检索结果-内蒙古大学图书馆

Weakly supervised semantic segmentation by iteratively refining optimal segmentation with deep cues guidance

NEURAL COMPUTING & APPLICATIONS 2021年第15期33卷 9035-9060页

作者： Al-Huda, Zaid Peng, Bo Yang, Yan Algburi, Riyadh Nazar Ali Ahmad, Muqeet Khurshid, Faisal Moghalles, Khaled Southwest Jiaotong Univ Sch Informat Sci & Technol Chengdu 610031 Sichuan Peoples R China Southwest Jiaotong Univ Natl Engn Lab Integrated Transportat Big Data App Chengdu 610031 Sichuan Peoples R China Southwest Jiaotong Univ Sch Mech Engn Chengdu 610031 Sichuan Peoples R China

Weakly supervised semantic segmentation under image-level label supervision has undergone impressive improvements over the past years. These approaches can significantly reduce human annotation efforts, although they remain inferior to fully supervised procedures. In this paper, we propose a novel framework that iteratively refines pixel-level annotations and optimizes segmentation network. We first produce initial deep cues using the combination of activation maps and a saliency map. To produce high-quality pixel-level annotations, a graphical model is constructed over optimal segmentation of high-quality region hierarchies to propagate information from deep cues to unmarked regions. In the training process, the initial pixel-level annotations are used as supervision to train the segmentation network and to predict segmentation masks. To correct inaccurate labels of segmentation masks, we use these segmentation masks with the graphical model to produce accurate pixel-level annotations and use them as supervision to retrain the segmentation network. Experimental results show that the proposed method can significantly outperform the weakly-supervised semantic segmentation methods using static labels. The proposed method has state-of-the-art performance, which are 66.7% mIoU score on PASCAL VOC 2012 test set and 27.0% mIoU score on MS COCO validation set.

关键词： Deep convolutional neural networks Graphical model hierarchical image segmentation Semantic segmentation Weakly supervised

来源：评论

学校读者我要写书评

暂无评论

image segmentation using dense and sparse hierarchies of superpixels

引用

PATTERN RECOGNITION 2020年 108卷 107532-107532页

作者： Galvao, Felipe Lemes Guimaraes, Silvio Jamil Ferzoli Falcao, Alexandre Xavier Univ Estadual Campinas Inst Comp Lab Image Data Sci Av Albert Einstein 1251 BR-13083852 Campinas SP Brazil Pontificia Univ Catolica Minas Gerais Comp Sci Dept BR-31980110 Belo Horizonte MG Brazil

We investigate the intersection between hierarchical and superpixel image segmentation. Two strategies are considered: (i) the classical region merging, that creates a dense hierarchy with a higher number of levels, and (ii) the recursive execution of some superpixel algorithm, which generates a sparse hierarchy with fewer levels. We show that, while dense methods can capture more intermediate or higher-level object information, sparse methods are considerably faster and usually with higher boundary adherence at finer levels. We first formalize the two strategies and present a sparse method, which is faster than its superpixel algorithm and with similar boundary adherence. We then propose a new dense method to be used as post-processing from the intermediate level, as obtained by our sparse method, upwards. This combination results in a unique strategy and the most effective hierarchical segmentation method among the compared state-of-the-art approaches, with efficiency comparable to the fastest superpixel algorithms. (C) 2020 Elsevier Ltd. All rights reserved.

关键词： Superpixel segmentation hierarchical image segmentation image foresting transform Iterative spanning forest Graph-based image segmentation Irregular image pyramid

来源：评论

学校读者我要写书评

暂无评论

Variational shape prior segmentation with an initial curve based on image registration technique

引用

image AND VISION COMPUTING 2020年第0期94卷 103865-000页

作者： Yeo, Doyeob Lee, Chang-Ock Korea Adv Inst Sci & Technol Dept Math Sci Daejeon 34141 South Korea

In general images, it is practically hard to distinguish only the desired object using the conventional image segmentation methods. In many cases, we can segment the desired object by using the shape information of the object in addition to the standard image segmentation. Chan and Zhu's model is not robust to the intensity changes of objects. In this paper, we propose a novel model for the shape prior segmentation that produces robust results using the hierarchical image segmentation and an attraction term. Moreover, we adopt an image registration technique and a multi-region image segmentation to get an initial for a given shape prior. Finally, we consider the free-form deformation in obtaining the shape function from the reference shape prior for real-world images. Numerical experiments demonstrate the results independent of intensities of objects and the location of the reference shape prior. All numerical calculations are automatic and progress without any user input. (C) 2019 Elsevier B.V. All rights reserved.

关键词： Shape prior segmentation hierarchical image segmentation image registration Free-form deformation

来源：评论

学校读者我要写书评

暂无评论

image Inpainting Based on Local Patch Search Supported by image segmentation 23rd

Image Inpainting Based on Local Patch Search Supported by Im...

引用

23rd Iberoamerican Congress on Pattern Recognition (CIARP)

作者： Carneiro, Sarah Almeida Pedrini, Helio Ferzoli Guimaraes, Silvio Jamil Pontificia Univ Catolica Minas Gerais Comp Sci Dept Belo Horizonte MG Brazil Univ Estadual Campinas Inst Comp Campinas SP Brazil

ISBN: (纸本)9783030134686;9783030134693

image inpainting can be defined as a restoration process in which damaged or selected regions are repaired by taking into account the image content. In this work, we employ a local-based strategy instead of a global one to identify the best existing patch with information to replace the damaged/selected patch. In order to properly identify the most representative patches, we propose a method based on the (i) creation of a local graph using a similarity of patches in the original image and (ii) partition of the image into regions according to hierarchical image segmentation to support the local patch identification. The experimental results demonstrate that our local search outperformed the results of image inpainting in terms of both qualitative and quantitative aspects, when compared to global search of patches.

关键词： image inpainting hierarchical image segmentation Patch graph

来源：评论

学校读者我要写书评

暂无评论

Convolutional Oriented Boundaries: From image segmentation to High-Level Tasks

引用

IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 2018年第4期40卷 819-833页

作者： Maninis, Kevis-Kokitsi Pont-Tuset, Jordi Arbelaez, Pablo Van Gool, Luc ETHZ Comp Vis Lab CH-8092 Zurich Switzerland Univ Los Andes Dept Biomed Engn Bogota 111711 Colombia

We present Convolutional Oriented Boundaries ( COB), which produces multiscale oriented contours and region hierarchies starting from generic image classification Convolutional Neural Networks (CNNs). COB is computationally efficient, because it requires a single CNN forward pass for multi-scale contour detection and it uses a novel sparse boundary representation for hierarchical segmentation;it gives a significant leap in performance over the state-of-the-art, and it generalizes very well to unseen categories and datasets. Particularly, we show that learning to estimate not only contour strength but also orientation provides more accurate results. We perform extensive experiments for low-level applications on BSDS, PASCAL Context, PASCAL segmentation, and NYUD to evaluate boundary detection performance, showing that COB provides state-of-the-art contours and region hierarchies in all datasets. We also evaluate COB on high-level tasks when coupled with multiple pipelines for object proposals, semantic contours, semantic segmentation, and object detection on MS-COCO, SBD, and PASCAL;showing that COB also improves the results for all tasks.

关键词： Contour detection contour orientation hierarchical image segmentation object proposals semantic contours

来源：评论

学校读者我要写书评

暂无评论

Exploring hierarchy simplification for non-significant region removal 32

Exploring hierarchy simplification for non-significant regio...

引用

32nd SIBGRAPI Conference on Graphics, Patterns and images (SIBGRAPI) / Symposium on Virtual and Augmented Reality (SVR)

作者： Barcelos, Isabela Borlido da Fonseca, Gabriel Barbosa Najman, Laurent Kenmochi, Yukiko Perret, Benjamin Cousty, Jean do Patrocinio, Zenilton K. G., Jr. Guimaraes, Silvio Jamil F. Univ Paris Est LIGM UPEMLV ESIEE ParisENPCCNRSUMR 8049 F-93162 Noisy Le Grand France Pontificia Univ Catolica Minas Gerais Audio Visual Proc Lab VIPLAB BR-31980110 Belo Horizonte MG Brazil

ISBN: (纸本)9781728152271

image segmentation is a classic subject in the field of digital image processing, and it can be used to solve a large variety of problems or serve as preprocessing for other methods of image analysis. hierarchical image segmentation methods provide a multiscale representation, therefore they produce a nested set of image segmentations in which a result at a given level can be produced by merging regions of the segmentation at its previous level. However, a hierarchical representation may produce small components at its coarser levels, leading to oversegmentations on such scales. To solve this problem, we explore strategies to simplify hierarchies in order to remove non-significant regions, in terms of area, while trying to preserve the hierarchical structure. We evaluate the proposed simplification strategies with different hierarchical segmentation methods on the Pascal Context dataset by using precision-recall measures and fragmentation curves, along with a qualitative assessment showing that the simplification of hierarchies can lead to visually better image segmentations.

关键词： hierarchical image segmentation Hierarchy simplification Non significant region removal

来源：评论

学校读者我要写书评

暂无评论

hierarchical Graph-Based segmentation in Detection of Object-Related Regions 23rd

Hierarchical Graph-Based Segmentation in Detection of Object...

引用

23rd Iberoamerican Congress on Pattern Recognition (CIARP)

作者： Ribeiro, Rafael Machado Guimaraes, Silvio Jamil Ferzoli Patrocinio, Zenilton Kleber G., Jr. Pontificia Univ Catolica Minas Gerais Comp Sci Dept Belo Horizonte MG Brazil

ISBN: (纸本)9783030134686;9783030134693

Object detection is an important task in computer vision. Recently, several unsupervised approaches have been proposed to cope with this problem in a category-independent manner. This work evaluates the adoption of a hierarchical graph-based segmentation along with an state-of-the-art method to detect object-related regions. A hierarchical segmentation approach produces a set of partitions at different detail levels, in a way that a coarser level can be obtained by a simple merge of finer ones. Experimental results show that our proposal obtains an increase of 11% in object detection rate.

关键词： Object detection hierarchical image segmentation Object segmentation Object localization

来源：评论

学校读者我要写书评

暂无评论

hierarchical segmentations with Graphs: Quasi-flat Zones, Minimum Spanning Trees, and Saliency Maps

引用

JOURNAL OF MATHEMATICAL IMAGING AND VISION 2018年第4期60卷 479-502页

作者： Cousty, Jean Najman, Laurent Kenmochi, Yukiko Guimaraes, Silvio Univ Paris Est CNRS Lab Informat Gaspard Monge ESIEE ParisA3SI Marne La Vallee France PUC Minas ICEI DCC VIPLAB Belo Horizonte MG Brazil

Hierarchies of partitions are generally represented by dendrograms (direct representation). They can also be represented by saliency maps or minimum spanning trees. In this article, we precisely study the links between these three representations. In particular, we provide a new bijection between saliency maps and hierarchies based on quasi-flat zones as often used in image processing and we characterize saliency maps and minimum spanning trees as solutions to constrained minimization problems where the constraint is quasi-flat zones preservation. In practice, these results make up a toolkit for designing new hierarchical methods where one can choose the most convenient representation. They also invite us to process non-image data with morphological hierarchies. More precisely, we show the practical interest of the proposed framework for: (i) hierarchical watershed image segmentations, (ii) combinations of different hierarchical segmentations, (iii) hierarchicalizations of some non-hierarchical image segmentation methods based on regional dissimilarities, and (iv) hierarchical analysis of geographic data.

关键词： Mathematical morphology Hierarchy of partitions hierarchical image segmentation Watershed Saliency maps Minimum spanning trees hierarchical classification

来源：评论

学校读者我要写书评

暂无评论

hierarchical segmentation of Remote Sensing images by Unsupervised Deep Learning Features 10

Hierarchical Segmentation of Remote Sensing Images by Unsupe...

引用

10th International Symposium on Computational Intelligence and Design (ISCID)

作者： Li, Yuhui Hong, Huo Fang, Tao Shanghai Jiao Tong Univ Dept Automat Key Lab Syst Control & Informat Proc Minist Educ Shanghai 200240 Peoples R China

ISBN: (纸本)9781538636756

Due to the scale diversity of geographical objects, hierarchical remote sensing image segmentation plays an important role in object-oriented image analysis. In this paper, a hierarchical remote sensing images segmentation method with unsupervised deep learning features is proposed. The unsupervised deep learning features of an image are extracted with a sparse convolutional auto-encoder. Both deep features and the clustering information of deep features are utilized in hierarchical image segmentation which is a bottom-up region merging process. The iterative region merging process starts from an initial partition of an image under a merging criterion. Finally, a tree-like image segmentations hierarchy which contains ground objects of different scales is obtained. The proposed method integrates the advantages of unsupervised deep learning features with region merging-based hierarchical image segmentation. The experiment results have shown the proposed method is superior to the methods using only spectral or spatial features.

关键词： hierarchical image segmentation sparse convolutional auto-encoder region adjacency graph nearest neighbor graph remote sensing images

来源：评论

学校读者我要写书评

暂无评论

Fuzzy image segmentation based on the hierarchical Divide and Link clustering algorithm 10

Fuzzy Image Segmentation based on the hierarchical Divide an...

引用

10th International Conference on Intelligent Systems and Knowledge Engineering (ISKE)

作者： Guada, Carely Gomez, Daniel Tinguaro Rodriguez, J. Yanez, Javier Montero, Javier Univ Complutense Fac Ciencias Matemat E-28040 Madrid Spain Univ Complutense Fac Estudios Estadist E-28040 Madrid Spain Univ Complutense Fac Ciencias Matemat Inst IGEO UCM CSIC UCM E-28040 Madrid Spain

ISBN: (纸本)9781467393232

In this paper, we present a method to obtain a Fuzzy image segmentation from the hierarchical clustering algorithm Divide and Link. The Divide and Link algorithm consists on treating a digital image as a graph, then building spanning forest through a Kruskal scheme to successively sort the edges while partitions are obtained. This process is driven until all the pixels of the image are segmented, that is, there are as many regions as pixels.

关键词： Fuzzy image segmentation hierarchical image segmentation image network spanning forest clustering

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：