检索结果-内蒙古大学图书馆

2nd International Conference on Foundation and Large Language Models, FLLM 2024

作者： Ferdi, Abdesselam Nakib, Amir Constantine 1 - Frères Mentouri University Laboratoire LISSI Université Paris Est Créteil Signal Processing Laboratory Vitry Sur Seine94400 France Université Paris Est Créteil Laboratoire LISSI Vitry Sur Seine94400 France

ISBN: (纸本)9798350354799

Object detection remains a critical challenge with extensive real-time applications, including autonomous vehicles, medical imaging, and surveillance systems. The field has experienced significant progress, particularly with the advent of state-of-the-art detectors employing convolutional neural network architectures. Among these, the You Only Look Once (YOLO) framework has emerged as a benchmark, excelling in balancing detection accuracy and real-time performance. Nevertheless, the intrinsic linearity of conventional convolution operations constrains the network's capacity to model complex and hierarchical data representations. In this work, we address this critical limitation by proposing a novel nonlinear operation termed quadratic convolution. Unlike standard linear convolutions, quadratic convolution involves squaring the input image elements within the convolution process, thereby augmenting the representational power of the feature maps. We incorporated this quadratic convolution into the latest YOLOv8 detector architecture to assess its effectiveness. Experiments conducted on the widely recognized MS COCO dataset indicate that our approach yields significant improvements in detection performance of the standard YOLOv8 detector. These findings underscore the potential of quadratic convolution to enhance object detection tasks, offering a promising direction for future advancements in deep learning architecture design. © 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

deep learning Segmentation Models Evaluation for Deforestation Monitoring Embedded Systems 4

Deep Learning Segmentation Models Evaluation for Deforestati...

引用

4th International Conference on Intelligent Cybernetics Technology and Applications, ICICyTA 2024

作者： Careli, Álvaro S. Boas, Evandro C. Vilas Teixeira, Eduardo H. Silva, Elaine C. C. Aquino, Guilherme P. Figueiredo, Felipe A. P. National Institute of Telecommunication Inatel Cyber Security and IoT Laboratory Cs&i Lab. Santa Rita do Sapucaí Brazil National Institute of Telecommunication Inatel IoT Research Group Laboratory Santa Rita do Sapucaí Brazil National Institute of Telecommunication Inatel Wireless and Artificial Intelligence Laboratory Wai Lab. Santa Rita do Sapucaí Brazil

ISBN: (纸本)9798331506490

This work evaluates deep learning segmentation models to propose a deforestation monitoring embedded system. The approach stands for environmental monitoring using remote sensing imagery, edge computing, and a deep learning segmentation model. Thus, the performance of you only look once architecture version 8 (YOLOv8) and Mask Region-based convolutional neural networks (Mask R-CNN) embedded in Raspberry Pi Model 4 regarding Intersection over Union (IoU), mean Average Precision (mAP), and time per image processing metrics is compared. The models are combined with a pixel-based algorithm that analyzes the temporal segmented images to define their forest area percentage for deforestation monitoring and detection. The results demonstrate YOLOv8x model achieved an IoU of 0.762, with a time per image of 0.4777 seconds, while Mask R-CNN R101 FPN 3x obtained an IoU of 0.763, with a time per image of 0.2669 seconds. The average times for YOLOv8 ranged from 0.0434 to 0.4777 seconds, and for Mask R-CNN from 0.1969 to 0.2669 seconds. Finally, this work proposes evaluating the model's performance when working with generative AI models Dall-e, Craiyon, and Tess-AI to create a synthetic dataset to augment the initial one with synthetic samples and improve the model's training with a large dataset. The Dall-e has been shown to outperform the others regarding the IoU metric, which was suggested to augment datasets with synthetic samples. © 2024 IEEE.

关键词： Deforestation Environmental Monitoring Mask R-CNN Segmentation YOLOv8

来源：评论

学校读者我要写书评

暂无评论

AIS Data-Guided Geolocation Correction Method for Low-Orbit Satellite Remote Sensing imagery

引用

IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING 2024年 17卷 18703-18726页

作者： Wu, Haoyang Huang, Zishuo Hu, Qinyou Ran, Xin Mei, Qiang Shanghai Maritime Univ Merchant Marine Coll Shanghai 201306 Peoples R China Jimei Univ Nav Coll Xiamen Peoples R China

Accurate geolocation of maritime objects in satellite imagery is challenging due to geometric distortions, atmospheric conditions, and sensor inaccuracies in low-Earth orbit satellites. This study presents a novel automatic identification system (AIS) data-guided geolocation correction method that integrates real-time AIS data with satellite imagery to rectify geolocation errors. The approach utilizes the GeoAISNet neural network, which enhances positional accuracy without relying on ground control points. By incorporating a modified YOLOv8 architecture with orientation parameters and the convolutional block attention module, detection performance improved significantly, achieving precision, recall, and F1 scores of 91.82%, 89.56%, and 90.67%, respectively. Ablation studies demonstrated the crucial impact of feature integration and attention mechanisms. Results indicate a mean average precision of 89%, with general cargo ships achieving 99.9% AP50. Localization accuracy saw a notable improvement, with root-mean-squared error reduced from 12 to 3 m, and layer normalization further enhanced stability, increasing precision, recall, and F1 scores to 94.23%, 92.67%, and 93.44%, respectively. The use of differential AIS data decreased maximum positional errors by 30%, achieving errors around 2 m. Computational efficiency was also enhanced, with processing time reduced from 2 to 0.5 s per image. This method effectively addresses oil spills and non-AIS vessel detection, expanding maritime surveillance capabilities. The global training dataset, validated with data from the South China Sea, ensures the method's applicability across diverse conditions.

关键词： Satellites Satellite images Accuracy Surveillance Remote sensing Machine learning real-time systems Geology Marine vehicles Environmental monitoring Automatic identification system (AIS) geolocation correction maritime surveillance neural networks satellite imagery

来源：评论

学校读者我要写书评

暂无评论

Spatio-temporal enhanced anomaly detection in FRP bridge monitoring using MPCA, biGRU, and attention mechanisms

引用

JOURNAL OF CIVIL STRUCTURAL HEALTH MONITORING 2025年 1-19页

作者： Dibiantara, Dimas Pustaka Adha, Augusta Darmawan, Muhammad Sigit Imjai, Thanongsak Russell, Justin Laory, Irwanda Univ Warwick Sch Engn Lib Rd Coventry CV4 7AL West Midlands England Inst Teknol Sepuluh Nopember Dept Civil Infrastruct Engn Surabaya 60116 East Java Indonesia Walailak Univ Sch Engn & Technol 222 Thaiburi Nakhon Si Thammarat 80160 Thailand

Accurate structural behavior interpretation via finite element models is often disrupted by uncertainties, while data-driven approaches can struggle with long datasets, complex fluctuations, and the omission of essential spatio-temporal features. Additionally, these methods are limited by their reliance on interpolative predictions. This paper introduces a novel, model-free approach that integrates Moving Principal Component Analysis (MPCA), bidirectional gated recurrent units (biGRU), and attention mechanisms (AM) within an encoder-decoder (ED) architecture. MPCA reduces dimensional complexity, extracts spatial features, and consolidates them into new time-series data for subsequent analysis. The biGRU module captures past and future dependencies, while AM emphasizes most relevant information. Validated on a full-scale pedestrian bridge dataset, the presented MPCA-biGRU-AM model converges 19% faster than MPCA-GRU and reduces anomaly detection lag by 46-78%. Although its per-step processing time (8 ms) slightly exceeds that of MPCA-GRU (3 ms), the model demonstrates greater robustness across diverse damage scenarios. These results highlight its potential for real-time structural health monitoring by effectively capturing spatio-temporal patterns with computational efficiency.

关键词： Spatio-temporal Anomaly detection Extrapolative prediction Structural behavior deep learning

来源：评论

学校读者我要写书评

暂无评论

DcsNet: a real-time deep network for crack segmentation

引用

SIGNAL image AND VIDEO processing 2022年第4期16卷 911-919页

作者： Pang, Jie Zhang, Hua Zhao, Hao Li, Linjing Southwest Univ Sci & Technol Sch Informat Engn Mianyang 621000 Sichuan Peoples R China Special Environm Robot Technol Key Lab Sichuan Pr Mianyang 621000 Sichuan Peoples R China Univ Sci & Technol China Dept Automat Hefei 230000 Peoples R China

Detecting cracks are a great significance for the maintenance of the man-made buildings, and deep learning methods such as semantic segmentation have greatly boosted this process in recent years. However, the existing crack segmentation methods often sacrifice feature resolution to achieve real-time inference speed which leads to poor performance, or use complex network module to improve the accuracy which leads to lower inference speed. In this paper, we propose a novel deep Crack Segmentation Network (DcsNet) that incorporates two feature extraction branches to achieve the balance of speed and accuracy. We first design a morphology branch (MB) to preserve the morphology information of scale invariance that consists of a lightweight convolution network, a pyramid pooling module (PPM), and an attention module (CSA). Meanwhile, a shallow detail branch (DB) with a small stride is constructed to supplement detailed information. Extensive experiments are conducted on five challenging datasets (Crack500, deepcrack, Gaps384, Structure, and Damcrack), and the results demonstrated that the proposed network achieves a good trade-off between accuracy and inference speed and outperforms state-of-the-art methods.

关键词： Crack segmentation real-time Morphology information Detailed information

来源：评论

学校读者我要写书评

暂无评论

TransFiLM: An Efficient and Lightweight Audio Enhancement Network for Low-Cost Wearable Sensors 21

TransFiLM: An Efficient and Lightweight Audio Enhancement Ne...

引用

21st IEEE International Conference on Mobile Ad-Hoc and Smart Systems (MASS)

作者： Chen, Shilin Weng, Jianhao Hong, Shicong He, Yanbo Zou, Yongpan Wu, Kaishun Shenzhen Univ Coll Comp Sci & Software Engn Shenzhen Peoples R China Hong Kong Univ Sci & Technol Guangzhou Informat Hub Guangzhou Peoples R China

ISBN: (纸本)9798350363999;9798350364002

Wearable devices such as headphones are increasingly popular in people's lives, and there is an increasing focus on how to achieve continuous and reliable information input using these devices. However, due to constraints in computing power, low power consumption, and low operating frequencies, such devices often record and transmit signals at lower sampling rates, the resultant lower-quality signals often have catastrophic implications for system performance. Efficient real-time conversion of low-resolution speech signals to full-resolution high-quality signals using low-cost wearable sensors on edge devices is a challenging research endeavor. To address this, this paper designs TransFiLM, a mobile deep learning network. It allows users to obtain full-resolution high-quality audio signals using low-cost wearable sensors on edge devices. TransFiLM integrates residual learning and super-resolution networks and employs effective signal processing strategies to achieve audio upscaling and noise reduction, significantly improving audio quality. We implement a prototype on commercial devices and conduct a series of experiments to evaluate its performance. Using signal-to-noise ratio (SNR) and log-spectral distance (LSD) as evaluation metrics, TransFiLM exhibits superior performance compared to other time-domain methods in cross-user, cross-corpus, and cross-noise environment testing. Additionally, our TransFiLM network handles 8192 samples with a response time of 181 ms, which meets the requirement to run in real-time on edge devices.

关键词： Wearable Sensors Audio Super-Resolution Mobile deep-learning

来源：评论

学校读者我要写书评

暂无评论

Intelligent elastic resource allocation strategy based on supervised learning 9

Intelligent elastic resource allocation strategy based on su...

引用

9th International Symposium on Computer and Information processing Technology, ISCIPT 2024

作者： Li, Lina Hu, Chuncheng Wang, Shuxin Lv, Junan Xu, Yiming Li, Nianfeng Changchun University College of Computer Science and Technology Changchun China

ISBN: (纸本)9798350388404

In streaming computing systems, flexible resource allocation for time-varying data is the key to ensure application and system performance. Traditional resource allocation methods and existing intelligent methods have the problems of limited practicability and high training costs. In order to minimize application delay, this paper proposes an intelligent elastic resource allocation strategy based on supervised learning. The strategy uses the deep neural network as the strategy model and conducts model training through supervised learning. After obtaining the basic system status and application load, it automatically performs near-optimal application resource allocation. We conducted policy performance evaluation on three real datasets. The experimental results show that compared with reinforcement learning strategy, our strategy has better performance in terms of delay violation rate and throughput, and is close to the performance of the optimal strategy. © 2024 IEEE.

关键词： Self-supervised learning

来源：评论

学校读者我要写书评

暂无评论

Coarse-Super-Resolution-Fine Network (CoSF-Net): A Unified End-to-End Neural Network for 4D-MRI With Simultaneous Motion Estimation and Super-Resolution

引用

IEEE TRANSACTIONS ON MEDICAL IMAGING 2024年第1期43卷 162-174页

作者： Zhi, Shaohua Wang, Yinghui Xiao, Haonan Bai, Ti Li, Bing Tang, Yunsong Liu, Chenyang Li, Wen Li, Tian Ge, Hong Cai, Jing Hong Kong Polytech Univ Dept Hlth Technol & Informat Hong Kong Peoples R China Univ Texas Southwestern Med Ctr Dept Radiat Oncol Med Artificial Intelligence & Automa MAIA Lab Dallas TX 75239 USA Zhengzhou Univ Dept Radiat Oncol Zhengzhou 450008 Peoples R China Zhengzhou Univ Affiliated Canc Hosp Zhengzhou 450008 Peoples R China Henan Canc Hosp Zhengzhou 450008 Peoples R China

Four-dimensional magnetic resonance imaging (4D-MRI) is an emerging technique for tumor motion management in image-guided radiation therapy (IGRT). However, current 4D-MRI suffers from low spatial resolution and strong motion artifacts owing to the long acquisition time and patients' respiratory variations. If not managed properly, these limitations can adversely affect treatment planning and delivery in IGRT. In this study, we developed a novel deep learning framework called the coarse-super-resolution-fine network (CoSF-Net) to achieve simultaneous motion estimation and super-resolution within a unified model. We designed CoSF-Net by fully excavating the inherent properties of 4D-MRI, with consideration of limited and imperfectly matched training datasets. We conducted extensive experiments on multiple real patient datasets to assess the feasibility and robustness of the developed network. Compared with existing networks and three state-of-the-art conventional algorithms, CoSF-Net not only accurately estimated the deformable vector fields between the respiratory phases of 4D-MRI but also simultaneously improved the spatial resolution of 4D-MRI, enhancing anatomical features and producing 4D-MR images with high spatiotemporal resolution.

关键词： Coarse-to-fine registration deep learning four-dimensional magnetic resonance imaging super-resolution

来源：评论

学校读者我要写书评

暂无评论

deep learning and Optical Character Recognition-Based Automatic UIDAI Details Extraction System

Deep Learning and Optical Character Recognition-Based Automa...

引用

International Conference on Advances and Applications of Artificial Intelligence and Machine learning, ICAAAIML 2022

作者： Kulkarni, Sayali Ramesh Sadakale, Ranjit College of Engineering Pune India

ISBN: (纸本)9789819959730

The Automatic UIDAI Details Extraction System extracts details from Aadhaar cards and stores them into a document. The system can be utilized in banking, in government agencies, at vaccination centers. The presented system uses deep learning to detect the name and number on Aadhaar cards and recognize the text in real-time using optical character recognition (OCR). Three sub-blocks make up the proposed system: image acquisition, name and number detection, and OCR. For image Acquisition, a setup employing a webcam with a reasonable quality was devised. This approach presents the extraction of the name and number from an image using a MobileNet V1 architecture based on Single-Shot Detector (SSD). EasyOCR is utilized for text recognition. The system was deployed on the NVIDIA Jetson Nano hardware platform. The System’s real-time performance has been optimized by adjusting these three sub-block parameters. The system provides accuracy of more than 90%. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd 2023.

关键词： Optical character recognition

来源：评论

学校读者我要写书评

暂无评论

Enhanced Lung Cancer Detection Using CNNs for Accurate and Efficient CT image Classification 1

Enhanced Lung Cancer Detection Using CNNs for Accurate and E...

引用

1st International Conference on Computing, Sciences and Communications, ICCSC 2024

作者： Bdair, Mohammad Ruby, E D Kanmani Juyal, Prachi Taseen, Rakheeba Abirami, A. Soujanya, A. School of Architecture Computing & Engineering University of East London London United Kingdom Vel Tech Rangarajan Dr Sagunthala R&D Institute of Science and Technology Department of ECE Tamil Nadu Chennai India Uttarakhand Dehradun India School of computer science and Engineering Presidency university Bangalore India Sona College of Technology Department of Mathematics Tamil Nadu Salem India Ibrahimpatnam India

ISBN: (纸本)9798350353648

Lung cancer must be detected as early as possible, but it may be difficult with existing methods since they often depend on human judgment and outdated image processing. These techniques take a lot of time and are prone to errors, which produces uneven outcomes. The research provides a unique deep learning-based method for classifying lung nodules from CT images using convolutional neural networks (CNNs). By automating feature extraction and categorization, the suggested technique greatly increases diagnostic precision. The result shows superior performance over existing systems is demonstrated by evaluation measures such as AUC-ROC, accuracy, sensitivity, and specificity. With a training duration of 12 hours, an AUC-ROC of 0.95, accuracy of 92%, sensitivity of 94%, specificity of 90%, and inference speed of 0.5 seconds per scan, the results demonstrate notable improvements in diagnostic efficiency and accuracy. deep learning has the potential to revolutionize the diagnosis of lung cancer and enhance patient care, as demonstrated by the study. © 2024 IEEE.

关键词： Lung cancer

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：