检索结果-内蒙古大学图书馆

4th IEEE International conference on Power, Electronics and Computer Applications, ICPECA 2024

作者： Fang, Quanrong Investment Co. Ltd Beijing China

ISBN: (纸本)9798350359589

In various fields such as medical imaging, object detection, and video surveillance, multi view natural language query systems utilize image data to provide a more comprehensive perspective, allowing users to intuitively query and obtain information. Due to the lack of a deep understanding of natural language in the hard coded matching rule method, the query results do not match the user's intentions and are difficult to meet practical application needs. Therefore, this article introduces machine vision algorithms for optimization and improvement. This article first discusses the system architecture of four modules: data input and preprocessing, visual feature extraction, natural language understanding and matching, and result generation and feedback. Then, the application of machine vision technology in the system was analyzed using two calculation formulas: grayscale conversion and binarization, and natural language processing technology was briefly discussed. Subsequently, a context understanding module was added to construct a multi view natural language query system based on machine vision. Finally, two sets of simulation experiments were conducted to draw the following conclusion: compared with traditional methods, the overall average improvement in image recognition accuracy indicators is about 14.3%, while the overall average improvement in response speed indicators is about 26.5%. This research system can effectively process images from different perspectives and match them with natural language queries. © 2024 IEEE.

关键词： machine vision technology multi view processing natural language query system design

来源：评论

学校读者我要写书评

暂无评论

Keyword-Based Diverse image Retrieval by Semantics-aware Contrastive Learning and Transformer 23

Keyword-Based Diverse Image Retrieval by Semantics-aware Con...

引用

46th International ACM SIGIR conference on Research and Development in Information Retrieval (SIGIR)

作者： Zhao, Minyi Wang, Jinpeng Liao, Dongliang Wang, Yiru Duan, Huanzhong Zhou, Shuigeng Fudan Univ Sch Comp Sci Shanghai Peoples R China Tsinghua Univ Tsinghua Shenzhen Intl Grad Sch Shenzhen Peoples R China Tencent Inc Wechat Grp Guangzhou Peoples R China

ISBN: (纸本)9781450394086

In addition to relevance, diversity is an important yet less studied performance metric of cross-modal image retrieval systems, which is critical to user experience. Existing solutions for diversity-aware image retrieval either explicitly post-process the raw retrieval results from standard retrieval systems or try to learn multi-vector representations of images to represent their diverse semantics. However, neither of them is good enough to balance relevance and diversity. On the one hand, standard retrieval systems are usually biased to common semantics and seldom exploit diversity-aware regularization in training, which makes it difficult to promote diversity by post-processing. On the other hand, multi-vector representation methods are not guaranteed to learn robust multiple projections. As a result, irrelevant images and images of rare or unique semantics may be projected inappropriately, which degrades the relevance and diversity of the results generated by some typical algorithms like top-k To cope with these problems, this paper presents a new method called CoLT that tries to generate much more representative and robust representations for accurately classifying images. Specifically, CoLT first extracts semantics-aware image features by enhancing the preliminary representations of an existing one-toone cross-modal system with semantics-aware contrastive learning. Then, a transformer-based token classifier is developed to subsume all the features into their corresponding categories. Finally, a post-processing algorithm is designed to retrieve images from each category to form the final retrieval result. Extensive experiments on two real-world datasets Div400 and Div150Cred show that CoLT can effectively boost diversity, and outperforms the existing methods as a whole (with a higher F1 score).

关键词： Cross-modal retrieval Keyword-based image retrieval Diversification retrieval Transformer

来源：评论

学校读者我要写书评

暂无评论

5th International conference on 3D Imaging Technologies—Multidimensional Signal processing and Deep Learning, 3DIT-MSP and DL 2023

5th International Conference on 3D Imaging Technologies—Mul...

引用

5th International conference on 3D Imaging Technologies—Multidimensional Signal processing and Deep Learning, 3DIT-MSP and DL 2023

ISBN: (纸本)9789819751808

The proceedings contain 59 papers. The special focus in this conference is on 3D Imaging Technologies—Multidimensional Signal processing and Deep Learning. The topics include: Digital Twin Technology Approach Based on the Hierarchical IDP Tensor Decomposition;modeling Technology for Complex Dynamic Operating Environment of Power Grid Based on Digital Twins;design of a Digital Twin Platform Based on Distributed Computing and Resource Optimization algorithms;deep Learning Network Optimization Combining 3D Imaging and Multidimensional Signal processing;time Series Prediction Application of Deep Learning in Multidimensional Signal processing;improving Abstractive Summarization with Graph Sequence Model;advancing Semantic Segmentation and Interpretation of 3D images Through Integrated Deep Learning and Natural Language processing Techniques;children’s Toy Product Design Based on Augmented Reality Technology;based on the Neural Network Classification of Human Behavior Research;development of an Independent Adversarial Sample Detection Model, Based on image Features;Visualization and Analysis of CNN Adversarial Training;video Violence Detection Method Based on Multi-Feature and Graph Convolutional Network;Enhancing Realized Volatility Prediction: An Exploration into LightGBM Baseline Models;terminal Anomaly Discovery Technology Based on Service Behavior Deviation;An Efficient CRNN Model with the Multi-scale Feature Fusion for Text Recognition from Chinese Medical Reports;advancing Precision Drug Screening: Integrating Imaging Technology and Artificial Intelligence for Novel Models;a Short Video Retrieval Method Based on Spatio-Temporal Feature Fusion;spatial Data Transformation and Fusion Expression Based on Geometric Algebra;EM2-YOLO: Lightweight Remote Sensing image Detection;Tower-Type Detection of UAV Aerial image Based on YOLOV5 Network Model.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Survey of Text Detection algorithms in images Based on Deep Learning 4

A Survey of Text Detection Algorithms in Images Based on Dee...

引用

4th International conference on Natural Language processing, ICNLP 2022

作者： Li, Linna Hu, Cuicui Liu, Ying Xi'an University of Posts and Telecommunications Network Public Opinion Monitoring and Analysis Center Shaanxi Xi'an710121 China Xi'an University of Posts and Telecommunications College Key Laboratory of Electronic Information Application Technology for Scene Investigation Ministry of Public Security Center for Image and Information Processing Shaanxi Xi'an710121 China

ISBN: (数字)9781665495448

ISBN: (纸本)9781665495448

Nowadays, text detection has infiltrated various industries like banking, education, criminal investigation, network public opinion, and more. However, the traditional way of text detection is largely dependent on the characteristics of manual design, which is time-consuming, laborious and inaccuracy. With the emergence of deep learning technology, text detection has already been transformed from text detection in static documents to more practical text detection in natural scenes. To address such challenges as complex background, improper acquisition, and multilingual text in natural scene images, there have been various improved neural network structures constructed. Especially, the performance of text detection in various scenes has been significantly improved. In learning in different scenes are summarized in detail. According to the different text objects in detection, the existing methods are classified into two categories as top-down and bottom-up, and the logic structure, advantages and disadvantages of various algorithms are summarized. In addition, the commonly used data sets and performance evaluation indexes are analyzed and explained, the practical application of text detection in different fields is introduced, and the future trend of research and development are indicated. © 2022 IEEE.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

Enhanced Navigation Based Obstacle Avoidance and Target Tracking with Raspberry-Pi for Mobile Robots 6th

Enhanced Navigation Based Obstacle Avoidance and Target Trac...

引用

6th conference on Computing systems and Applications

作者： Belaidi, Hadjira Demim, Fethi Mezhary, Adel Rouigueb, Abdenebi Yahoui, Sofiane Nemra, Abdelkrim Univ MHamed Bougara Boumerd Signals & Syst Lab Inst Elect Elect Engn Boumerdes Algeria Ecole Mil Polytech Lab Guidance & Nav Algiers Algeria Ecole Mil Polytech Lab Artificial Intelligence & Virtual Real Algiers Algeria

ISBN: (纸本)9783031718472;9783031718489

Our study focuses on developing a self-navigating mobile robot using advanced computer vision techniques and a microprocessor. The innovative integration of these technologies results in an advanced robotic system. Our main objective is to create a seamless navigation system from a starting point to a predetermined target, with real-time obstacle avoidance and dynamic decision-making capabilities. The robot efficiently detects and adjusts its trajectory to avoid obstacles by continuously scanning its surroundings with computer vision. Precise obstacle identification is ensured through intricate embedded algorithms, enabling secure and collision-free navigation. The robot autonomously halts upon reaching the target, and if the target is hidden, it initiates comprehensive environmental exploration. Powered by the Raspberry Pi, our system offers a robust and cost-effective solution for mobile robot navigation. This strategic amalgamation of technologies greatly enhances the robot's operational prowess, enabling adept navigation and focused target pursuit in complex environments.

关键词： Autonomous navigation mobile robot computer vision system collision avoidance image processing and sensor fusion

来源：评论

学校读者我要写书评

暂无评论

An Improved Neural Classification Logic to Predict Kidney Disease Identification Using Medical image processing Principles

An Improved Neural Classification Logic to Predict Kidney Di...

引用

2024 International conference on Innovative Computing, Intelligent Communication and Smart Electrical systems, ICSES 2024

作者： Balaji, C.G. Balachandran, G. Imran, Md. Ujwala, S.Vamsi Edlabadkar, Ajinkya P. Fawareh, Hamed Department of Digital and Telecom Management Lavale Maharashtra Pune412115 India Jeppiaar Engineering College Department of Ece Rajiv Gandhi salai Chennai119 India Qis College of Engineering and Technology Department of S&h Pondur Road Vengamukkapalem Ongole Prakasam AndhraPradesh PrakasamDistrict523272 India Qis College of Engineering and Technology Department of Mba Pondur Road Vengamukkapalem Ongole Prakasam Andhra Pradesh PrakasamDistrict523272 India Yeshwantrao Chavan College of Engineering Department of Mechanical Engineering Nagpur India College of Information Technology Zarqa University Department of Cyber Security zarqa Jordan University of Business and Technology Jeddah21448 Saudi Arabia

ISBN: (纸本)9798331543617

Kidney disease is a significant global health concern, requiring accurate and early diagnosis to prevent progression and ensure timely treatment. This study proposes an improved neural classification logic combining an AutoEncoder and an Artificial Neural Network (ANN) to predict kidney disease using medical image processing techniques. A dataset comprising kidney images from CT, MRI, and ultrasound was utilized to train and test the model. The AutoEncoder was employed for unsupervised feature extraction, reducing dimensionality while retaining key information related to kidney abnormalities. The ANN then classified these features into healthy or diseased categories. image preprocessing techniques such as noise reduction, resizing, normalization, and contrast enhancement were applied to ensure data consistency. Additionally, advanced segmentation algorithms like U-Net were used to isolate the kidney regions from the background, focusing on the Regions of Interest (ROI). Data augmentation was applied to improve generalization. The proposed model achieved a classification accuracy of 94.8%, outperforming baseline models such as CNN (92.1 %) and SVM (88.7%). The AutoEncoder + ANN combination demonstrated superior precision, recall, and Fl-score, with an AUC-ROC of 97.4%. The results highlight the model's robustness and potential for practical applications in the early detection of kidney disease. © 2024 IEEE.

关键词： Noise abatement

来源：评论

学校读者我要写书评

暂无评论

image Enhancement by Various Segmentation Techniques Using Machine Learning 6th

Image Enhancement by Various Segmentation Techniques Using M...

引用

6th International conference on Information and Communication Technology for Intelligent systems, ICTIS 2021

作者： Chanchlani, Akshita Thakare, Vilas M. Wadhai, Vijay M. Department of Computer Science and Engineering SGBA University Amravati India Department of Electronics and Telecommunications Engineering D. Y. Patil College of Engineering Akurdi Pune India

ISBN: (纸本)9789811935701

Brain Cancer is a disease where abnormal cells grow without control and can invade other tissues, if it is identified at the early stage and treated appropriately then it can save human life. In developing countries cause of disease like brain cancer is effecting many people. Brain Cancer Medical image processing is most important and challenging field now a days. image preprocessing and its segmentation on brain cancer Magnetic Resonance Imaging (MRI) and its analysis is process of finding the objects and understanding their significance. image segmentation have become most popular due to its usage into various applications. The objective of this paper is exploring and implementation of various algorithms of image segmentation on medical image specifically on brain cancer MRI. image segmentation is very important step of understanding the images for further processing and analysis. This paper represents implementation of various image segmentation techniques that includes image Thresholding, Applying Various Morphological Operations, edge and region-based segmentation for enhancement of brain cancer images and detection of contours from brain cancer MRI. © 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

关键词： image analysis

来源：评论

学校读者我要写书评

暂无评论

Feature Rectification and Distribution Correction for Few-Shot image Classification

Feature Rectification and Distribution Correction for Few-Sh...

引用

Natural Language processing (ICNLP), International conference on

作者： Qiping Cheng Ying Liu Weidong Zhang Center for Image and Information Processing Xi’an University of Posts and Telecommunications Xi’an China

ISBN: (数字)9798350349115

ISBN: (纸本)9798350349122

Few-shot learning aims to train a model that can effectively capture data relationships using a limited amount of data. Although few-shot learning has achieved some success in the field of image classification, its generalization still has a large room for improvement. To this end, this paper proposes a few-shot learning strategy named Feature Rectification and Distribution Correction (FRDC). FRDC incorporates a feature rectification function to improve the model adaptability after the change of discriminative features, thereby improving the generalization ability of the model. By adding tukey_yeo-johnson transform, the model aligns more closely with the normal distribution in the feature space and the difference with the original data distribution is minimized. Hence, the classification performance of the model is improved. This strategy avoids over-reliance on complex model structures and is applicable to any existing model. Experiments show that the test accuracy of the proposed strategy is slightly better than that of the current advanced algorithms on two datasets. In the cross-dataset classification task, the classification accuracy is improved by about 7% compared with the Distribution Calibration (DC) algorithm.

关键词： Adaptation models Accuracy Transforms Gaussian distribution Data models Natural language processing Classification algorithms Complexity theory Few shot learning image classification

来源：评论

学校读者我要写书评

暂无评论

A new method for non-cooperative signal parameter estimation 4

A new method for non-cooperative signal parameter estimation

引用

4th International conference on Sensors and Information Technology, ICSI 2024

作者： Yang, Han Rao, Yongnan Lu, xiaochun Xi'an710600 China Key Laboratory of Time Reference and Applications Chinese Academy of Sciences Xi'an710600 China University of Chinese Academy of Sciences Beijing100049 China

ISBN: (数字)9781510678699

ISBN: (纸本)9781510678682

Navigation signals are one of the most crucial components of the Global Navigation Satellite System (GNSS), acting as a hub among satellites, users, and ground control systems. Conducting space signal quality assessments and monitoring the navigation system are essential means to ensure the normal functioning of navigation system capabilities such as positioning, speed measurement, and timing. These assessments not only provide high-quality services to users but also offer theoretical support for navigation system design and technical validation. In scenarios lacking prior information, it is necessary to estimate parameters such as the initial value of the carrier phase and carrier frequency to achieve signal demodulation and recover the pseudocode for subsequent capture and tracking, thus enabling signal assessment in non-cooperative settings. Existing methods for parameter estimation, particularly under conditions of low signal-to-noise ratios and the presence of interference, do not provide satisfactory accuracy and fail to meet the requirements for capture and tracking. This paper introduces a method for estimating the parameters of QPSK signals, utilizing image information combined with unsupervised clustering algorithms. Simulations show significantly improved estimation accuracy under low signal-to-noise ratio conditions. This method can also be applied to the estimation of signal modulation characteristic parameters and eye diagram parameters, providing important technical support for signal quality assessment. © 2024 SPIE.

关键词： Signal to noise ratio

来源：评论

学校读者我要写书评

暂无评论

A one-stage deep learning based method for automatic analysis of droplet-based digital PCR images

引用

ANALYST 2023年第13期148卷 3065-3073页

作者： Yao, Yuanyang Zhao, Shuhao Liang, Yan Hu, Fei Peng, Niancai Xi An Jiao Tong Univ State Key Lab Mfg Syst Engn Xian 710054 Shaanxi Peoples R China

Droplet-based dPCR offers many advantages over chip-based dPCR, such as lower processing cost, higher droplet density, higher throughput, while requiring less sample. However, the stochastic nature of droplet locations, uneven illuminations, and unclear droplet boundaries make automatic image analysis challenging. Most methods currently used to count a large amount of microdroplets rely on flow detection. Conventional machine vision algorithms cannot extract all information of the targets from complex backgrounds. Some two-stage methods, which first locate and then classify droplets according to their grayscale values, require high-quality imaging. In this study, we addressed these limitations by improving a one-stage deep learning algorithm named YOLOv5 and applying it to the detection task to realize one-stage detection. We introduced an attention mechanism module to increase the detection rate of small targets and used a new loss function to speed up the training process. Furthermore, we employed a network pruning method to facilitate the deployment of the model on mobile devices while preserving its performance. We validated the model with captured droplet-based dPCR images and found that the improved model accurately identified negative and positive droplets in complex backgrounds with an error rate of 0.65%. This method is characterized by its fast detection speed, high accuracy, and ability to be used on mobile devices or cloud platforms. Overall, the study presents a novel approach for detecting droplets in large-scale microdroplet images and provides a promising solution for accurate and efficient droplet counting in droplet-based dPCR.

关键词： Stochastic systems

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：