检索结果-内蒙古大学图书馆

MSEB: Plug and play multi-scale image embedding block for vision backbone

NEUROCOMPUTING 2025年 617卷

作者： Yuan, Hao Zhang, Bin Wang, Yachuan Xi An Jiao Tong Univ 28 Xian Ning West Rd Xian 710049 Shaanxi Peoples R China

Mainstream backbone networks in computer vision, leveraging sophisticated CNN or Transformer architectures, have achieved remarkable performance across various tasks. However, these works primarily focus on downsample ranges of 4x to 32x on input images, with limited exploration into the embedding portion of the network. In this paper, we propose a novel Multi-Scale embedding Block (MSEB) to enhance embedding features. MSEB is constructed through a combination of Self-Calibrated Channel Dilate Unit (SCDU) and Multi-Scale Encoder Unit (MSEU). SCDU projects images into high-dimensional space, incorporating channel self-calibration to preserve fine-grained information in shallow feature maps, thereby providing amore comprehensive representation for subsequent network layers. MSEU employs a multi-scale cascade design with implicit feature sharing characteristics, expanding the receptive field scale range to enhance embedding representation. Extensive experiments on mainstream benchmarks demonstrate that our proposed multi-scale embedding block can be seamlessly integrated into most popular CNN and Transformer architectures such as ResNet, Res2Net, Swin-Transformer, and MViT. By replacing the existing embedding structure in these methods, significant improvements on imageNet-1k datasets are observed, e.g., CNN-based ResNet-50 and further adopting a multi-scale design Res2Net-50 achieve Top-1 accuracy improvements of 0.47% and 1.22%, respectively. Attention-based Swin-Tiny and MViT-Tiny, both adopting multi-scale design, achieved Top-1 accuracy improvements of 2.57% and 1.05%, respectively.

关键词： Multi-scale Vision backbone image embedding Computer vision

来源：评论

学校读者我要写书评

暂无评论

image embedding for Detecting Irregularity 3rd

Image Embedding for Detecting Irregularity

引用

3rd International Conference on Computer Vision and image Processing (CVIP)

作者： Sharma, M. K. Sheet, D. Biswas, Prabir Kumar IIT Kharagpur Adv Technol Dev Ctr Kharagpur W Bengal India IIT Kharagpur Elect Engn Kharagpur W Bengal India IIT Kharagpur Elect & Elect Commun Engn Kharagpur W Bengal India

ISBN: (纸本)9789813292918;9789813292901

Detecting irregularity in an image or video is an important task in quality control or automatic visual inspection. This paper presents an image embedding technique for detecting an irregularity or abnormality in images. This can further be utilized in image screening application. In the proposed architecture, deep adversarial autoencoder is trained to extract the features from images. Using these features and skip-gram model, we develop the image2vec architecture to capture contextual probability in an image. Various score aggregation techniques are explored and its performance is reported. As a case study, we present a scenario of foreign body object detection in clinical-grade X-ray images. The proposed approach is found to correctly detect and localize abnormality in images.

关键词： Visual inspection image embedding Contextual abnormality detection Adversarial autoencoder Skip-gram Negative sampling

来源：评论

学校读者我要写书评

暂无评论

QR images: Optimized image embedding in QR Codes

引用

IEEE TRANSACTIONS ON image PROCESSING 2014年第7期23卷 2842-2853页

作者： Garateguy, Gonzalo J. Arce, Gonzalo R. Lau, Daniel L. Villarreal, Ofelia P. Mathworks Natick MA 01760 USA Univ Delaware Dept Elect & Comp Engn Newark DE 19711 USA Univ Kentucky Dept Elect & Comp Engn Lexington KY 40506 USA

This paper introduces the concept of QR images, an automatic method to embed QR codes into color images with bounded probability of detection error. These embeddings are compatible with standard decoding applications and can be applied to any color image with full area coverage. The QR information bits are encoded into the luminance values of the image, taking advantage of the immunity of QR readers against local luminance disturbances. To mitigate the visual distortion of the QR image, the algorithm utilizes halftoning masks for the selection of modified pixels and nonlinear programming techniques to locally optimize luminance levels. A tractable model for the probability of error is developed and models of the human visual system are considered in the quality metric used to optimize the luminance levels of the QR image. To minimize the processing time, the optimization techniques proposed to consider the mechanics of a common binarization method and are designed to be amenable for parallel implementations. Experimental results show the graceful degradation of the decoding rate and the perceptual quality as a function the embedding parameters. A visual comparison between the proposed and existing methods is presented.

关键词： QR codes image embedding halftoning

来源：评论

学校读者我要写书评

暂无评论

VISAtlas: An image-Based Exploration and Query System for Large Visualization Collections via Neural image embedding

引用

IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS 2024年第7期30卷 3224-3240页

作者： Ye, Yilin Huang, Rong Zeng, Wei Hong Kong Univ Sci & Technol Guangzhou Guangzhou 511400 Guangdong Peoples R China Hong Kong Univ Sci & Technol Hong Kong Peoples R China

High-quality visualization collections are beneficial for a variety of applications including visualization reference and data-driven visualization design. The visualization community has created many visualization collections, and developed interactive exploration systems for the collections. However, the systems are mainly based on extrinsic attributes like authors and publication years, whilst neglect intrinsic property (i.e., visual appearance) of visualizations, hindering visual comparison and query of visualization designs. This paper presents VISAtlas, an image-based approach empowered by neural image embedding, to facilitate exploration and query for visualization collections. To improve embedding accuracy, we create a comprehensive collection of synthetic and real-world visualizations, and use it to train a convolutional neural network (CNN) model with a triplet loss for taxonomical classification of visualizations. Next, we design a coordinated multiple view (CMV) system that enables multi-perspective exploration and design retrieval based on visualization embeddings. Specifically, we design a novel embedding overview that leverages contextual layout framework to preserve the context of the embedding vectors with the associated visualization taxonomies, and density plot and sampling techniques to address the overdrawing problem. We demonstrate in three case studies and one user study the effectiveness of VISAtlas in supporting comparative analysis of visualization collections, exploration of composite visualizations, and image-based retrieval of visualization designs. The studies reveal that real-world visualization collections (e.g., Beagle and VIS30K) better accord with the richness and diversity of visualization designs than synthetic collections (e.g., Data2Vis), inspiring composite visualizations are identified in real-world collections, and distinct design patterns exist in visualizations from different sources.

关键词： Data visualization Visualization Feature extraction Task analysis Layout Taxonomy Semantics Visualization collection image embedding visual query image visualization design pattern

来源：评论

学校读者我要写书评

暂无评论

image embedding into Generative Adversarial Networks

Image Embedding into Generative Adversarial Networks

引用

作者： Abdal, Rameen King Abdullah University of Science and Technology

学位级别：硕士

We propose an e cient algorithm to embed a given image into the latent space of StyleGAN. This embedding enables semantic image editing operations that can be applied to existing photographs. Taking the StyleGAN trained on the FFHQ dataset as an example, we show results for image morphing, style transfer, and expression transfer. Studying the results of the embedding algorithm provides valuable insights into the structure of the StyleGAN latent space. We propose a set of experiments to test what class of images can be embedded, how they are embedded, what latent space is suitable for embedding, and if the embedding is semantically meaningful.

关键词： Generative modeling GANs image embedding image editing StyleGAN Deep Learning

来源：评论

学校读者我要写书评

暂无评论

Colored-QRNet: Fast QR Code Color image embedding 30

Colored-QRNet: Fast QR Code Color Image Embedding

引用

30th European Signal Processing Conference (EUSIPCO)

作者： Pena-Pena, Karelia Lau, Daniel L. Arce, Gonzalo R. Univ Delaware Elect & Comp Engn Newark DE 19716 USA Univ Kentucky Elect & Comp Engn Lexington KY 40506 USA

ISBN: (纸本)9789082797091

Quick Response (QR) codes are widely used to connect offline and online content, and thus many efforts have aimed at improving the visual quality of QR codes to be easily included in publicity designs in billboards and magazines. The most successful approaches, however, are slow since optimization algorithms are required for the generation of each beautified QR code, hindering its online customization. The aim of this paper is the fast generation of visually pleasant and robust QR codes. The proposed framework leverages state-of-the-art deep-learning algorithms to embed a color image into a baseline QR code in seconds while keeping a maximum probability of error during the decoding procedure. Halftoning techniques that exploit the human visual system (HVS) are used to smooth the embedding of the QR code structure in the final QR code image while reinforcing the decoding robustness. Compared to optimization-based methods, our framework provides similar qualitative results but is significantly faster.

关键词： QR codes image embedding deep learning

来源：评论

学校读者我要写书评

暂无评论

Effectiveness of Deep image embedding Clustering Methods on Tabular Data 15

Effectiveness of Deep Image Embedding Clustering Methods on ...

引用

15th International Conference on Advanced Computational Intelligence (ICACI)

作者： Abrar, Sakib Sekmen, Ali Samad, Manar D. Tennessee State Univ Dept Comp Sci Nashville TN 37203 USA

ISBN: (纸本)9798350321456

Deep learning methods are commonly benchmarked on image data sets, which may not be suitable or effective baselines for non-image tabular data. In this paper, we take a data-centric view to perform one of the first studies on deep embedding clustering of tabular data. Eight clustering and state-of-the-art embedding clustering methods proposed for image data sets are tested on seven tabular data sets. Our results reveal that traditional clustering ranks second out of eight methods on tabular data and is superior to most deep embedding clustering baselines. Our observation aligns with the literature that traditional machine learning of tabular data is still more effective than deep learning. Therefore, state-of-the-art embedding clustering methods should consider data-centric custom learning architectures and algorithms for tabular data.

关键词： tabular data embedding clustering image embedding convolutional neural network deep clustering

来源：评论

学校读者我要写书评

暂无评论

Garbage Classification Using Inception V3 as image embedding and Extreme Gradient Boosting

Garbage Classification Using Inception V3 as Image Embedding...

引用

2024 ASU International Conference in Emerging Technologies for Sustainability and Intelligent Systems, ICETSIS 2024

作者： Yulita, Intan Nurma Ardiansyah, Firman Sholahuddin, Asep Rosadi, Rudi Trisanto, Agus Ramdhani, Muhammad Rasyid Universitas Padjadjaran Department of Computer Science Sumedang Indonesia Institut Teknologi Dan Bisnis Ahmad Dahlan Lamongan Management Lamongan Indonesia Universitas Padjadjaran Department of Electrical Engineering Sumedang Indonesia

ISBN: (纸本)9798350372229

Employing the image embedding method with the Inception V3 model, this study explored the possibility of utilizing visual features for garbage classification. Several algorithms, including Extreme Gradient Boosting (XGBoost), Naive Bayes, AdaBoost, and Random Forest, were evaluated for their effectiveness in classifying garbage based on extracted visual characteristics. Our investigation commenced with the collection of low-quality image data before conducting feature extraction with the Inception V3 model. For assessment, a dataset that contains the above characteristics is utilized as input. This is then fed into various classification algorithms, and assessment is done based on evaluation measures such as F1 Score and Area under the ROC Curve (AUC). Given the class imbalance in the garbage dataset, the focus is mainly on the AUC measure, as it is more effective in handling such imbalances. Garbage classification can be efficiently carried out by combining the Inception V3 and XGboost algorithms, as per the research. A valuable insight into effective waste management through the use of machine learning is obtained herein. Such a technique could pave the way for ecologically conscious waste disposal measures. © 2024 IEEE.

关键词： Garbage Classification image embedding Inception V3 XGBoost

来源：评论

学校读者我要写书评

暂无评论

An improved region-based embedding technique for data hiding and image recovery using multiple ROI and RONI

引用

INTERNATIONAL JOURNAL OF ELECTRONIC SECURITY AND DIGITAL FORENSICS 2023年第2期15卷 101-113页

作者： Paikaray, Bijay Kumar Swain, Debabala Chakravarty, Sujata Centurion Univ Technol & Management Dept CSE R Sitapur Odisha India Rama Devi Womens Univ Dept Comp Sci Bhubaneswar Odisha India Centurion Univ Technol & Management Dept CSE R Sitapur Odisha India

To preserve the sensitive contents of digital images during their transmission, it is essential to hide them with maximum imperceptibility so that intruders will not be able to identify them visually. The image recovery at the receiver end is equally significant because of sensitive images, like medical diagnosis images, satellite images, etc. This paper proposes an improved image hiding technique where the sensitive contents of the image are located in multiple regions. These regions get embedded based on histogram analysis of the region of interest (ROI) pixels then hidden in the region of non-interest (RONI). Further, the reverse operations can be applied to the embedded regions and the hidden data are retrieved from RONI. Using this technique, the embedded regions can be easily extracted, recovered, and fully restored without any loss in data. The proposed work is on multiple ROI with the reliability, integrity, and confidentiality of transmitted images.

关键词： multiple ROI region of non-interest RONI medical image image embedding hiding imperceptibility recovery

来源：评论

学校读者我要写书评

暂无评论

A deep learning-driven multi-layered steganographic approach for enhanced data security

引用

SCIENTIFIC REPORTS 2025年第1期15卷 1-30页

作者： Sanjalawe, Yousef Al-E'mari, Salam Fraihat, Salam Abualhaj, Mosleh Alzubi, Emran Univ Jordan JU King AbdullahSchool Informat Technol 2 Dept Informat Technol Amman 11942 Jordan Univ Petra UoP Fac Informat Technol Dept Informat Secur Amman 11196 Jordan Ajman Univ Coll Engn & Informat Technol Artificial Intelligence Res Ctr AIRC Ajman U Arab Emirates Al Ahliyya Amman Univ Fac Informat Technol Dept Networks & Informat Secur Amman 19328 Jordan Northern Border Univ NBU Coll Business Adm Ar Ar 91431 Saudi Arabia

In the digital era, ensuring data integrity, authenticity, and confidentiality is critical amid growing interconnectivity and evolving security threats. This paper addresses key limitations of traditional steganographic methods, such as limited payload capacity, susceptibility to detection, and lack of robustness against attacks. A novel multi-layered steganographic framework is proposed, integrating Huffman coding, Least Significant Bit (LSB) embedding, and a deep learning-based encoder-decoder to enhance imperceptibility, robustness, and security. Huffman coding compresses data and obfuscates statistical patterns, enabling efficient embedding within cover images. At the same time, the deep learning encoder adds layer of protection by concealing an image within another. Extensive evaluations using benchmark datasets, including Tiny imageNet, COCO, and CelebA, demonstrate the approach's superior performance. Key contributions include achieving high visual fidelity with Structural Similarity Index Metrics (SSIM) consistently above 99%, robust data recovery with text recovery accuracy reaching 100% under standard conditions, and enhanced resistance to common attacks such as noise and compression. The proposed framework significantly improves robustness, security, and computational efficiency compared to traditional methods. By balancing imperceptibility and resilience, this paper advances secure communication and digital rights management, addressing modern challenges in data hiding through an innovative combination of compression, adaptive embedding, and deep learning techniques.

关键词： Data security Huffman encoding image embedding Steganography LSB embedding

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：