检索结果-内蒙古大学图书馆

Debiased Visual Question Answering via the perspective of question types

PATTERN RECOGNITION LETTERS 2024年 178卷 181-187页

作者： Huai, Tianyu Yang, shuwen Zhang, Junhang Zhao, Jiabao he, Liang East China Normal Univ Sch Comp Sci & Technol Shanghai Peoples R China East China Normal Univ Inst AI Educ Shanghai Peoples R China East China Normal Univ Shanghai Key Lab Multidimens Informat Proc Shanghai Peoples R China

Visual Question Answering (VQA) aims to answer questions according to the given image. However, current VQA models tend to rely solely on textual information from the questions and ignore the visual information in the images to get answers, which is caused by bias that is generated during the training phase. Previous studies have shown that bias in VQA is mainly caused by the text modality, and our analysis suggests that question type is a crucial factor in bias formation. To address this bias, we proposed a self -supervised method including the Against Biased Samples(ABS) module that performs targeted debiasing by selecting samples that are prone to bias, and the Shuffle Question types(SQT) module that constructs negative samples by randomly replacing the question types of the samples selected by the ABS, to interrupting the shortcuts from question type to answer. Our approach mitigates the question -to -answer bias without using external annotations, overcoming the prior language problem. Additionally, we designed a new objective function for negative samples. Experimental results indicate that our method outperforms both self -supervised -based and supervised -based state-of-the-art approaches, achieving 70.36% accuracy on the VQA-CP v2 dataset.

关键词： Visual Question Answering De-biasing Self-supervised

来源：评论

学校读者我要写书评

暂无评论

Effect on In³⁺ co-doping on photoluminescence and scintillation properties of LYSO:Ce crystal

引用

PHYSICA B-CONDENSED MATTER 2025年 699卷

作者： Yu, Jiaqi Li, Mingqing Zhao, shuwen Xue, Zhongjun Ding, Dongzhou Feng, he Shanghai Univ Sch Mat Sci & Engn Shanghai 200444 Peoples R China Chinese Acad Sci Shanghai Inst Ceram Shanghai 201800 Peoples R China

In this work, a series of LYSO:0.03 at.% Ce,xat.%In (x = 0, 0.001, 0.005 and 0.01) crystals were successfully grown for the first time by the optical floating zone (FZ) method. The effect of In3+ co-doping on photoluminescence and scintillation properties of LYSO:Ce scintillation single crystal was investigated. It was found that at x = 0.001, the overall scintillation performance was optimized. This was attributed to the fact that In3+ co-doping reduces the concentration of Ce2 emission center and accelerates the photoluminescence decay time of Ce3+. The temperature dependence of photoluminescence of LYSO:Ce,xat.%In samples was also investigated. Combined with density functional theory calculation, the phenomenon of In3+ co-doping leading to a decrease in quenching temperature of decay time was explained in terms of both crossover relaxation and introduction of impurity energy levels.

关键词： Scintillation crystal LYSO:Ce In co-doping Figure of merit Temperature dependence

来源：评论

学校读者我要写书评

暂无评论

Centrality and bridge connections between cognitive emotion regulation strategies and professional identity among Chinese undergraduate nursing students: A network analysis

引用

NURSE EDUCATION IN PRACTICE 2024年 80卷 104151页

作者： Liu, Jingya Zhang, Haoran Tao, Siyu he, Jie Li, shuwen Anhui Med Univ Sch Nursing 15 Feicui Rd Hefei 230601 Peoples R China

Aim: To construct a network structure for cognitive emotion regulation among Chinese undergraduate nursing students and identify central nodes and to explore the bridge connections between cognitive emotion regulation strategies and professional identity from the perspective of network analysis. Background: Undergraduate nursing students are currently in a poor psychological condition and cognitive emotion regulation strategies can help them use positive approaches to regulate their emotions. There may be a link between cognitive emotion regulation strategies and professional identity. Design: This was a cross-sectional study that used network analysis. Method: A total of 218 Chinese undergraduate nursing students were selected and surveyed using the Cognitive Emotion Regulation Questionnaire and the Professional Identity Questionnaire. A network analysis model was constructed and the related indices were calculated using R 4.3.0 software. Results: Network analysis showed that the central nodes of undergraduate nursing students' cognitive emotion regulation strategies were positive reappraisal, refocusing on planning and catastrophising;In the bridge network of cognitive emotion strategies and professional identity, professional self-concept, positive reappraisal, benefits of stay and risk of resignation and refocusing on planning were the nodes with the strongest bridge strength. Conclusion: The salient central and bridge nodes can serve as potential targets for interventions aimed at improving the mental health of undergraduate nursing students. Nursing educators must be trained in cognitive emotion regulation and appropriately guided to use positive emotion regulation strategies in their studies and work. Educators should focus on the relation between cognitive emotion regulation and professional identity to improve the mental health of nursing students and stabilise the nursing workforce.

关键词： Undergraduate nursing students Cognitive emotion regulation Professional identity Network analysis Bridge networks

来源：评论

学校读者我要写书评

暂无评论

Maritime Target Radar Detection and Tracking via DTNet Transfer Learning Using Multi-Frame Images

引用

REMOTE SENSING 2025年第5期17卷 836-836页

作者： he, Xiaoyang Chen, Xiaolong Du, Xiaolin Wang, Xinghai Xu, shuwen Guan, Jian Yantai Univ Sch Comp & Control Engn Yantai 264005 Peoples R China Naval Aviat Univ Marine Target Detect Res Grp Yantai 264001 Peoples R China Xidian Univ Sch Elect Engn Xian 710071 Peoples R China

Traditional detection and tracking methods struggle with the complex and dynamic maritime environment due to their poor generalization capabilities. To address this, this paper improves the YOLOv5 network by integrating Transformer and a Convolutional Block Attention Module (CBAM) with the multi-frame image information obtained from radar scans. It proposes a detection and tracking method based on the Detection Tracking Network (DTNet), which leverages transfer learning and the DeepSORT tracking algorithm, enhancing the detection capabilities of the model across various maritime environments. First, radar echoes are preprocessed to create a dataset of Plan Position Indicator (PPI) images for different marine conditions. An integrated network for detecting and tracking maritime targets is then designed, utilizing the feature differences between moving targets and sea clutter, along with the coherence of inter-frame information for moving targets, to achieve multi-target detection and tracking. The proposed method was validated on real maritime targets, achieving a precision of 99.06%, which is a 7.36 percentage point improvement over the original YOLOv5, demonstrating superior detection and tracking performance. Additionally, the impact of maritime regions and weather conditions is discussed, showing that, when transferring from Region I to Regions II and III, the precision reached 92.2% and 89%, respectively, and, when facing rainy weather, although there was interference from the sea clutter and rain clutter, the precision was still able to reach 82.4%, indicating strong generalization capabilities compared to the original YOLOv5 network.

关键词： maritime moving targets detection and tracking radar images deep learning inter-frame information

来源：评论

学校读者我要写书评

暂无评论

Enzyme-mimetic antioxidant nanomaterials for ROS scavenging: Design, classification, and biological applications

引用

COORDINATION CheMISTRY REVIEWS 2024年 500卷

作者： Zhou, shuwen Cai, Huijuan he, Xinjuan Tang, Zhiyong Lu, Siyu Zhengzhou Univ Green Catalysis Ctr Zhengzhou 450001 Henan Peoples R China Zhengzhou Univ Coll Chem Zhengzhou 450001 Henan Peoples R China Natl Ctr Nanosci & Technol CAS Ctr Excellence Nanosci CAS Key Lab Nanosyst & Hierarch Fabricat Beijing 100190 Peoples R China

Reactive oxygen species (ROS) perform vital functions in numerous life processes and significantly affect physiological activities in different organ systems in the human body. However, excessive ROS accumulation is harmful and causes oxidative damage or ROS-related diseases. Recently, various natural enzyme-like nanomaterials with ROS-scavenging activities have been designed. These nanomaterials can be used as alternative substitutes for ideal antioxidants. This review summarizes various antioxidant nanomaterials in terms of their design, classification, and biological applications. First, different design strategies (e.g., modulating the size, shape, composition, ligand, or surface chemistry of nanomaterials or forming atomically distributed metal active centres) for regulating antioxidant activity are analysed. Second, various antioxidant nanomaterials based on the enzymatic properties and types of scavenged ROS are classified and summarized. Third, the representative biological applications of antioxidant enzyme-mimetic nanomaterials, including therapeutics, protective effects, and other ROS-related applications, are described. Finally, the challenges and perspectives regarding the use of these enzyme-like antioxidant nanomaterials are outlined. This review provides a deeper understanding of the design and biological application of antioxidant nanomaterials and offers guidance for improving biocatalytic performance and selecting future research directions.

关键词： Antioxidant Nanozyme Nanomaterial Reactive oxygen species scavenging Biological application

来源：评论

学校读者我要写书评

暂无评论

Exogenous selenium foliar application on nutrition, grain yield and quality of rice ( Oryza sativa L.)

引用

JOURNAL OF FOOD COMPOSITION AND ANALYSIS 2024年 130卷

作者： Su, Dan Xue, shuwen Tie, Mei Sun, Jifeng he, Zhenli Liaoning Univ Sch Environm Sci Shenyang 110036 Peoples R China Univ Florida Indian River Res & Educ Ctr Dept Soil Water & Ecosyst Sci IFAS Ft Pierce FL 34945 USA

Selenium (Se) supplementation through plant-based foods is easy to implement and effective;therefore, it is important to explore the Se-enrichment ability of staple grains and analyze their quality. The purpose of this study was to investigate the effects of foliar application of Se on the quality, Se-enriched capacity, organic Se composition and content of rice under field cultivation. The results showed that exogenous Se application at 80 mg L-1 contributed the most to the yield, and eight elements differed accordingly to the application of Se, promoted the accumulation of Se-containing organic matter in rice grain. This indicates that rational foliar application of Se can effectively improve the quality of rice grains. Moreover, selenomethionine (SeMet), selenocysteine (SeCys2), and Se-methylselenocysteine (Se-MeSeCys) are important components of Se-enriched in hydrolyzed proteins in grains. Based on these results, foliar spraying of Se is a suitable method to supplement rice with additional Se and can improve the yield and quality of rice grains, particularly the quality of organic Se and Se-containing amino acids.

关键词： Se-enrichment HG -AFS ICP-MS HPLC-MS Seleno-amino acid Se-containing organic matter

来源：评论

学校读者我要写书评

暂无评论

LandslideNet: A landslide semantic segmentation network based on single-temporal optical remote sensing images

引用

ADVANCES IN SPACE RESEARCH 2024年第10期74卷 4616-4638页

作者： Zhu, Xinyu Zhang, Zhihua he, Yi Wang, Wei Yang, shuwen Hou, Yuhao Lanzhou Jiaotong Univ Fac Geomatics Lanzhou 730070 Gansu Peoples R China Natl Local Joint Engn Res Ctr Technol & Applicat N Lanzhou 730070 Gansu Peoples R China Key Lab Sci & Technol Surveying & Mapping Gansu Pr Lanzhou 730070 Gansu Peoples R China

Swiftly and accurately acquiring the spatial distribution, location, and magnitude of landslides while documenting them in a landslide cataloging database can furnish crucial information for precise disaster mitigation measures and secondary hazard prevention. The extraction of landslides using existing semantic segmentation algorithms may give rise to issues such as false detection and missed detection due to the diverse shape and texture features of landslides in remote sensing images, the abundance of spectral features, and the complexity of the environment. In this article, we proposed LandslideNet, a novel model specifically designed for accurate segmentation of landslides in single-temporal high spatial resolution optical remote sensing images. By constructing a landslide image dataset and employing the LandslideNet model, we successfully identify and segment landslides with high precision. Quantitative experimental results demonstrate that our LandslideNet achieves superior performance compared to widely used semantic segmentation models including U-Net, PSPNet, Deeplabv3+, HRNetv2, Segformer and GELAN-c with F1-score, mIoU, FWIoU, mPA and OA reaching 72.53 %, 78.41 %, 99.86 %, 83.33 % and 99.93 % respectively. Moreover, our model exhibits lower complexity while demonstrating improved capability in detecting landslides with complex shapes and different sizes. (c) 2024 COSPAR. Published by Elsevier B.V. All rights are reserved, including those for text and data mining, AI training, and similar technologies.

关键词： Remote sensing Landslide extraction Deep learning YOLOv8 Google earth

来源：评论

学校读者我要写书评

暂无评论

Real-Time Ultraviolet Flame Detection System Based on 4H-SiC Phototransistor

引用

IEEE TRANSACTIONS ON ELECTRON DEVICES 2024年第3期71卷 1580-1586页

作者： Huang, Danyang Zhao, Xiaolong Li, Quan Liang, Zhaozhao Guo, shuwen he, Yongning Xi An Jiao Tong Univ Sch Microelect Key Lab Micro Nano Elect & Syst Integrat Xian City Xian 710049 Shaanxi Peoples R China

High-sensitivity ultraviolet (UV) photoelectric detection technology is significant in UV communication, missile tail flame radiation detection, and fire early warning. This article utilized a 4H-SiC phototransistor detector to respond to weak UV light. A corresponding detection circuit system was developed to convert the photogenerated analog current signal to a digital voltage signal and transmit the digital signal to the host computer. A trans-impedance amplifier amplifies and transfers the weak photocurrent into a large voltage signal, which possesses a stronger antijamming ability. The analog-to-digital converter converts the analog voltage signal into a digital signal for convenient transmission in the following circuits. A universal serial bus interface module transmits the digital signal to the host computer using a serial transmission method with a field-programmable gate array controlling the timing. A graphical software interface based on LabVIEW was developed to display the collected digital data that stand for real-time UV intensity. Monitoring and recording the weak optical power of the UV light emitted by candlelight can be realized via the system. The UV power density range was between 50 nW/cm(2) and 3 mu W/cm(2) , and the corresponding output voltages ranged from 0.095 to 3.45 V. Compared with traditional high-voltage biased p-i-n avalanche diode detection system, the flame detection system demonstrated in this article was based on a 4H-SiC phototransistor, which has the advantages of low voltage and high stability.

关键词： Detectors Fires Phototransistors Photoconductivity Voltage Anodes Real-time systems 4H-SiC phototransistor data acquisition circuit ultraviolet (UV) detection visible-blind UV detection wide-bandgap detection system

来源：评论

学校读者我要写书评

暂无评论

Simple contrastive learning in a self-supervised manner for robust visual question answering

引用

COMPUTER VISION AND IMAGE UNDERSTANDING 2024年 241卷

作者： Yang, shuwen Xiao, Luwei Wu, Xingjiao Xu, Junjie Wang, Linlin he, Liang East China Normal Univ Sch Comp Sci & Technol Shanghai 200062 Peoples R China Fudan Univ Sch Comp Sci Shanghai 200433 Peoples R China

Recent observations have revealed that Visual Question Answering models are susceptible to learning the spurious correlations formed by dataset biases, i.e., the language priors, instead of the intended solution. For instance, given a question and a relative image, some VQA systems are prone to provide the frequently occurring answer in the dataset while disregarding the image content. Such a preferred tendency has caused them to be brittle in real -world settings, harming the robustness of VQA models. We experimentally found that conventional VQA methods often confuse negative samples that with identical questions but different images, which results in the generation of linguistic bias. In this paper, we propose a simple contrastive learning scheme, namely SCLSM, to mitigate the above issues in a self -supervised manner. We construct several special negative samples and introduce a debiasing-aware contrastive learning approach to help the model learn more discriminative multimodal features, thus improving the ability of debiasing. The SCLSM is compatible with numerous VQA baselines. Experimental results on the widely -used public datasets VQA-CP v2 and VQA v2 validate the effectiveness of our proposed model.

关键词： Visual question answering Deep learning Contrastive learning Information extraction

来源：评论

学校读者我要写书评

暂无评论

Adaptive convolutional sparsity with sub-band correlation in the NSCT domain for MRI image fusion

引用

PHYSICS IN MEDICINE AND BIOLOGY 2024年第5期69卷 055022-055022页

作者： Hu, Qiu Cai, Weiming Xu, shuwen Hu, Shaohai Wang, Lang he, Xinyi NingboTech Univ Sch Informat Sci & Engn Ningbo 315100 Peoples R China Zhejiang Engn Res Ctr Intelligent Marine Ranch Eq Ningbo 315100 Peoples R China China Elect Technol Grp Corp Res Inst 3 Beijing 100846 Peoples R China Beijing Jiaotong Univ Inst Informat Sci Beijing 100044 Peoples R China Beijing Key Lab Adv Informat Sci & Network Techno Beijing 100044 Peoples R China Ningbo Xiaoshi High Sch Ningbo 315100 Peoples R China

Objective. Multimodal medical image fusion (MMIF) technologies merges diverse medical images with rich information, boosting diagnostic efficiency and accuracy. Due to global optimization and single-valued nature, convolutional sparse representation (CSR) outshines the standard sparse representation (SR) in significance. By addressing the challenges of sensitivity to highly redundant dictionaries and robustness to misregistration, an adaptive convolutional sparsity scheme with measurement of the sub-band correlation in the non-subsampled contourlet transform (NSCT) domain is proposed for MMIF. Approach. The fusion scheme incorporates four main components: image decomposition into two scales, fusion of detail layers, fusion of base layers, and reconstruction of the two scales. We solved a Tikhonov regularization optimization problem with source images to obtain the base and detail layers. Then, after CSR processing, detail layers were sparsely decomposed using pre-trained dictionary filters for initial coefficient maps. NSCT domain's sub-band correlation was used to refine fusion coefficient maps, and sparse reconstruction produced the fused detail layer. Meanwhile, base layers were fused using averaging. The final fused image was obtained via two-scale reconstruction. Main results. Experimental validation of clinical image sets revealed that the proposed fusion scheme can not only effectively eliminate the interference of partial misregistration, but also outperform the representative state-of-the-art fusion schemes in the preservation of structural and textural details according to subjective visual evaluations and objective quality evaluations. Significance. The proposed fusion scheme is competitive due to its low-redundancy dictionary, robustness to misregistration, and better fusion performance. This is achieved by training the dictionary with minimal samples through CSR to adaptively preserve overcompleteness for detail layers, and constructing fusion activity

关键词： multimodal medical image image fusion sparse representation convolutional sparse representation non-subsampled contourlet transform misregistration

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：