检索结果-内蒙古大学图书馆

6th International Conference on Robotics and computer Vision, ICRCV 2024

作者： Luo, Die Yu, Qiuhui He, Chengwan Chen, Wei School of Computer Science and Engineering School of Artificial Intelligence Wuhan Institute of Technology Wuhan China

ISBN: (纸本)9798331527426

The present scene text extraction methods generally include two independent steps of text detection and text recognition, which is not beneficial to the interaction of feature information and may result in the accumulation of errors. Aiming at the above problems, an algorithm is proposed to achieve end-to-end text extraction by interacting information between text detection and recognition. In the process of extracting scene text, Swin-Transformer and CE-FPN are adopted as the base network to extract abundant feature representations at various scales which reduce the impact of information loss on text detection and recognition;Using Recognition Conversion to guide text localization by recognition loss enables feature interaction of detection and recognition;CTC is utilized in recognition to achieve the constraint of Attention decoding, which guarantees the full extraction of long text character information. Experimental results on several datasets show that the method in this paper demonstrates better performance compared to the existing typical models. © 2024 IEEE.

关键词： Text mining

来源：评论

学校读者我要写书评

暂无评论

Research on Named Entity Recognition in the Diabetes field Based on Improved LEBERT 7

Research on Named Entity Recognition in the Diabetes Field B...

引用

7th International Conference on computer Information Science and Application technology, CISAT 2024

作者： Li, Zhuli Yao, Feng Huang, Yucheng Li, Dianzhen Wuhan Institute of Technology School of Computer Science & Engineering School of Artificial Intelligence Wuhan China

ISBN: (纸本)9798350375107

The complexity of medical terminologies, variations in professional terms, and the presence of nested entities in the diabetes domain pose challenges for named entity recognition, which has been a focal point of research in intelligent medicine. To address these issues, this study proposes the Lexicon and Syntax Information Enhanced BERT (SLEBERT) model for named entity recognition in the diabetes domain. Building upon the LEBERT model, the SLEBERT model introduces contextual features and syntactic information instances. The SLEBERT model matches the sentence with external medical dictionaries to construct character-word pairs and analyzes syntactic information based on characters. The lexicon-syntax adapter then integrates character information vectors, word information vectors, and syntactic information vectors for feature fusion, which is subsequently incorporated between the initial layers of BERT. Additionally, a diabetes domain corpus dataset is constructed and annotated using the BIOE sequence labeling method. On the self-built DiaNer dataset and the CCKS2019 Chinese electronic medical record dataset, SLEBERT outperforms other comparative models in this study, demonstrating superior entity recognition performance and validating the effectiveness of this approach. © 2024 IEEE.

关键词： Electronic health record

来源：评论

学校读者我要写书评

暂无评论

A Study on Lightweight Algorithms for Detecting Helmet-Wearing in Electric Scooters 6

A Study on Lightweight Algorithms for Detecting Helmet-Weari...

引用

6th International Conference on Robotics and computer Vision, ICRCV 2024

作者： Hu, Po Wang, Haihui Wang, Yuzhuo Wang, Chenghui School of Computer Science and Engineering School of Artificial Intelligence Wuhan Institute of Technology Wuhan China

ISBN: (纸本)9798331527426

In urban transportation, electric bicycles serve as a primary mode of non-motorized transport, leading to increasing concerns about traffic safety, especially the helmet-wearing habits of riders. Helmets are crucial for rider safety, effectively reducing injuries in traffic accidents. Existing helmet detection algorithms struggle with small objects, occluded targets, and complex environments. To address these issues, we propose a helmet-wearing detection algorithm based on YOLOv8n. We enhance the model's feature extraction by incorporating the BiFormer attention mechanism before the first detection head, P3, improving the detection of small objects and overall performance. Additionally, in the Neck part of YOLOv8n, we replace traditional convolution modules with the lightweight G-GSConv module for efficient feature aggregation without sacrificing performance. Building on G-GSConv, we introduce the G-Ghost-Bottleneck layer, using a one-shot aggregation method to design the cross-level part network (GSCSP) module, G-GhostCSP-Bottleneck, replacing the original three CSP modules to reduce model parameters. We further integrate the BSAM attention mechanism based on the Transformer architecture to enhance the detection of dense small objects, thereby improving model accuracy. We also refine the loss function using the MDPIOU method to enhance detection flexibility and robustness. Experimental results on the SHWD dataset show that our improved model achieves an average precision of 88.9%, a 3.6% improvement over the original model. The proposed algorithm effectively improves helmet-wearing detection accuracy in complex traffic scenarios, demonstrating its potential for enhancing urban traffic safety management. © 2024 IEEE.

关键词： Nonmotorized transportation

来源：评论

学校读者我要写书评

暂无评论

Global Context-Aware Method for Infrared Image Non-Uniformity Correction 3

Global Context-Aware Method for Infrared Image Non-Uniformit...

引用

3rd International Conference on artificial intelligence and computer Information technology, AICIT 2024

作者： Ma, Zhiwei Zhang, Shuiping Wang, Lei Rao, Qiang School of Computer Science and Engineering School of Artificial Intelligence Wuhan Institute of Technology Wuhan China

ISBN: (纸本)9798350365443

Infrared imaging systems are often affected by non-uniformity, which can significantly reduce the quality of captured images and hinder subsequent image-processing tasks. Existing infrared image non-uniformity correction methods mainly suffer from incomplete detail recovery and easy ghosting. This paper proposes a global context-aware infrared image non-uniformity correction method, which effectively captures long-distance dependencies and complex feature patterns by combining residual networks and self-attention mechanisms. To verify the effectiveness of our method, we conducted extensive experiments on multiple public datasets. Experimental results show that compared with the existing state-of-the-art technology, our method performs better in both qualitative and quantitative evaluations and can effectively remove infrared image stripe noise in complex scenes while retaining key details of the image. © 2024 IEEE.

关键词： Infrared imaging

来源：评论

学校读者我要写书评

暂无评论

A Deep Insight on Cricket Video to Text Summarization Using Neural Networks 10

A Deep Insight on Cricket Video to Text Summarization Using ...

引用

10th International Conference on Advanced Computing and Communication Systems, ICACCS 2024

作者： Gopal, J. Vijay Prasad, K. Sai Kumar, Dhatrika Kamal Prashanth, Maddhe Sai Kumar, Vuppala Praneeth Teja, I. Nikhil Sri Sai Computer Science Engineering Artificial Intelligence and Machine Learning Mlr Institute of Technology Hyderabad India

ISBN: (纸本)9798350384369

In this modern era, the demand for efficient and automated cricket video summarization techniques is rapidly increasing. This paper introduces an innovative and advance neural network system that transforms the way cricket match videos are summarized. Cricket video-to-text summarization system overcomes the limitations of traditional manual summarization approach by utilizing various deep learning techniques to completely automate the summarization process, our system can extract crucial insights from lengthy cricket match footage and convert them into easily readable text formats. The system employs three-pronged approach which involves extraction of visual features using VGG16 Convolutional Neural Network (CNN), scoreboard information is extracted through Optical Character Recognition (OCR) technology, and Text Summarization performed by Long Short-Term Memory (LSTM) network. Our system revolutionizes the way cricket enthusiasts engage with match videos, providing a game-changing experience for fans worldwide. © 2024 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Drowsy Driver Detection System using computer Vision and Transfer Learning for Preventing Road Accidents 3

Drowsy Driver Detection System using Computer Vision and Tra...

引用

3rd IEEE International Conference on Blockchain and Distributed Systems Security, ICBDS 2024

作者： Bagal, Naisargi Dingreja, Kashish Atterkar, Atharva Bachhav, Prasad Sable, Nilesh P. Yenkikar, Anuradha Alkunte, Ritesh Vishwakarma Institute of Information Technology Department of Computer Science and Engineering - Artificial Intelligence Pune India

ISBN: (纸本)9798350354348

Major traffic accidents are attributed to driver fatigue, according to study on the topic. Driver drowsiness is a state in which the driver of a car is on the verge of falling asleep or losing consciousness. It can be brought on by a number of variables, including biological, physical, psychological, and other factors, all of which can impede and inhibit safe driving. The several methods that can be applied to the implementation of a system for the detection of driver sleepiness will be reviewed in this paper. Improving the ability to identify drowsiness in real time through computer vision is the main objective of all technologies. Thus, our work also pertains to the identification of driver drowsiness, whereby a motorist's tiredness can be ascertained by first recognizing their face, followed by eye tracking. The system compares the extracted eye image with the dataset. The system used the dataset and predicts using transfer learning with our base model as InceptonV3 to identify that it could alert the driver with an alarm if the driver's eyes were closed for a predetermined amount of time, and it could resume monitoring if the driver's eyes were open following the alarm alert. We established a score that grew if the eyes were closed and dropped if they were open. This study aims to reduce traffic accidents by solving the sleepiness detection problem with InceptionV3 with an accuracy of 95.46% compare to 92.11% of ***, driving drowsiness detection - which assesses a driver's degree of fatigue using facial recognition and eye tracking becomes the main focus of our research. © 2024 IEEE.

关键词： Face recognition

来源：评论

学校读者我要写书评

暂无评论

Research on predicting legal judgments by incorporating key information 7

Research on predicting legal judgments by incorporating key ...

引用

7th International Conference on computer Information Science and Application technology, CISAT 2024

作者： Xu, Fan Liu, Wei Guo, Kao Wang, Zihao Wuhan Institute of Technology School of Computer Science & Engineering School of Artificial Intelligence Wuhan China

ISBN: (纸本)9798350375107

The prediction of legal judgments is based on the description of case facts to predict the final charges. Through judgment prediction technology, the judicial system can handle a large number of cases more efficiently and provide more accurate judgment results. However, in reality, the types of charges are numerous and complex. To address the issues of severe data imbalance and easily confusable charges, this paper proposes a legal judgment model that incorporates key information. Firstly, it combines the ALBERT (A Lite BERT) pre-trained model and the bidirectional Gated Recurrent Unit (GRU) model to encode cases, efficiently capturing the contextual semantic information in case descriptions. Then, the concept of key information is introduced, using attention mechanisms and statistical methods to extract two distinguishable parts of key information from the cases, which are then concatenated with the case encoding for the final prediction. Additionally, category classification weights are employed to enhance the model's ability to predict low-frequency samples. The proposed method has achieved good results on real legal judgment datasets. © 2024 IEEE.

关键词： Prediction models

来源：评论

学校读者我要写书评

暂无评论

Video-Based Depth Estimation Autoencoder With Weighted Temporal Feature and Spatial Edge Guided Modules

IEEE Transactions on Artificial Intelligence

引用

IEEE Transactions on artificial intelligence 2024年第2期5卷 613-623页

作者： Yang, Wei-Jong Tsung, Wan-Nung Chung, Pau-Choo Department of Artificial Intelligence and Computer Engineering National Chin-Yi University of Technology Taichung411030 Taiwan Department of Electrical Engineering National Cheng Kung University Tainan701 Taiwan

Convolutional neural networks with encoder and decoder structures, generally referred to as autoencoders, are used in many pixelwise transformation, detection, segmentation, and estimation applications, for example, which can be applied for face swapping, lane detection, semantic segmentation, and depth estimation, respectively. However, traditional autoencoders, which are based on single-frame inputs, ignore the temporal consistency between consecutive frames, and may, hence, produce unsatisfactory results. Accordingly, in this article, a video-based depth estimation (VDE) autoencoder is proposed to improve the quality of depth estimation through the inclusion of two weighted temporal feature (WTF) modules in the encoder and a single spatial edge guided (SEG) module in the decoder. The WTF modules designed with channel weighted block submodule effectively extract the temporal similarities in consecutive frames, whereas the SEG module provides spatial edge guidance of the object contours. Through the collaboration of the proposed modules, the accuracy of the depth estimation is greatly improved. The experimental results confirm that the proposed VDE autoencoder achieves a better monocular depth estimation performance than the existing autoencoders with only a slight increase in the computational cost. © 2023 IEEE.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Classifying Misinformation of User Credibility in Social Media Using Supervised Learning

引用

computers, Materials & Continua 2023年第5期75卷 2921-2938页

作者： Muhammad Asfand-e-Yar Qadeer Hashir Syed Hassan Tanvir Wajeeha Khalil Department of Computer Science Center of Excellence in Artificial Intelligence(CoE-AI)Bahria UniversityIslamabad44000Pakistan Department of Computer Science and Information Technology University of Engineering and TechnologyPeshawar25000Pakistan

The growth of the internet and technology has had a significant effect on social *** information has become an important research topic due to the massive amount of misinformed content on social *** is very easy for any user to spread misinformation through the ***,misinformation is a problem for professionals,organizers,and ***,it is essential to observe the credibility and validity of the News articles being shared on social *** core challenge is to distinguish the difference between accurate and false *** studies focus on News article content,such as News titles and descriptions,which has limited their ***,there are two ordinarily agreed-upon features of misinformation:first,the title and text of an article,and second,the user *** the case of the News context,we extracted different user engagements with articles,for example,tweets,i.e.,read-only,user retweets,likes,and *** calculate user credibility and combine it with article content with the user’s *** combining both features,we used three Natural language processing(NLP)feature extraction techniques,i.e.,Term Frequency-Inverse Document Frequency(TF-IDF),Count-Vectorizer(CV),and Hashing-Vectorizer(HV).Then,we applied different machine learning classifiers to classify misinformation as real or ***,we used a Support Vector Machine(SVM),Naive Byes(NB),Random Forest(RF),Decision Tree(DT),Gradient Boosting(GB),and K-Nearest Neighbors(KNN).The proposed method has been tested on a real-world dataset,i.e.,“fakenewsnet”.We refine the fakenewsnet dataset repository according to our required *** dataset contains 23000+articles with millions of user *** highest accuracy score is 93.4%.The proposed model achieves its highest accuracy using count vector features and a random forest *** discoveries confirmed that the proposed classifier would effectively classify misinformat

关键词： Misinformation user credibility fake news machine learning

来源：评论

学校读者我要写书评

暂无评论

Selective arguments representation with dual relation-aware network for video situation recognition

引用

Neural Computing and Applications 2024年第17期36卷 9945-9961页

作者： Liu, Wei He, Qing Wang, Chao Peng, Yan Xie, Shaorong School of Computer Engineering and Science Shanghai University Shanghai200444 China Shanghai Artificial Intelligence Laboratory Shanghai201114 China School of Future Technology Shanghai University Shanghai200444 China Institute of Artificial Intelligence Shanghai University Shanghai200444 China

Argument visual states are helpful for detecting structured components of events in videos, and existing methods tend to use object detectors to generate their candidates. However, directly leveraging object features captured by bounding boxes overlooks a deep understanding of object relations and differences between them and real arguments. In this work, we propose a novel framework to generate selective contextual representations of videos, thereby reducing the interference of useless or incorrect object features. Firstly, we construct grid-based object features as graphs based on the internal grid connection and then use graph convolutional network to execute feature aggregation. Secondly, a weighted geometric attention module is designed to obtain the contextual representation of objects, which explicitly combines visual similarity and geometric correlation with different importance proportions. Then, we propose a dual relation-aware selection module for further feature selection. Finally, we utilize labels as the ladder to bridge the gap between object features and semantic roles, while considering the proximity in the semantic space. Experimental results and extensive ablation studies on the VidSitu indicate that our method effectively obtains a deep understanding of events in videos and outperforms state-of-the-art models. © The Author(s), under exclusive licence to Springer-Verlag London Ltd., part of Springer Nature 2024.

关键词： Object 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：