检索结果-内蒙古大学图书馆

9th International Conference on Signal and image processing (ICSIP)

作者： Li, Dongrui Zhao, Zhichun Shenzhen MSU BIT Univ Fac Engn Shenzhen Peoples R China

ISBN: (纸本)9798350350920

Accurate classification and identification of vessels in remote sensing satellite imagery is critical for ocean monitoring and resource management. The ability to extract information from remote-sensing data is of paramount importance. To exploit the non-stationary characteristics of synthetic aperture radar (SAR) target, a comprehensive SAR ship recognition framework is designed by combing the second-order synchrosqueezing transform (SST), an effective non-stationary signal processing tool, with the histogram of oriented gradient (HOG) feature in this paper. Firstly, the second-order SST is performed on SAR images to describe the non-stationary characteristics of ships at different times and frequencies. Secondly, HOG features are utilized to effectively extract the non-stationary information of SAR ships and provide more discriminative input for the deep learning network. Then, the optimal ResNet model is selected as the convolutional neural network (CNN) classifier to automatically fuse the non-stationary features and abstract features of SAR ships. Experiments on two open SAR ship datasets (OpenSARShip and FUSAR-Ship) show that the proposed method achieves accurate classification and outperforms the state-of-the-art (SOTA) CNN-based methods in terms of robustness and generalization ability. The positive effect of non-stationary characteristics on SAR ship classification is verified.

关键词： Synthetic aperture radar (SAR) ship classification non-stationary signal processing time-frequency analysis (TFA) histogram of oriented gradient (HOG) convolutional neural network (CNN)

来源：评论

学校读者我要写书评

暂无评论

End-to-end Trainable Dual Fisheye image Compression 24

End-to-end Trainable Dual Fisheye Image Compression

引用

7th International Conference on Signal processing and Machine learning, SPML 2024

作者： Zhang, Xueting Hao, Jianjun Tang, Weijie Yan, Dong Chen, Xiaozhong School of Electronic and Information Engineering Shandong University of Science and Technology China Co. Ltd China Co. Ltd China

ISBN: (纸本)9798400717192

Dual-fisheye photos are the most efficient and economical way for 3D image/video display and VR applications. The photos are captured by camera with fisheye lens and stored in left and right views individually, and these two images overlap in most fields of view, thus lead to a large amount of redundant information, especially in the edges of images. Generally the two images are compressed in Jpeg or PNG individually, but this scheme ignores the correlation between left and right images. Aiming to develop a more efficient method, a jointed compression strategy by using an end-to-end trainable deep network for dual fisheye images is proposed this paper. In addition, to better study dual fisheye image compression, real scene photographs indoor and outdoor are taken to establish the data of dual fisheye image and verify the algorithm. The results show that the proposed compression algorithm improves by 11.86% in compression ratio for dual fisheye images, which is more advantageous compared to the direct input stereo image compression network. © 2024 ACM.

关键词： image compression

来源：评论

学校读者我要写书评

暂无评论

A Transfer learning Based Approach For American Sign Language Recognition Using deep Convolutional Neural Network 1

引用

7th International Conference on Intelligent Computing and Optimization, ICO 2023

作者： Islam, Aminul Habiba, Sultana Umme Mahmud, Tanjim Rahman, Habibur Sumi, Mahmuda Akter Basnin, Nanziba Hossain, Mohammad Shahadat Andersson, Karl Green University of Bangladesh Kanchon Bangladesh Khulna University of Engineering & Technology Khulna Bangladesh Rangamati Science and Technology University Rangamati Bangladesh Leeds Beckett University Leeds United Kingdom University of Chittagong Chittagong Bangladesh Luleå University of Technology Luleå Sweden

ISBN: (数字)9783031733185

ISBN: (纸本)9783031733178

American Sign Language (ASL), a visual language utilizing hand gestures, facial expressions, and body movements, remains less recognized than spoken languages, resulting in communication challenges between deaf and hearing individuals. This pioneering research paper introduces an exceptionally effective method for ASL gesture recognition through image processing and computer vision. By capturing webcam images of users signing and applying advanced algorithms, the system extracts crucial features like hand position, shape, and movement to classify signs accurately. The image processing pipeline employs techniques like background subtraction, hand detection, tracking, and feature extraction, utilizing a self-prepared dataset of around 10,000 images. This holistic approach achieves an impressive average recognition accuracy of 99.2% for 26 ASL signs in real-time. This research has the potential to greatly enhance accessibility and the quality of life for the deaf and hard-of-hearing community. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.

关键词： Visual languages

来源：评论

学校读者我要写书评

暂无评论

The Enhanced UGV Cluster Navigation Methodology Based on MARL and LSTM 5

The Enhanced UGV Cluster Navigation Methodology Based on MAR...

引用

5th Asia-Pacific Conference on image processing, Electronics and Computers, IPEC 2024

作者： Cheng, Yinjun Wei, Siwei Chen, Zexi Hubei University of Technology School of Computer and Science Wuhan430070 China CCCC Second Highway Consultants Co. Ltd. Wuhan430056 China Consultants Co. Ltd Wuhan430205 China

ISBN: (纸本)9798350374407

The main focus of this paper is to address the obstacle avoidance and path planning challenges in Unmanned Ground Vehicle (UGV) cluster navigation tasks within unknown environments of a certain scale, proposing a Multi-Agent Reinforcement learning (MARL) based method for UGV cluster navigation. The approach leverages the MATD3 algorithm, which features an architecture with two sets of Critic networks to mitigate the issue of overestimation in reinforcement learning. Additionally, an LSTM network is integrated to address the inefficiency in utilizing time-series information in sample data, guiding UGV clusters to achieve global mission objectives without relying on explicit environmental maps. Furthermore, to expedite neural network training for UGV clusters in unknown environments, the paper employs Pybullet to simulate unknown environments. Subsequently, this paper compares this algorithm with MADDPG and MATD3 methods, demonstrating that our proposed approach outperforms navigation tasks within unknown environments. UGV clusters utilizing LSTM-MATD3 achieve higher cumulative rewards and success rates, along with lower cluster collision rates. Experimental results affirm the feasibility and effectiveness of our proposed method. © 2024 IEEE.

关键词： deep reinforcement learning

来源：评论

学校读者我要写书评

暂无评论

Gan-based data augmentation to improve breast ultrasound and mammography mass classification

引用

BIOMEDICAL SIGNAL processing AND CONTROL 2024年 94卷

作者： Jimenez-Gaona, Yuliana Carrion-Figueroa, Diana Lakshminarayanan, Vasudevan Rodriguez-Alvarez, Maria Jose Univ Tecn Particular Loja Dept Quim & Ciencias Exactas San Cayetano Alto S-N CP1101608 Loja Ecuador Hosp Gen Sur Quito IESS Calle Moraspungo & Pinllopata Quito 170111 Ecuador Univ Waterloo Sch Optometry & Vis Sci Theoret & Expt Epistemol Lab Waterloo ON N2L 3G1 Canada Univ Waterloo Dept Syst Design Engn Phys & Elect & Comp Engn Waterloo ON N2L 3G1 Canada Univ Politecn Valencia Inst Instrumentac Imagen Mol I3M E-46022 Valencia Spain

Data imbalance is a common problem in breast cancer diagnosis, to address this challenge, the research explores the use of Generative Adversarial Networks (GANs) to generate synthetic medical data. Various GAN methods, including Wasserstein GAN with Gradient Penalty (WGAN-GP), Cycle GAN, Conditional GAN, and Spectral Normalization GAN (SNGAN), were tested for data augmentation in breast regions of interest (ROIs) using mammography and ultrasound databases. The study employed real, synthetic, and hybrid ROIs (128x128 pixels) to train a Resnet network for classifying as benign (B) or malignant (M) classes. The quality and diversity of the synthetic data were assessed using several metrics: Fre chet Inception Distance (FID), Kernel Inception Distance (KID), Structural Similarity Index (SSIM), Multi -Scale SSIM (MS-SSIM), Blind Reference image Spatial Quality Evaluator (BRISQUE), Naturalness image Quality Evaluator (NIQE), and Perception -based image Quality Evaluator (PIQE).Results revealed that the SNGAN model (FID = 52.89) was most effective for augmenting mammography data, while CGAN (FID = 116.03) excelled with ultrasound data. Cycle GAN and WGAN-GP, though demonstrating lower KID values, did not perform better than SNGAN and CGAN. The lower average MS-SSIM values suggested that SNGAN and CGAN produced a high diversity of synthetic images. However, lower SSIM, BRISQUE, NIQE, and PIQE values indicated poor quality in both real and synthetic images. Classification results showed high accuracy without data augmentation in both US (93.1 %B/94.9 %M) and mammography (80.9 %B/76.9 %M). The research concludes that preprocessing and characterizing ROIs by abnormality type is crucial to generate diverse synthetic data and improve accuracy in the classification process using combined GANs and CNN models.

关键词： Breast cancer Data augmentation deep learning algorithms Generative Adversarial Networks (GAN) Mammography Ultrasound

来源：评论

学校读者我要写书评

暂无评论

Online learning of image Recognition Task Offloading Policies over Wireless Links

Online Learning of Image Recognition Task Offloading Policie...

引用

IEEE International Conference on Communications (IEEE ICC)

作者： Koutsopoulos, Iordanis Athens Univ Econ & Business Dept Informat Athens Greece

ISBN: (纸本)9781538674628

We study the problem of online learning of optimal offloading policies for image processing tasks, for minimizing a cost that is weighted sum of transmit energy and object recognition error rate. A mobile node generates image processing tasks that involve object recognition. There exist three options: (i) transmit the image to a remote server for processing with a deep-learning (DL) model, (ii) process locally with a simpler model, (iii) apply a lightweight, error-prone technique for object detection, and if objects are detected, then send image to the server. The proper offloading decision requires knowledge of the transmit energy cost and object recognition error rate for each option. However, these processes are non-stationary due to unpredictable object occurrence, mobility and propagation dynamics, and they depend on the object inference result which is unknown at decision time. We cast the problem as an adversarial multi-armed bandit, in which the EXP3 algorithm achieves sublinear regret. For the constrained problem, we propose an algorithm that extends EXP3 and achieves good regret in the objective and constraint, thus asymptotically learning the optimal static randomized offloading policy, while satisfying the error constraint. Performance is validated via numerical experiments informed by real-life object recognition measurements and models.

关键词： Object recognition

来源：评论

学校读者我要写书评

暂无评论

Neural Network based Pesticides Recommender System for Leaf Disease Detection 15

Neural Network based Pesticides Recommender System for Leaf ...

引用

15th International Conference on Advances in Computing, Control, and Telecommunication Technologies, ACT 2024

作者： Ramesh, M. Haswitha, Juluru Potluri, Syam Pratap Vijaya, Pentapati Sri Dept of Information Technology VRSEC Kannuru Vijayawada India

ISBN: (纸本)9798331300579

The majority of tropical and subtropical nations in the world eat rice as their main meal. This involves hectare- sized paddy fields, whose upkeep and care becomes a tiresome Undertaking for the farmers. The caregivers are unable to recognize specific problems and cannot do the time-consuming task of crop care in such a short amount of time. Brown spot, Leaf Blast and Tungro are the diseases that affect rice leaves most frequently. Therefore, disease detection in leaves is a crucial topic that offers numerous advantages for keeping an eye on vast fields of crops. Rice leaf disease damages the green layer of the leaves, which might have an impact on yield and quality. It will take time to speak with doctors and professionals to detect sickness. Plant diseases can be discovered via image processing. Steps including image acquisition, image pre-processing, picture segmentation, feature extraction, and classification are involved in disease detection in order to obtain crucial data for more research. It will be more effective to build a system for classifying and identifying diseases. To address this issue, deep learning models are developed using convolutional neural networks to pinpoint problematic areas. There are loaded image data sets of many common leaf diseases. By comparing data that was established using features and segments, our farmers can classify the illness and administer pesticides. Society will gain from automating the disease diagnosis process, and crop production will rise. © Grenze Scientific Society, 2024.

关键词： image acquisition

来源：评论

学校读者我要写书评

暂无评论

Satellite image–Based Ecosystem Monitoring with Sustainable Agriculture Analysis Using Machine learning Model

引用

Remote Sensing in Earth Systems Sciences 2024年第4期7卷 764-773页

作者： Mulakaledu, Ajjanna Swathi, Baswaraju Jadhav, Makarand Mohan Shukri, Shakeerah Mohd Bakka, Vinod Jangir, Pradeep Department of English Koneru Lakshmaiah Education Foundation AP Vaddeswaram India Department of Computer Science and Engineering (Data Science) New Horizon College of Engineering Bengaluru India Department of Electronics and Telecommunication NBN Sinhgad Technical Institutes Campus Pune 411041 India Management and Science University Selangor Shah Alam Malaysia Koneru Lakshmaiah Education Foundation AP Vaddeswaram India Department of Biosciences Saveetha School of Engineering Saveetha Institute of Medical and Technical Sciences Chennai 602 105 India Applied Science Research Center Applied Science Private University Amman Jordan

Understanding the variations in soil fertility and crop growth across time and geography is crucial for understanding the agricultural environment. Satellite and unmanned aerial remote sensing are the two main types of remote sensing methods used in agroecosystem monitoring. Through the collection of remote sensing photos, it is able to monitor as well as control agro-ecosystem environment in real-time. Spatial choice constraints are lessened by high-decision satellites, and satellite imagery processing is improved by use of artificial intelligence (AI) and machine learning methods, which enables computerised crop identification, disease diagnosis, and yield calculation. This research proposes novel technique in satellite image–based sustainable agriculture analysis with ecosystem monitoring using machine learning techniques. Here, the input is collected as satellite image and processed for noise removal and normalisation. Then, this image feature has been selected for analysis of crop growth using Gaussian belief kernel component analysis and classified using recursive Bayes probabilistic vector neural networks. Experimental analysis has been carried out in terms of training accuracy, precision, sensitivity, F-1 score, and AUC for various satellite image dataset. Proposed technique attained precision of 94%, sensitivity of 96%, AUC of 93%, training accuracy of 97%, and F1-score of 95%. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2024.

关键词： Crop Growth Analysis Ecosystem Monitoring Machine learning Techniques Satellite image Sustainable Agriculture Analysis

来源：评论

学校读者我要写书评

暂无评论

A Method for Surface Wave Suppression Based on Multimodal Constraints Neural Networks 4

A Method for Surface Wave Suppression Based on Multimodal Co...

引用

4th International Meeting for Applied Geoscience and Energy, image 2024

作者： Yan, Yixuan Wang, Mengxiu Cui, Dong Wang, Chunming Hou, Sian Research Institute of Petroleum Exploration and Development CNPC China

Surface waves represent low-frequency regular interference waves in onshore seismic exploration, exerting a significant influence on the seismic data processing quality. Despite the classic method for surface wave suppression such as singular value decomposition (Bereford and Rango, 1989;Lu, 2006), deep learning has been playing an important role in denoising (Sun and et al., 2019;Yu and et al., 2019;Tian and et al., 2020;Alkhalifah and et al., 2021;Liu and et al., 2022;Wang and et al., 2022;). This paper introduces a surface wave suppression technique based on the deep Residual Fourier Neural Operator (DRFNO). The approach integrates multimodal constraints with an intelligent data-driven methodology, utilizing neural network operations in both the time-space and frequency-wavenumber domains to jointly suppress surface waves and monochromatic noise in pre-stack seismic data. Testing on simulated and actual data showcases the effectiveness of the DRFNO in efficiently eliminating surface waves and reconstructing meaningful signals. Comparative results with commercial software demonstrate the superior performance of the deep neural network method in suppressing surface waves and monochromatic noise. © 2024 Society of Exploration Geophysicists and the American Association of Petroleum Geologists.

关键词： deep neural networks

来源：评论

学校读者我要写书评

暂无评论

Automatic non-destructive UAV-based structural health monitoring of steel container cranes

引用

APPLIED GEOMATICS 2024年第1期16卷 125-145页

作者： Lopez, Vanessa De Arriba Maboudi, Mehdi Achanccaray, Pedro Gerke, Markus Tech Univ Carolo Wilhelmina Braunschweig Inst Geodesy & Photogrammetry Bienroder Weg 81 D-38106 Braunschweig Niedersachsen Germany

Container cranes are of key importance for maritime cargo transportation. The uninterrupted and all-day operation of these container cranes, which directly affects the efficiency of the port, necessitates the continuous inspection of these massive hoisting steel structures. Due to the large size of cranes, the current manual inspections performed by expert climbers are costly, risky, and time-consuming. This motivates further investigations on automated non-destructive approaches for the remote inspection of fatigue-prone parts of cranes. In this paper, we investigate the effectiveness of color space-based and deep learning-based approaches for separating the foreground crane parts from the whole image. Subsequently, three different ML-based algorithms (k-Nearest Neighbors, Random Forest, and Naive Bayes) are employed to detect the rust and repainting areas from detected foreground parts of the crane body. Qualitative and quantitative comparisons of the results of these approaches were conducted. While quantitative evaluation of pixel-based analysis reveals the superiority of the k-Nearest Neighbors algorithm in our experiments, the potential of Random Forest and Naive Bayes for region-based analysis of the defect is highlighted.

关键词： Structural health monitoring Crane inspection image processing UAV Defects Machine learning deep learning

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：