检索结果-内蒙古大学图书馆

Frame-by-Frame Multi-object Tracking-Guided Video Captioning

IEEE Transactions on Circuits and Systems for Video Technology 2025年

作者： Luo, Huilan Cai, Xia Shark, Lik-Kwan Jiangxi University of Science and Technology Jiangxi Province Key Laboratory of Multidimensional Intelligent Perception and Control Ganzhou341000 China University of Central Lancashire School of Engineering and Computing PrestonPR1 2HE United Kingdom

Video captioning through deep learning presents a multifaceted challenge that encompasses the extraction of complex spatio-temporal visual features and the synthesis of meaningful natural language descriptions. Most of the existing deep learning models can be broadly grouped as either convolution-based or transformer-based encoder-decoder networks, with video captions generated from features encoded at the pixel level for the former, and from features encoded at grid, frame, or video levels depending on encoder complexity for the latter. This paper advocates frame-level features as a more balanced and compact representation for fast caption generation, and introduces the Tracking-guided Information Augmentation for Captioning (Track4Cap) model, which integrates tracking-guided information augmentation to enhance frame-level features without relying on complex architectures or additional data modalities. Specifically, Track4Cap employs the Frame-by-Frame Multi-object Tracking module (FMoT) to identify the most relevant objects in the input video and the Object Relation Encoder (ORE) to model inter-object relationships as supplementary high-level cues for caption generation. By avoiding time-consuming end-to-end training and leveraging compact representations, Track4Cap achieves computational efficiency while improving captioning performance. Extensive experiments on two commonly used benchmark datasets demonstrate that Track4Cap not only achieves faster inference times but also outperforms state-of-the-art convolution-based and transformer-based video captioning models. The implementation of our method is publicly available at https://***/ccc000-png/Tracker4Cap. © 1991-2012 IEEE.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Excellent sensitivity and selectivity of g-C₄N₃ monolayer based gas senors for efficiently detecting nitrogen oxides: Spin-resolved quantum transport behaviors

引用

IEEE Sensors Journal 2025年

作者： Zhou, Wensheng Luo, Cheng Tang, Peng Xiao, Xianbo Xionga, Songbo Chen, Tong Jiangxi University of Science and Technology School of Energy and Mechanical Engineering Jiangxi Provincial Key Laboratory of Multidimensional Intelligent Perception and Control Nanchang 330013 China Jiangxi University of Chinese Medicine School of Computer Science Nanchang 330004 China

NO and NO2 are important gases that can lead to the formation of photochemical smog and acid rain. Therefore, it is important to develop high-performance nitrogen oxide sensors. Here, the electronic structure, spin tr... 详细信息

NO and NO₂ are important gases that can lead to the formation of photochemical smog and acid rain. Therefore, it is important to develop high-performance nitrogen oxide sensors. Here, the electronic structure, spin transport characteristics, and gas sensitivity of two-dimensional graphitic carbon nitride (2D g-C₄N₃) monolayers with doped magnetic metal atoms were studied via first-principles methods. A g-C₄N₃ monolayer exhibited ferromagnetic half-metallic and high-sensitivity properties for NO and NO₂ detection. The monolayer nanodevices also exhibited perfect single-spin conduction with a spin-filtering efficiency of up to 100%. A designed g-C₄N₃ monolayer sensor strongly chemisorbed both NO and NO₂, resulting in the formation of electrons in the local scattering region. This directly affected spin transport properties, resulting in NO and NO₂ sensitivity up to 100%. A g-C₄N₃ sensor with transition metal doping (especially Fe) slightly improved NO detection and significantly reduced NO₂ detection. Hence, a Fe-doped g-C₄N₃ sensor in a complex gas environment containing both NO and NO₂ largely eliminated NO₂ interference and increased NO detection. A g-C₄N₃ monolayer with excellent half-metal properties and high gas sensitivity will have potential applications in high-performance spin devices and spin-gas sensitive devices. © 2001-2012 IEEE.

关键词： Fe-doped Gas sensor Half-metal properties Spin-filtering efficiency

来源：评论

学校读者我要写书评

暂无评论

Direct Z-scheme CdTe/g-C3N4 van der Waals heterojunction for enhanced solar-to-hydrogen efficiency and spontaneous photocatalytic water splitting

引用

Molecular Catalysis 2025年 582卷

作者： Li, Yi Qiu, Xin Gong, Cheng Chen, Tong Zhang, Dong-Lan Wang, Ling-Ling Dong, Kejun Xu, Liang Jiangxi Provincial Key Laboratory of Multidimensional Intelligent Perception and Control School of Energy and Mechanical Engineering Jiangxi University of Science and Technology Jiangxi Province Nanchang330013 China Key Laboratory for Micro-Nano Optoelectronic Devices of Ministry of Education School of Physics and Electronics Hunan University Changsha410082 China Centre for Infrastructure Engineering School of Engineering Design and Built Environment Western Sydney University PenrithNSW2751 Australia

The development of an efficient photocatalyst for water splitting to produce hydrogen represents a promising solution to the energy crisis and environmental pollution. Using first-principles calculations, a CdTe/g-C3N4 van der Waals heterojunction, which exhibits outstanding stability with a direct band gap of 1.67 eV, is proposed and investigated in this work. The CdTe/g-C3N4 heterojunction functions as a direct Z-scheme heterojunction with band edges favorable for water splitting at pH = 0–14. Gibbs free energy (ΔG) calculations demonstrate the potential provided by photogenerated carriers can effectively drive spontaneous overall water splitting. Compared with CdTe and g-C3N4 monolayers, the CdTe/g-C3N4 heterojunction demonstrates superior light absorption efficiency in the visible light region, and its high electron mobility (7648.10 cm²/Vs) significantly improves photocatalytic performance. Furthermore, the solar-to-hydrogen (STH) efficiency is as high as 28.42 %. Biaxial strain ranging from -10 % to 10 % was applied to further modulate its electronic and optical properties. Notably, when the tensile strain is applied, the band gap decreases and the heterojunction exhibits a redshifted absorption peak, which enhances the absorption of visible light. Therefore, the CdTe/g-C3N4 vdW heterojunction is expected to be developed as an efficient photocatalyst for water splitting hydrogen evolution. © 2025 Elsevier B.V.

关键词： Van der Waals forces

来源：评论

学校读者我要写书评

暂无评论

The Indoor Fusion Algorithm Based on INS and UWB 6th

The Indoor Fusion Algorithm Based on INS and UWB

引用

6th International Conference on Computer Science and Educational Informatization, CSEI 2024

作者： Meng, Yu Huang, Kangni Hui, Jin Long, Keliu Jiangxi Province Key Laboratory of Multidimensional Intelligent Perception and Control Jiangxi University of Science and Technology Ganzhou341000 China School of Information Engineering Jiangxi University of Science and Technology Ganzhou341000 China School of Information Engineering Hangzhou Dianzi University Hangzhou310018 China

ISBN: (纸本)9789819637348

Ultra-wideband (UWB) positioning is highly susceptible to environmental factors, whereas inertial navigation system (INS) positioning suffers from cumulative errors over time. To address these limitations, this paper investigates the fusion of UWB and INS positioning techniques to increase the positioning accuracy. This study provides a comprehensive analysis of UWB and INS positioning principles and presents a novel approach for their integration. In the experimental setup, the time of arrival (TOA) method was employed for UWB positioning evaluation, and separate experiments were conducted to assess the performance of the INS system. The experimental results confirm that UWB positioning is prone to environmental disturbances, causing interruptions in continuous signal transmission. In contrast, the INS system experiences progressive error accumulation, which significantly degrades the positioning accuracy over time. To overcome these challenges, this paper proposes a fusion approach based on a Kalman filter algorithm aimed at combining the strengths of both positioning technologies. Comparative analysis with standalone positioning methods demonstrates that the proposed fusion algorithm significantly enhances accuracy and mitigates error accumulation, thereby validating its effectiveness. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.

关键词： Inertial navigation systems

来源：评论

学校读者我要写书评

暂无评论

Multi-Class Token Attention Learning for Weakly Supervised Semantic Segmentation 23

Multi-Class Token Attention Learning for Weakly Supervised S...

引用

23rd International Conference on Machine Learning and Cybernetics, ICMLC 2024

作者： Zeng, Zhen Luo, Huilan School of Information Engineering Jiangxi University of Science and Technology Ganzhou341000 China Jiangxi Provincial Key Laboratory of Multidimensional Intelligent Perception and Control Ganzhou341000 China

ISBN: (纸本)9798331528041

Weakly supervised semantic segmentation (WSSS) using only image-level class labels is a challenging task. Due to the local receptive fields of convolutional neural networks (CNNs), CAMs applied to CNNs often suffer from partial activation-highlighting the most discriminative parts of the image. To capture both local features and global representations, we propose an approach called MCTRM, which is a WSSS solution based on an improved Conformer as backbone. To ensure each class token's discriminative ability for the respective target class, we design a class-aware training strategy (CAT) to establish a one-to-one correspondence between the output class tokens and the ground-truth class labels. To enhance the class discrimination ability between each class token, we propose a class token regularization module (CTRM) to enhance the learning of discriminative class tokens so that the model can better capture the unique features and attributes of each class. To obtain a fine-grained CAM, we propose the refinement module (RM) to combine the class-specific attention and inherent patch-level pairwise affinity between image patches learned by the Transformer branch with the CAM of the CNN branch. Despite its simplicity, MCTRM achieves the state-of-the-art performance of 69.0% and 68.4% on the PASCAL VOC 2012 validation and test sets, respectively. © 2024 IEEE.

关键词： Semantic Segmentation

来源：评论

学校读者我要写书评

暂无评论

Gravitational wave search by time-scale-recursive denoising and matched filtering

引用

Science China(Physics,Mechanics & Astronomy) 2024年第12期67卷 166-182页

作者： Cunliang Ma Chenyang Ma Zhoujian Cao Mingzhen Jia School of Information Engineering Jiangxi University of Science and TechnologyGanzhou341000China Institute for Frontiers in Astronomy and Astrophysics Beijing Normal UniversityBeijing102206China School of Fundamental Physics and Mathematical Sciences Hangzhou Institute for Advanced StudyUCASHangzhou310024China Jiangxi Provincial Key Laboratory of Multidimensional Intelligent Perception and Control Ganzhou341000China

In our previous work [Physical Review D,2024,109(4):043009],we introduced MSNRnet,a framework integrating deep learning and matched filtering methods for gravitational wave(GW) *** with end-to-end classification methods,MSNRnet is physically *** denoising models and astrophysical discrimination models corresponding to different parameter space were operated independently for the template prediction and *** the MSNRnet has a lot of computational *** this study,we propose a new framework for template prediction,which significantly improves our previous *** new framework consists of the recursive application of denoising models and waveform classification models,which solve the problem of computational *** waveform classification network categorizes the denoised output based on the signal's time *** enhance the denoising performance for long-time-scale data,we upgrade the denoising model by incorporating Transformer and ResNet ***,we introduce a novel training approach that allows for the simultaneous training of the denoising network and waveform classification network,eliminating the need for manual annotation of the waveform dataset required in our previous ***-data analysis results demonstrate that our new method decreases the false alarm rate by approximately 25%,boosts the detection rate by roughly 5%,and slashes the computational cost by around 90%.The new method holds potential for future application in online GW data processing.

关键词： gravitational wave matched filtering deep learning

来源：评论

学校读者我要写书评

暂无评论

Multi-Class Token Attention Learning for Weakly Supervised Semantic Segmentation

Multi-Class Token Attention Learning for Weakly Supervised S...

引用

International Conference on Machine Learning and Cybernetics (ICMLC)

作者： Zhen Zeng Huilan Luo School of Information Engineering Jiangxi University of Science and Technology Ganzhou China Jiangxi Provincial Key Laboratory of Multidimensional Intelligent Perception and Control Ganzhou China

ISBN: (数字)9798331528041

ISBN: (纸本)9798331528058

关键词： Location awareness Training Deep learning Computer vision Weak supervision Semantic segmentation Transformers Cams Convolutional neural networks Cybernetics

来源：评论

学校读者我要写书评

暂无评论

Using deep learning to denoise and detect gravitational waves

引用

Physical Review D 2024年第6期110卷 063010页

作者： CunLiang Ma ShuoQiu Li Zhoujian Cao Mingzhen Jia School of Information Engineering Jiangxi Province Key Laboratory of Multidimensional Intelligent Perception and Control Ganzhou 341000 China Institute of Applied Mathematics School of Fundamental Physics and Mathematical Sciences Hangzhou Institute for Advanced Study

We have upgraded the MSNRnet framework to MSNRnet-2 by refining the training strategy, drawing inspiration from generative adversarial networks for data generation. The astrophysical discrimination network enforces constraints on the denoising output, ensuring that the “signal+noise” case conforms to an astrophysical-origin shape, while limiting the denoised output of the “noise” case to a nonastrophysical origin shape. This improvement has led to enhanced denoising and astrophysical discrimination performance in most scenarios. For the analysis of confident events, we observed a 1.3% increase in the detection rate. Remarkably, no false triggers were generated during the analysis of the data jointly processed by the Hanford and Livingston interferometers in O3b (about five months of data). These findings suggest that the denoising-discrimination-matched filtering framework holds promising potential for gravitational wave searches in the future.

关键词： Classical black holes General relativity Gravitational waves

来源：评论

学校读者我要写书评

暂无评论

Extraction of binary neutron star gravitational wave waveforms from Einstein Telescope using deep learning

arXiv

引用

arXiv 2024年

作者： Ma, CunLiang Yu, XinYao Cao, Zhoujian Jia, Mingzhen School of Information Engineering Jiangxi University of Science and Technology Ganzhou341000 China Jiangxi Provincial Key Laboratory of Multidimensional Intelligent Perception and Control Ganzhou341000 China Institute of Applied Mathematics Academy of Mathematics and Systems Science Chinese Academy of Sciences Beijing100190 China School of Fundamental Physics and Mathematical Sciences Hangzhou Institute for Advanced Study UCAS Hangzhou310024 China

In the future, the third generation (3G) gravitational wave (GW) detectors, exemplified by the Einstein Telescope (ET), will be operational. The detection rate of GW from binary neutron star (BNS) is expected to reach approximately 104 per year. To address the challenges posed by BNS GW data processing for 3G GW detectors, this paper explores the extraction of BNS waveforms from ET. Drawing inspiration from SPIIR’s matched filtering approach, we introduce a novel framework leveraging deep learning for BNS waveform extraction. By integrating denoised outputs of time-delayed strain, we can reconstruct the embedded BNS waveform. We have established three distinct BNS GW denoising models, each tailored to address the early inspiral, later inspiral, and merger phases of BNS GW, respectively. To further regulate the waveform shape, we propose the Amplitude Regularity Model that takes denoised output as input and regulated waveform as output. The experiments conducted on test data demonstrate the efficacy of the denoising models, the Amplitude Regularity Models, as well as the overall waveform construction method. To the best of our knowledge, this marks the first instance of deep learning being utilized for the task of BNS waveform extraction. We believe that the proposed method holds promise for early warning, searching, and localization of BNS GWs. Copyright © 2024, The Authors. All rights reserved.

关键词： Gravity waves

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：