检索结果-内蒙古大学图书馆

5th International Conference on Computer Vision, image and deep learning, CVIDL 2024

作者： Wang, Guoen Lin, Hongsheng Wang, Qingbo Naval Petty Officer School Anhui Province Bengbu233012 China Naval Research Institute Beijing100161 China

ISBN: (纸本)9798350373820

Detection and tracking of underwater targets is a prerequisite for autonomous operations of underwater robots. The optical imaging is often used for close range environmental perception for underwater robots because of its high resolution. deep learning is widely used in object detection tasks, but it takes a long time and is rarely applied in tracking tasks with high real-time requirements. Since correlation filters have been applied to target tracking, their ultra-high speeds have attracted widespread attention. In this paper, a deep learning target detection method is combined with a kernel correlation filter to solve the real-time problem of underwater target tracking. In this paper, experiments are conducted using the fish image dataset Labelled Fishes in the Wild and the results show that the method in this paper is feasible. © 2024 IEEE.

关键词： Target tracking

来源：评论

学校读者我要写书评

暂无评论

Enhanced fault feature extraction and bearing fault diagnosis using shearlet transform and deep learning

引用

SIGNAL image AND VIDEO processing 2024年第12期18卷 9285-9293页

作者： Swami, Preety D. Jha, Rakesh Kumar Jat, Anuradha RGPV Dept Elect & Commun Engn Bhopal 462033 India Aartech Solon Ltd Ind Res Design & Dev Div Bhopal 462016 India

Accurate bearing fault diagnosis is essential for ensuring the health and longevity of mechanical systems. Traditional methods often struggle with the dynamic operating conditions of machinery, including variations in speed, load, and noise. This paper proposes a novel deep learning-based approach for robust bearing fault diagnosis. The method utilizes a combination of Shearlet Transform, Autoencoder, and Softmax Classifier. Vibration signals from healthy and faulty bearings are transformed into 2D image representations, capturing intricate details of the underlying mechanical state. Shearlet Transform is then employed to enhance these images, specifically targeting and amplifying subtle fault signatures, leading to improved diagnostic accuracy. The enhanced images are subsequently fed to an autoencoder, where the encoder compresses the data into a lower-dimensional feature space. These compressed features are then used to train and optimize a Softmax Classifier for effective fault classification. The proposed methodology is evaluated under diverse speed and load conditions, mimicking real-world operating scenarios. The achieved high classification accuracy across various operating points demonstrates the robustness and effectiveness of the proposed approach.

关键词： Autoencoder Bearing fault diagnosis Machine learning Shearlet transform Softmax classifier

来源：评论

学校读者我要写书评

暂无评论

Multi-Feature Fusion for Enhanced Feature Representation in Automatic Modulation Recognition

引用

IEEE ACCESS 2025年 13卷 1164-1178页

作者： Cao, Jiuxiao Zhu, Rui Wu, Lingfeng Wang, Jun Shi, Guohao Chu, Peng Zhao, Kang Xijing Univ Sch Elect Informat Xian 710123 Peoples R China

Modulation recognition plays a crucial role in the efficient management of spectrum resources. However, traditional methods have long posed challenges for researchers due to their excessive reliance on manual effort. With the advancement of deep learning, automatic modulation recognition has emerged as a promising solution. Nonetheless, most existing deep learning-based modulation recognition studies consider only single-domain feature information of the signal, such as time-domain or frequency-domain features. To further enhance recognition performance, this paper proposes a novel multi-feature input fusion network. By utilizing different representations and processing methods of the signal, the proposed approach designs distinct feature extraction networks tailored to specific processed signals, leveraging the characteristics of convolution kernels with varying sizes and receptive fields. Additionally, the attention mechanism module is improved to enhance network performance while minimizing the increase in parameter count. Experimental results on the publicly available RML2016.10a dataset demonstrate that the proposed model achieves highly efficient recognition above 2 dB, with accuracy approaching 100%. Testing with real-world signals collected under realistic channel conditions using USRP devices further confirms the model's superior performance, achieving an effective recognition rate of 99.27% when the two USRP devices are in close proximity. These results validate the model's high efficiency and robustness in both simulated and real-world environments.

关键词： Modulation Convolution Feature extraction Kernel Clocks time-domain analysis Attention mechanisms deep learning Accuracy Neural networks Modulation recognition deep learning attention mechanism multi-input network concatenated USRP

来源：评论

学校读者我要写书评

暂无评论

Efficient One-Stage deep learning for Text Detection in Scene images

EEA - Electrotehnica, Electronica, Automatica

引用

EEA - Electrotehnica, Electronica, Automatica 2024年第4期72卷 65-71页

作者： Tibermacine, Ahmed Guettala, Walid Tibermacine, Imad Eddine Faculty of Exact Sciences Natural And Life Sciences University Mohamed Khider of Biskra Algeria Computer Science Department University Mohamed Khider of Biskra Algeria Department of Computer Automation and Management Engineering Sapienza University of Rome Italy

Text detection in natural images is a crucial task for extracting and recognizing valuable information, but it comes with significant challenges. Traditional image processing methods often rely on synthetic features, which may not be suitable for handling the diverse range of unstructured text scenarios encountered in real-world environments. To overcome this challenge, deep learning techniques have emerged as a powerful tool for adaptively learning and extracting features to digitize text. In this paper, we introduce a novel approach based on the YOLOv5 architecture for text detection. Our method leverages the strengths of YOLOv5, including its efficient and lightweight architecture, multi-scale prediction capability, and advanced data augmentation techniques. By employing these features, our proposed model can accurately identify regions in an image that are more likely to contain text. Through extensive experimentation on popular benchmark datasets, we have evaluated the performance of our proposed model. The results have demonstrated that our method outperforms state-of-the-art approaches in terms of both detection efficiency and accuracy. This indicates the effectiveness of our approach in handling the challenges posed by text detection in natural images. © Editura ELECTRA 2024. All rights reserved.

关键词： deep learning Natural scene image Text detection Yolov5

来源：评论

学校读者我要写书评

暂无评论

Investigation of steganalysis performance of selected steganography methods with deep learning models

JOURNAL OF THE FACULTY OF ENGINEERING AND ARCHITECTURE OF GA...

引用

JOURNAL OF THE FACULTY OF ENGINEERING AND ARCHITECTURE OF GAZI UNIVERSITY 2024年第1期40卷 443-453页

作者： Bulus, Ercan Tekirdag Namik Kemal Univ Corlu Fac Engn Dept Comp Engn TR-59860 Tekirdag Turkiye

Purpose:It is to research whether steganalysis can be done with deep learning methods. Theory and Methods: 10,000 BOWS 2 and 4,000 BOSSBase 1.01 pairs were used for training, 1,000 BOSSBase 1.01 pairs forvalidation, and 5,000 BOSSBase 1.01 pairs for testing. HILL, MiPOD, S-UNIWARD and WOW steganography methods were used to information hiding. Steganalysis was performed with VGG16, Xu-Net, Ye-Net and Yeroudj-Net deep learning models. Results: At 0.2 bpp load there is not much difference between the results and not quite as expected. The best result in wow at 0.4 payload was obtained with the VGG16 (86.4%) method. However, VGG16 test times are twice as long compared to others. In this case, the Yedroudj method seems more suitable in terms of accuracyrate and processing time. Conclusion: When the test accuracy results for 0.2bpp load are examined in the study;It is seen that the lowest result is 50% (0.500) in the YEDROUDJ model with the MIPOD method, and the most efficient result is 69% (0.69) in the XU-net model for all methods. Since acceptable results could not be achieved with the VGG16 model under 0.2bpp load, it was not evaluated. In this case, the Xu-net model can be preferred to investigate whetherinformation is hidden in images with a low amount of information such as 0.2bpp. Again, when the test accuracy results for 0.4bpp load are examined in the study;It is seen that the lowest result is obtained withthe Xu-net model in the MIPOD method with 72% (0.72) and again with the Ye-Net model in the MIPOD method with 72% (0.72). On the other hand, the highest result was obtained with the VGG16 model in the WOW method with 86% (0.864) and with the VGG16 model with 84% (0.847). However, when the processing times in Table 11 are examined, it is clear that the processing time of the examinations made with the VGG16 model is almost twice as long as the others, and this is a very undesirable situation. In this case, the Yedroudj method is seen as a more useful method sin

关键词： image Steganography image Steganalysis Data Hiding deep learning Networks Cryptography

来源：评论

学校读者我要写书评

暂无评论

deep learning for rice leaf disease detection: A systematic literature review on emerging trends, methodologies and techniques

引用

Information processing in Agriculture 2025年第2期12卷 151-168页

作者： Simhadri, Chinna Gopi Kondaveeti, Hari Kishan Vatsavayi, Valli Kumari Mitra, Alakananda Ananthachari, Preethi School of Computer Science Engineering VIT-AP University Andhra Pradesh Amaravathi522237 India Andhra University Andhra Pradesh Visakhapatnam530003 India Nebraska Water Center Institute of Agriculture and Natural Resources University of Nebraska-Lincoln LincolnNE United States AI & Big Data Endicott College of International Studies Woosong University Endicott Building Daejeon Korea Republic of

Rice is an essential food crop that is cultivated in many countries. Rice leaf diseases can cause significant damage to crop cultivation, leading to reduced yields and economic losses. Traditional disease detection approaches are often time-consuming, labor-intensive, and require expertise. Automatic leaf disease detection approaches help farmers detect diseases without or with less human interference. Most of the earlier studies on rice leaf disease detection depended on image processing and machine learning techniques. image processing techniques are used to extract features from diseased leaf images, such as the color, texture, vein patterns, and shape of lesions. Machine learning techniques are used to detect diseases based on the extracted features. In contrast, deep learning techniques learn complex patterns from large datasets without explicit feature extraction techniques and are well-suited for disease detection tasks. This systematic review explores various deep learning approaches used in the literature for rice leaf disease detection, such as Transfer learning, Ensemble learning, and Hybrid approaches. This review also discusses the effectiveness of these approaches in addressing various challenges. This review discusses the details of various models and hyperparameter settings used, model fine-tuning techniques followed, and performance evaluation metrics utilized in various studies. This review also discusses the limitations of existing studies and presents future directions for further developing more robust and efficient rice leaf disease detection techniques. © 2024 The Author(s)

关键词： learning systems

来源：评论

学校读者我要写书评

暂无评论

A Hybrid Approach for Pavement Crack Detection Using Mask R-CNN and Vision Transformer Model

引用

Computers, Materials & Continua 2025年第1期82卷 561-577页

作者： Shorouq Alshawabkeh Li Wu Daojun Dong Yao Cheng Liping Li Faculty of Engineering China University of GeosciencesWuhan430074China

Detecting pavement cracks is critical for road safety and infrastructure *** methods,relying on manual inspection and basic image processing,are time-consuming and prone to *** deep-learning(DL)methods automate crack detection,but many still struggle with variable crack patterns and environmental *** study aims to address these limitations by introducing the Masker Transformer,a novel hybrid deep learning model that integrates the precise localization capabilities of Mask Region-based Convolutional Neural Network(Mask R-CNN)with the global contextual awareness of Vision Transformer(ViT).The research focuses on leveraging the strengths of both architectures to enhance segmentation accuracy and adaptability across different pavement *** evaluated the performance of theMaskerTransformer against other state-of-theartmodels such asU-Net,TransformerU-Net(TransUNet),U-NetTransformer(UNETr),SwinU-NetTransformer(Swin-UNETr),You Only Look Once version 8(YoloV8),and Mask R-CNN using two benchmark datasets:Crack500 and *** findings reveal that the MaskerTransformer significantly outperforms the existing models,achieving the highest Dice SimilarityCoefficient(DSC),precision,recall,and F1-Score across both ***,the model attained a DSC of 80.04%on Crack500 and 91.37%on deepCrack,demonstrating superior segmentation accuracy and *** high precision and recall rates further substantiate its effectiveness in real-world applications,suggesting that the Masker Transformer can serve as a robust tool for automated pavement crack detection,potentially replacing more traditional methods.

关键词： Pavement crack segmentation transportation deep learning vision transformer Mask R-CNN image segmentation

来源：评论

学校读者我要写书评

暂无评论

deep-learning-Based Strong Ground Motion Signal Prediction in real time

引用

BUILDINGS 2024年第5期14卷 1267-1267页

作者： AlHamaydeh, Mohammad Tellab, Sara Tariq, Usman Amer Univ Sharjah Coll Engn Dept Civil Engn POB 26666 Sharjah U Arab Emirates Amer Univ Sharjah Coll Engn Dept Mechatron POB 26666 Sharjah U Arab Emirates Amer Univ Sharjah Coll Engn Dept Elect Engn POB 26666 Sharjah U Arab Emirates

processing ground motion signals at early stages can be advantageous for issuing public warnings, deploying first-responder teams, and other time-sensitive measures. Multiple deep learning (DL) models are presented herein, which can predict triaxial ground motion accelerations upon processing the first-arriving 0.5 s of recorded acceleration measurements. Principal Component Analysis (PCA) and the K-means clustering algorithm were utilized to cluster 17,602 accelerograms into 3 clusters using their metadata. The accelerograms were divided into 1 million input-output pairs for training, 100,000 for validation, and 420,000 for testing. Several non-overlapping forecast horizons were explored (1, 10, 50, 100, and 200 points). Various architectures of Artificial Neural Networks (ANNs) were trained and tested, such as Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Long Short-Term Memory (LSTM) networks, and CNN-LSTMs. The utilized training methodology applied different aspects of supervised and unsupervised learning. The LSTM model demonstrated superior performance in terms of short-term prediction. A prediction horizon of 10 timesteps in the future with a Root Mean Squared Error (RMSE) value of 8.43 x 10-6 g was achieved. In other words, the LSTM model exhibited a performance improvement of 95% compared to the baseline benchmark, i.e., ANN. It is worth noting that all the considered models exhibited acceptable real-time performance (0.01 s) when running in testing mode. The CNN model demonstrated the fastest computational performance among all models. It predicts ground accelerations under 0.5 ms on an Intel Core i9-10900X CPU (10 cores). The models allow for the implementation of real-time structural control responses via intelligent seismic protection systems (e.g., magneto-rheological (MR) dampers).

关键词： Artificial Neural Networks (ANN) Convolutional Neural Networks (CNN) CNN-LSTMs deep learning (DL) Long Short-Term Memory (LSTM) networks Machine learning (ML) Recurrent Neural Networks (RNN)

来源：评论

学校读者我要写书评

暂无评论

Identification of Sign Language Alphabets on Static images Using deep learning 6

Identification of Sign Language Alphabets on Static Images U...

引用

6th IEEE International Conference on Emerging Smart Computing and Informatics, ESCI 2024

作者： Jayasingh, Bipin Bihari Mallick, Sumitra CVR College of Engineering IT Department Telangana Hyderabad India Guru Nanak Institute of Technical Campus Department of Computer Science Telangana Hyderabad India

ISBN: (纸本)9798350306613

Deaf and mute individuals often rely on sign language interpreters to facilitate communication. However, finding proficient interpreters can be challenging and costly. To address these challenges, a computerized interpreter could serve as a reliable and cost-effective alternative. Such a system would be capable of translating the sign language into plain text or audio in real-time, aiding in seamless communication. This paper integrates computer vision and employs deep learning techniques such as Convolutional Neural Networks (CNNs) to facilitate the real-time translation of sign language. These models quickly translate the user's sign into text by using computer vision to quickly record the user's sign. Through image processing, Convolutional Neural Networks facilitate the classification and prediction of various sign symbols. This integration of technology enables a seamless and instantaneous translation of sign language into textual form, promoting enhanced communication accessibility. By leveraging more amounts of data, deep learning algorithms aim to achieve high accuracy and generate predictions. The validation accuracy increased to 95.2% after 43 epochs and the validation loss dropped to 0.19. This demonstrates that the model was able to distinguish between Sign Language Recognition (SLR) and non-SLR photos with high accuracy and minimal error rates. The research intends to develop an automated system that can precisely recognize and translate sign language motions into written text by integrating deep learning and computer vision. © 2024 IEEE.

关键词： computer vision data learning deep learning gestures hearing impairments image processing motion detection neural networks sign language recognition textual translation

来源：评论

学校读者我要写书评

暂无评论

deep Reinforcement learning Model for Stock Portfolio Management Based on Data Fusion

引用

NEURAL processing LETTERS 2024年第2期56卷 108-108页

作者： Li, Haifeng Hai, Mo Cent Univ Finance & Econ Sch Informat Beijing 100081 Peoples R China

deep reinforcement learning (DRL) can be used to extract deep features that can be incorporated into reinforcement learning systems to enable improved decision-making;DRL can therefore also be used for managing stock portfolios. Traditional methods cannot fully exploit the advantages of DRL because they are generally based on real-time stock quotes, which do not have sufficient features for making comprehensive decisions. In this study, in addition to stock quotes, we introduced stock financial indices as additional stock features. Moreover, we used Markowitz mean-variance theory for determining stock correlation. A three-agent deep reinforcement learning model called Collaborative Multi-agent reinforcement learning-based stock Portfolio management System (CMPS) was designed and trained based on fused data. In CMPS, each agent was implemented with a deep Q-network to obtain the features of time-series stock data, and a self-attention network was used to combine the output of each agent. We added a risk-free asset strategy to CMPS to prevent risks and referred to this model as CMPS-Risk Free (CMPS-RF). We conducted experiments under different market conditions using the stock data of China Shanghai Stock Exchange 50 and compared our model with the state-of-the-art models. The results showed that CMPS could obtain better profits than the compared benchmark models, and CMPS-RF was able to accurately recognize the market risk and achieved the best Sharpe and Calmar ratios. The study findings are expected to aid in the development of an efficient investment-trading strategy.

关键词： deep reinforcement learning Multi-agents Quantitative trading Portfolio management

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：