Filters








1,045,029 Hits in 6.2 sec

Enhancing Prototypical Few-Shot Learning by Leveraging the Local-Level Strategy [article]

Junying Huang, Fan Chen, Keze Wang, Liang Lin, Dongyu Zhang
2021 arXiv   pre-print
We found that the existing works often build their few-shot model based on the image-level feature by mixing all local-level features, which leads to the discriminative location bias and information loss  ...  Specifically, we present (a) a local-agnostic training strategy to avoid the discriminative location bias between the base and novel categories, (b) a novel local-level similarity measure to capture the  ...  BACKGROUND KNOWLEDGE Given disjoint category sets {C base , C val , C novel } and the corresponding datasets {D base , D val , D novel }, the objective of few-shot learning (FSL) is to train a robust model  ... 
arXiv:2111.04331v1 fatcat:ljungdynm5dmpegaghev4cuckm

A Novel Image Segmentation Approach Based on Improved Level Set Evolution Algorithm

Xiaoliang JIANG, Bailin LI, Jiajia LIU, Shaojie CHEN
2014 Sensors & Transducers  
Image segmentation is a fundamental topic in image processing. And the method of level set based on curve evolving theory is widely applied in image segmentation.  ...  In this paper, we propose a novel region-based active contour model which bases on the region- scalable fitting (RSF) term and the new signed pressure force (SPF) term.  ...  In this paper, we propose a novel region-based active contour model which bases on the regionscalable fitting term and the new Signed Pressure Function term.  ... 
doaj:be13f77016204edba0df83ec2f620b79 fatcat:wndzrimdnfbtrfuetk4ytpajju

Fractional Differentiation-Based Active Contour Model Driven by Local Intensity Fitting Energy

Ming Gu, Renfang Wang
2016 Mathematical Problems in Engineering  
Secondly, we defined a new energy functional based on local image information and fractional order differentiation image; the proposed model not only can describe the input image more accurately but also  ...  A novel active contour model is proposed for segmentation images with inhomogeneity.  ...  Conclusion This paper proposed a novel active contour model based on the local image information and fractional order differentiation image.  ... 
doi:10.1155/2016/6098021 fatcat:k2jzwklujngizhk66lafgtkpsy

Multi-scale mesh saliency based on low-rank and sparse analysis in shape feature space

Shengfa Wang, Nannan Li, Shuai Li, Zhongxuan Luo, Zhixun Su, Hong Qin
2015 Computer Aided Geometric Design  
The technical core of our approach is a new shape descriptor that embraces both local geometry information and global structure information in an integrated way.  ...  Towards this goal, we exploit our novel shape descriptor to define local-to-global shape context in a vertex-wise fashion and concatenate all the shape contexts to form a feature space, which encodes both  ...  For the low level, a low frequency model M l is constructed in Eq. (1) (we use the m-th level model M m , where m is set to be 200 empirically for all examples).  ... 
doi:10.1016/j.cagd.2015.03.003 fatcat:fc3ltb2x7vboxdnt2tiujogwvi

Active contour with selective local or global segmentation for intensity inhomogeneous image

Thi-Thao Tran, Van-Truong Pham, Yun-Jen Chiu, Kuo-Kai Shyu
2010 2010 3rd International Conference on Computer Science and Information Technology  
The presented method introduces a signed pressure force function using the local information of the image to be segmented. Thus, this model can work with heterogeneous images.  ...  In this paper, a novel algorithm for intensity inhomogeneous image segmentation is proposed.  ...  In this paper, we proposed a novel segmentation algorithm based on active contour method. The classical model in active contour method is snakes [1] , [2] .  ... 
doi:10.1109/iccsit.2010.5564761 fatcat:n42y5f2y4jg6fgy4zjpebshyyq

Multi-level Similarity Learning for Low-Shot Recognition [article]

Hongwei Xv, Xin Sun, Junyu Dong, Shu Zhang, Qiong Li
2019 arXiv   pre-print
Our approach is achieved based on the fact that the image similarity learning can be decomposed into image-level, global-level, and object-level.  ...  According to this concept, we propose a multi-level similarity model (MLSM) to capture the deep encoded distance metric between the support and query samples.  ...  Problem Definition Assuming that there is a base train dataset D base , a novel test dataset D novel , where D base D novel = ∅.  ... 
arXiv:1912.06418v1 fatcat:vivmcpoq4fhqrdixwykq66jq6a

Active Contours Driven by Local and Global Region-based Information for Image Segmentation

Xiaojun Yang, Xiaoliang Jiang, Lingfei Zhou, Yong Wang, Yuliang Zhang
2020 IEEE Access  
Aiming at these problems, this work proposes a novel hybrid active contour method which combines local and global statistical information into an improved signed pressure force (SPF) function.  ...  INDEX TERMS Active contour, image segmentation, signed pressure force, level set. 6460 This work is licensed under a Creative Commons Attribution 4.0 License.  ...  Similarly, in [19] , Min et al. proposed a novel image processing method based on multi-scale local feature.  ... 
doi:10.1109/access.2019.2963435 fatcat:uyyqb7bsxbck3pwlkzmdq4jfjm

Commonality-Parsing Network across Shape and Appearance for Partially Supervised Instance Segmentation [article]

Qi Fan, Lei Ke, Wenjie Pei, Chi-Keung Tang, Yu-Wing Tai
2020 arXiv   pre-print
Incorporating both the shape and appearance commonalities, our model significantly outperforms the state-of-the-art methods on both partially supervised setting and few-shot setting for instance segmentation  ...  The learned models are expected to be generalizable to novel categories.  ...  Fig. 6 . 6 The segmentation performance of different models on a fixed set of novel categories as a function of number of mask-annotated (base) categories.  ... 
arXiv:2007.12387v1 fatcat:2xywj3cprzh5bbza55sqwyfjki

Implicit Active Contours Driven by Local and Global Image Fitting Energy for Image Segmentation and Target Localization

Xiaosheng Yu, Yuanchen Qi, Ziwei Lu, Nan Hu
2013 Journal of Sensors  
We propose a novel active contour model in a variational level set formulation for image segmentation and target localization.  ...  We combine a local image fitting term and a global image fitting term to drive the contour evolution.  ...  Conclusion This paper presents a novel region-based active contour model in a variational level set formulation.  ... 
doi:10.1155/2013/713536 fatcat:ytpcznpxnbhhxfsxubg6xwzgoy

Saliency Detection Using Texture and Local Cues [chapter]

Qiang Qi, Muwei Jian, Yilong Yin, Junyu Dong, Wenyin Zhang, Hui Yu
2017 Communications in Computer and Information Science  
Firstly, an effective method based on selective contrast (SC), which explores the most distinguishable component information in texture, is used to calculate the texture saliency map.  ...  Experimental results, based on a widely used and openly available database, demonstrate that the proposed method can produce competitive results and outperforms some existing popular methods.  ...  [11] proposed a novel saliency detection method based on frequency tuned model to detect salient object.  ... 
doi:10.1007/978-981-10-7305-2_58 fatcat:cxbb5skvmrgsvlbc364d3za3o4

Underwater object segmentation integrating transmission and saliency features

Zhe Chen, Yang Sun, Yupeng Gu, Huibin Wang, Hao Qian, Hao Zheng
2019 IEEE Access  
In this paper, novel cues and a suitable model formulation for object segmentation from underwater images are proposed.  ...  Various types of knowledge and features have been explored for level set-based segmentation.  ...  In order to prevent the local optimization problem, a re-initialization process is necessary for the edge-based models. However, this process largely reduces the automation of the level set model.  ... 
doi:10.1109/access.2019.2919711 fatcat:fsqr7not7vgjnj7bky2qbk7phq

A novel combined level set model for automatic MR image segmentation

Jianzhang Li, Sven Nebelung, Björn Rath, Markus Tingart, Jörg Eschweiler
2020 Current Directions in Biomedical Engineering  
In this paper, we introduce a novel combined level set model that mainly cooperates with an edge detector and a local region intensity descriptor.  ...  The noise and intensity inhomogeneities are eliminated by the local region intensity descriptor. The edge detector helps the level set model to locate the object boundaries more precisely.  ...  The Level Set Method (LSM) has become a popular technique in recent years [1, 2] . Two main classes represent the typical LSM, which are edge-based [2, 3] and region-based models [4] [5] [6] .  ... 
doi:10.1515/cdbme-2020-3006 fatcat:uofrop5ejrfovk7m543ji7i5ei

Scene Categorization Using Topic Model Based Hierarchical Conditional Random Fields [chapter]

Vikram Garg, Ehtesham Hassan, Santanu Chaudhury, Madan Gopal
2011 Lecture Notes in Computer Science  
We use Conditional Random Fields in a hierarchical setting for discovering the global context of these topics. The learned random fields are further used for categorization of a new scene.  ...  We propose a novel hierarchical framework for scene categorization. The scene representation is defined by latent topics extracted by Latent Dirichlet Allocation.  ...  The LDA assigns a topic distribution to each image patch ( figure (1 shows topics on face image using color based bag-of-words model).  ... 
doi:10.1007/978-3-642-21786-9_35 fatcat:gepjvxlyx5fi7f5jercbl5vi4q

Sparse Structure Learning via Graph Neural Networks for Inductive Document Classification

Yinhua Piao, Sangseon Lee, Dohoon Lee, Sun Kim
2022 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
To address these challenges, we propose a novel GNN-based sparse structure learning model for inductive document classification.  ...  However, most existing methods are based on static word co-occurrence graphs without sentence-level information, which poses three challenges:(1) word ambiguity, (2) word synonymity, and (3) dynamic contextual  ...  To this end, we construct a novel trainable document-level graph to jointly capture local and global contextual information.  ... 
doi:10.1609/aaai.v36i10.21366 fatcat:ek76uytcljddrhm5tuzdfgujfy

Biologically Motivated Novel Localization Paradigm by High-Level Multiple Object Recognition in Panoramic Images

Sungho Kim, Min-Sheob Shim
2015 The Scientific World Journal  
The metric global localization (position, viewing direction) was conducted based on the bearing information of recognized objects from just one panoramic image.  ...  This paper presents the novel paradigm of a global localization method motivated by human visual systems (HVSs).  ...  The object DB module contains learned local feature-based object models representing a 3D object as a set of views.  ... 
doi:10.1155/2015/465290 pmid:26457323 pmcid:PMC4589637 fatcat:uf6pcpzperaalavfhh2kjtzgse
« Previous Showing results 1 — 15 out of 1,045,029 results