural networks substructure-aware neighborhood hashing Transformer

来源：评论

学校读者我要写书评

暂无评论

Residual Hybrid Attention Enhanced Video Super-Resolution with Cross Convolution 7th

Residual Hybrid Attention Enhanced Video Super-Resolution wi...

引用

7th Chinese Conference on Pattern Recognition and Computer Vision

作者： Yuan, Shiqian Li, Boyue Zhao, Xin Lan, Rushi Luo, Xiaonan Guilin Univ Elect Technol Guangxi Key Lab Image & Graph Intelligent Proc Guilin 541004 Peoples R China Guilin Univ Elect Technol Int Joint Res Lab Spatiotemporal Informat & Intel Guilin 541004 Peoples R China

ISBN: (纸本)9789819785070;9789819785087

In video super-resolution reconstruction, traditional methods often fall short in capturing details, particularly in edges and occluded areas, which affects the realism and clarity of the images. To address this issue, we propose a novel model-the Residual Hybrid Attention-Enhanced Video Super-Resolution Model, augmented by Cross Convolution techniques, denoted as RCVSR. The model ingeniously integrates a residual hybrid attention mechanism, refining the learning of global and local features through parallel channel attention and self-attention mechanisms. Simultaneously, our model introduces overlapping cross-attention blocks to enhance dynamic interactions between frames, thereby boosting the model's performance. Furthermore, the design of the cross-convolution blocks allows for parallel processing of vertical and horizontal gradient information in images, effectively extracting edge details. In multiple benchmark tests, the RCVSR model demonstrated its excellent reconstruction effects and outstanding performance.

关键词： Video super-resolution Hybrid attention Cross convolution

来源：评论

学校读者我要写书评

暂无评论

Real-Time Detection Transformer with Bi-Level Routing Attention 7th

Real-Time Detection Transformer with Bi-Level Routing Attent...

引用

7th Chinese Conference on Pattern Recognition and Computer Vision

作者： Zhaol, Xin Yuan, Shiqian Li, Boyue Tian, Rushi Luo, Xiaonan Guilin Univ Elect Technol Guangxi Key Lab Image & Graph Intelligent Proc Guilin 541004 Peoples R China Guilin Univ Elect Technol Int Joint Res Lab Spatiotemporal Informat & Intel Guilin 541004 Peoples R China

ISBN: (纸本)9789819785049;9789819785056

In recent years, DEtection TRansformer (DETR) has achieved remarkable performance for traffic monitoring. However, due to the typically high computational complexity of transformers, DETR encounters performance bottlenecks in resource-constrained scenarios. To solve the above problems, Real-Time DEtection TRansformer (RT-DETR) was recently proposed. The efficient design of RT-DETR allows for real-time object detection without sacrificing accuracy. But the model still requires a large amount of computing resources, which may limit its deployment on resource constrained devices. To make the model more lightweight, we propose a Real-Time DEtection TRansformer with Bi-Level Routing Attention (RDETR-BRA). Specifically, we incorporated an efficient hybrid encoder based on bi-level routing attention, which efficiently processes multi-scale features by decoupling intra scale interactions and inter scale fusion. Experimental results show that the RDETR-BRA effectively improves the accuracy and precision of object detection in road scenarios. On the BDD100K and Udacity datasets, the average precision of RDETR-BRA is improved by 0.6 percentage points compared to RT-DETR and 18 percentage points compared to RetinaNet. Additionally, the parameter count is reduced from 42M (RT-DETR) to 34M.

关键词： Road Traffic End-to-End Object Detect Transformer

来源：评论

学校读者我要写书评

暂无评论

Enhanced Spatial Adaptive Fusion Network For Video Super-Resolution 7th

Enhanced Spatial Adaptive Fusion Network For Video Super-Res...

引用

7th Chinese Conference on Pattern Recognition and Computer Vision

作者： Li, Boyue Zhao, Xin Yuan, Shiqian Lan, Rushi Luo, Xiaonan Guilin Univ Elect Technol Guangxi Key Lab Image & Graph Intelligent Proc Guilin 541004 Peoples R China Guilin Univ Elect Technol Int Joint Res Lab Spatiotemporal Informat & Intel Guilin 541004 Peoples R China

ISBN: (纸本)9789819785070;9789819785087

In the field of Video Super-Resolution (VSR), recurrent structures are frequently used in the designed network architectures, yet they face significant challenges in effective information transmission and accurate alignment. To address this issue, this study introduces a concise and efficient network, named Enhanced Spatial Adaptive Fusion Network (ESAFN), to enhance information flow and accuracy in VSR. The core strategy of ESAFN involves a Coupled Propagation Spatially Adaptive Module (CPSAM) and an Implicit Alignment Mechanism (IAM), innovatively restructuring the classical BasicVSR architecture. The proposed method initially promotes the effective integration of forward and backward information, then it dynamically selects the most repr esentative features across multiple scales, and further explores contextual information through Spatially Mixed Convolution (SMC) technology. With the implicit alignment strategy, our method significantly enhances the capability to restore high-frequency details in super-resolved videos, markedly surpassing current advanced technologies.

关键词： Recurrent structures Spatially adaptive Implicit alignment

来源：评论

学校读者我要写书评

暂无评论

A Novel Dual-supervised Convolutional Network for Retinal Vessel Segmentation

A Novel Dual-supervised Convolutional Network for Retinal Ve...

引用

International Conference on image processing, Computer Vision and Machine Learning (ICICML)

作者： Li, Zizheng Guan, Junlin Wang, Huadeng Guilin Univ Elect Technol Guangxi Key Lab Image & Graph Intelligent Proc Guilin 541004 Peoples R China

ISBN: (纸本)9781665464680

Retinal vascular images play an essential role in the early diagnosis of ophthalmic diseases. Automatic segmentation of retinal blood vessels is a challenging task due to the complex vascular interlacing of fundus images and the noise problems during acquisition. Because of the above problems, we propose a new network structure, which is effective for vascular tree segmentation and can segment more fine blood vessels. Firstly, to improve the segmentation ability of the network to blood vessels, we design a novel convolutional structure of FMCD from the primary level. Then, to focus on the learning of vascular characteristics from the perspectives of deep segmentation and shallow segmentation, we design a deep-shallow dual-level network from the local level. Finally, to assist in supervising the deep and shallow networks to improve the segmentation accuracy of blood vessels and complement the segmentation output of the deep and shallow networks, to segment more microvessels, we use the multi-level semantic fusion module (MSFM) from the overall level. Our Acc, Sen, and F1 on the DRIVE dataset achieve high accuracy of 0.9706, 0.8404, and 0.8332, respectively, outperforming mainstream methods. The proposed method has important clinical implications.

关键词： Retinal vessel segmentation Deep-light dual level networking Multi-level semantic fusion module Residual network

来源：评论

学校读者我要写书评

暂无评论

Multi-scale error feedback network for low-light image enhancement

引用

NEURAL COMPUTING & APPLICATIONS 2022年第23期34卷 21301-21317页

作者： Qian, Yi Jiang, Zetao He, Yuting Zhang, Shaoqin Jiang, Shenming Guilin Univ Elect Technol Guangxi Key Lab Image & Graph Intelligent Proc Guilin 541004 Peoples R China Nanchang Hangkong Univ Nanchang 330063 Jiangxi Peoples R China

Low-light image enhancement is a challenging task because brightness, contrast, noise and other factors must be considered simultaneously. However, most of the existing studies focus on improving illumination, and it is difficult to obtain natural images when the images of complex scenes are enhanced. To address this issue, we propose a neural network-a multi-scale error feedback network (MSEFN)-to enhance low-light images. The proposed network consists of an error feedback encoder module (EFEM), an error feedback decoder module (EFDM) and a feature integration module (FIM). As the main component of EFEM and EFDM, the error feedback feature extraction module can effectively retain spatial information by using the shuffle attention fusion block (SAFB) to fuse the acquired multi-scale features and nonadjacent features. FIM has the ability to capture contextual information that can compensate for the lack of global features in the network. Furthermore, the local uneven illumination (LUI) dataset and polynomial loss function constructed in this paper make our network more stable. Extensive experiments demonstrate that the proposed network outperforms state-of-the-art methods both qualitatively and quantitatively. The LUI dataset is publicly available at: https://***/Qyizos/LUI-dataset.

关键词： Low-light image enhancement Error feedback mechanism Transformer Attention mechanism

来源：评论

学校读者我要写书评

暂无评论

A Method Generating Adversarial Mark Based on Convolutional Neural Networks 13th

A Method Generating Adversarial Mark Based on Convolutional ...

引用

13th International Conference on Computer Engineering and Networks (CENet)

作者： Deng, Zhengjie Liu, Meijun Li, Xiyan Hainan Normal Univ Sch Informat Sci & Technol Haikou Hainan Peoples R China Haikou Key Lab Artificial Intelligence & Digital Haikou Hainan Peoples R China Guangxi Key Lab Image & Graph Intelligent Proc Guilin Peoples R China

ISBN: (纸本)9789819992454;9789819992430;9789819992423

Deep neural networks (DNNs) have achieved impressive results in image classification tasks. Recent studies have shown that adding imperceptible perturbations to original images can cause image recognition models to make erroneous judgments. Additionally, visible marks, which add text or image information to images, serve as a reminder of copyright ownership and can help preventing image theft and infringement. In a sense, visible text mark can be seen as meaningful noise added to clean images. In this article, we propose a method (GAM) that combines text mark and adversarial examples by focusing perturbations on meaningful text that does not affect human judgment, while causing image classification models to produce incorrect results.

关键词： Adversarial mark Convolutional neural networks Adversarial samples Perturbation

来源：评论

学校读者我要写书评

暂无评论

Focusing on diacritics to improve Vietnamese scene text detection

Focusing on diacritics to improve Vietnamese scene text dete...

引用

International Joint Conference on Neural Networks (IJCNN)

作者： Huang, Wenhui Shi, Shaoliang Wen, Yimin Guilin Univ Elect Technol Guilin Peoples R China China Acad Sci & Technol Dev GuangXi Branch Nanning Peoples R China Guilin Univ Elect Technol Guangxi Key Lab Image & Graph Intelligent Proc Guilin Peoples R China

ISBN: (纸本)9798350359329;9798350359312

Scene text detection is usually recognized as one of the fundamental challenges in computer vision and robotics. While the detection of scene text for English and Chinese has achieved great progress, Vietnamese scene text detection is still a relatively new field. In Vietnamese scene text detection, the diacritics of Vietnamese characters are often too small to be detected, making it difficult to accurately extract their features. The bounding boxes are usually placed close to Latin letters, resulting in incomplete covering for diacritics. To overcome these challenges, we propose a method of focusing on diacritics to detect Vietnamese scene text. Firstly, we propose detailed information embedding to enhance the features of diacritics. Secondly, we introduce IoU-v to allow the proposals to match the boundary of Vietnamese text instances more accurately. Lastly, in mask branch, we incorporate a diacritics feature fusion module to further enhance the features of diacritics. The experimental results illustrate the proposed method outperforms the existing approaches. The code of the proposed algorithm is available at https://***/mlmmwym/FD2VSTD.

关键词： Vietnamese characters Diacritics Scene text 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：