Filters








105,502 Hits in 5.3 sec

Image-to-Class Distance Metric Learning for Image Classification [chapter]

Zhengxiang Wang, Yiqun Hu, Liang-Tien Chia
2010 Lecture Notes in Computer Science  
Our I2C distance is adaptive to different class by combining with the learned metric for each class.  ...  Image-To-Class (I2C) distance is first used in Naive-Bayes Nearest-Neighbor (NBNN) classifier for image classification and has successfully handled datasets with large intra-class variances.  ...  We name our method as I2CDML, short for Image-To-Class distance metric learning.  ... 
doi:10.1007/978-3-642-15549-9_51 fatcat:m757vokwnbgdzewgub2lmd7yqq

Aggregated Distance Metric Learning (ADM) for Image Classification in Presence of Limited Training Data [chapter]

Gaoyu Xiao, Anant Madabhushi
2011 Lecture Notes in Computer Science  
The focus of image classification through supervised distance metric learning is to find an appropriate measure of similarity between images.  ...  We present a novel image classification method called aggregated distance metric (ADM) learning for situations where the training image data are limited.  ...  During image classification, all training images are used to learn distance metric d, which is then used to compute the distance between a test image t and each training image.  ... 
doi:10.1007/978-3-642-23626-6_5 fatcat:zsaihhwqnrc35bosij34xlcjhe

Deep Metric Learning for Few-Shot Image Classification: A Review of Recent Developments [article]

Xiaoxu Li, Xiaochen Yang, Zhanyu Ma, Jing-Hao Xue
2022 arXiv   pre-print
In this paper, we provide an up-to-date review of deep metric learning methods for few-shot image classification from 2018 to 2022 and categorize them into three groups according to three stages of metric  ...  One main solution to few-shot image classification is deep metric learning.  ...  In the case of few-shot classification, the metric is learned on the base dataset; query images of the novel class are classified by computing their distances to novel support images with respect to the  ... 
arXiv:2105.08149v2 fatcat:yxsvfdspbrhfpcrzgnny27vgjy

From Point to Set: Extend the Learning of Distance Metrics

Pengfei Zhu, Lei Zhang, Wangmeng Zuo, David Zhang
2013 2013 IEEE International Conference on Computer Vision  
Most of the current metric learning methods are proposed for point-to-point distance (PPD) based classification.  ...  In this paper, we extend the PPD based Mahalanobis distance metric learning to PSD and SSD based ones, namely point-to-set distance metric learning (PSDML) and set-to-set distance metric learning (SSDML  ...  With metric learning, PSDML can further utilize the class label to learn a discriminative metric for the pointto-set distance, and thus may result in better classification performance.  ... 
doi:10.1109/iccv.2013.331 dblp:conf/iccv/ZhuZZZ13 fatcat:ypbr75hvqbhozjdjadopfdleku

Distance Metric with Kullback–Leibler Divergence for Classification

Dongyun Qian, Huifeng Jin
2017 International Journal of Signal Processing, Image Processing and Pattern Recognition  
Therefore, learning an excellent distance metric is of vital importance but challenging. Up to now, many distance metric learning methods have been proposed using various techniques.  ...  DMKD defines the divergences between two different classes using KL-divergence. Then, a constraint is added to ensure DMKD to obtain a feasible distance metric.  ...  Acknowledgments The authors would like to thank the reviewers for their comments which has improved the quality of the work. This work is supported by the project of the Domestic Visitor  ... 
doi:10.14257/ijsip.2017.10.7.14 fatcat:xaqzjvnmava6jacjo7ohou75ue

Large Scale Metric Learning for Distance-Based Image Classification on Open Ended Data Sets [chapter]

Thomas Mensink, Jakob Verbeek, Florent Perronnin, Gabriela Csurka
2013 Advanced Topics in Computer Vision  
For the NCM classifier we introduce a new metric learning approach, and we also introduce an extension to allow for richer class representations.  ...  Experimentally we study the generalization performance to classes that were not used to learn the metrics.  ...  Other methods aim at learning metrics for verification problems and essentially learn binary classifiers that threshold the learned distance to decide whether two images belong to the same class or not  ... 
doi:10.1007/978-1-4471-5520-1_9 dblp:series/acvpr/MensinkVP13 fatcat:ntz7nljgrbdlxbyuss4f3dbdee

Metric Learning for Large Scale Image Classification: Generalizing to New Classes at Near-Zero Cost [chapter]

Thomas Mensink, Jakob Verbeek, Florent Perronnin, Gabriela Csurka
2012 Lecture Notes in Computer Science  
We are interested in large-scale image classification and especially in the setting where images corresponding to new or existing classes are continuously added to the training set.  ...  We learn metrics on the ImageNet 2010 challenge data set, which contains more than 1.2M training images of 1K classes.  ...  Performance of 1,000-way classification among test images of 200 classes not used for metric learning, and control setting with metric learning using all classes.  ... 
doi:10.1007/978-3-642-33709-3_35 fatcat:z62gr264vjd43pf4rah5e433la

Distance-Based Image Classification: Generalizing to New Classes at Near-Zero Cost

T. Mensink, J. Verbeek, F. Perronnin, G. Csurka
2013 IEEE Transactions on Pattern Analysis and Machine Intelligence  
To this end we consider two distance-based classifiers, the k-nearest neighbor (k-NN) and nearest class mean (NCM) classifiers, and introduce a new metric learning approach for the latter.  ...  Experimentally we study the generalization performance to classes that were not used to learn the metrics.  ...  TABLE 6 : 6 Results for 1,000-way classification among test images of 200 classes not used for metric learning, and control setting when learned on all classes. metric is trained on all 1,000 classes.  ... 
doi:10.1109/tpami.2013.83 pmid:24051724 fatcat:vjoafmof4nfozbxoxcuns6sjvu

Learning a metric for class-conditional KNN [article]

Daniel Jiwoong Im, Graham W. Taylor
2016 arXiv   pre-print
An empirical evaluation on classification and retrieval tasks demonstrates that our proposed method clearly outperforms existing learned distance metrics across a variety of image and non-image datasets  ...  To extend its applicability outside of image-based domains, we propose to learn a metric which captures perceptual similarity.  ...  CONCLUSION Inspired by Neighbourhood Components Analysis which optimizes a distance metric for 1-nearest neighbour classification, we show how to learn a distance metric optimized for class-conditional  ... 
arXiv:1607.03050v1 fatcat:hgahsjviyzawzb7dwww5ufdfpm

Maximum Margin Multi-Instance Learning

Hua Wang, Heng Huang, Farhad Kamangar, Feiping Nie, Chris H. Q. Ding
2011 Neural Information Processing Systems  
C2B distance by introducing the class specific distance metrics and the locally adaptive significance coefficients.  ...  To tackle this, in this paper we approach MIL problems from a new perspective using the Class-to-Bag (C2B) distance, which directly assesses the relationships between the classes and the bags.  ...  Our task is to learn class specific distance metrics M k and significance coefficients w j k from the training data, with which we compute the C2B distances from the classes to a query image X for classification  ... 
dblp:conf/nips/WangHKND11 fatcat:tojbrxcvw5fgfd7o5wru4hnh6u

Convex Optimizations for Distance Metric Learning and Pattern Classification [Applications Corner

Kilian Weinberger, Fei Sha, Lawrence Saul
2010 IEEE Signal Processing Magazine  
Algorithms for distance metric learning attempt to improve on ad-hoc or default choices of distance metrics.  ...  METRIC LEARNING FOR NEAREST-NEIGHBOR CLASSIFICATION Convex Optimizations for Distance Metric Learning and Pattern Classification by the majority label of their kNN in the training set.  ...  Looking forward, we anticipate many such applications given the ubiquitous role of distance metrics in both nonparametric and parametric models of classification.  ... 
doi:10.1109/msp.2010.936013 fatcat:ffx2slovabbelmoil7nz2iptia

Improving Image Distance Metric Learning by Embedding Semantic Relations [chapter]

Fang Wang, Shuqiang Jiang, Luis Herranz, Qingming Huang
2012 Lecture Notes in Computer Science  
Learning a proper distance metric is crucial for many computer vision and image classification applications.  ...  To overcome this problem, integrating concrete semantic relations of images into the distance metric learning procedure can be a useful solution.  ...  In image classification, most distance metric learning approaches try to employ the information contained in visual features and class labels, so that an appropriate Mahalanobis distance metrics are obtained  ... 
doi:10.1007/978-3-642-34778-8_39 fatcat:lgmmku6zbbcutcm7g5d5wiuquq

Reconstruction Regularized Deep Metric Learning for Multi-label Image Classification [article]

Changsheng Li and Chong Liu and Lixin Duan and Peng Gao and Kai Zheng
2020 arXiv   pre-print
In this paper, we present a novel deep metric learning method to tackle the multi-label image classification problem.  ...  To capture the relationships between image features and labels, we aim to learn a two-way deep distance metric over the embedding space from two different views, i.e., the distance between one image and  ...  In [37] , a novel metric learning framework was presented to integrate class-specific distance metrics and explicitly take into account inter-class correlations for multilabel prediction.  ... 
arXiv:2007.13547v1 fatcat:6slei7dv25ej5hcs5vqsztc7uq

Multi-label learning by Image-to-Class distance for scene classification and image annotation

Zhengxiang Wang, Yiqun Hu, Liang-Tien Chia
2010 Proceedings of the ACM International Conference on Image and Video Retrieval - CIVR '10  
In this paper, we propose a multi-label learning framework based on Imageto-Class (I2C) distance, which is recently shown useful for image classification.  ...  For each image, we constrain its weighted I2C distance to the relevant class to be much less than its distance to other irrelevant class, by the use of a margin in the optimization problem.  ...  The I2C distance is firstly introduced in [2] for classifying multi-class single label images and achieves excellent performance for such multi-class classification problems.  ... 
doi:10.1145/1816041.1816060 dblp:conf/civr/WangHC10 fatcat:cpobbeyow5dflg27ii5taakzzq

Revisiting Metric Learning for Few-Shot Image Classification [article]

Xiaomeng Li, Lequan Yu, Chi-Wing Fu, Meng Fang, Pheng-Ann Heng
2020 arXiv   pre-print
Once trained, our network is able to extract discriminative features for unseen novel categories and can be seamlessly incorporated with a non-linear distance metric function to facilitate the few-shot  ...  generalize to unseen classes.  ...  Non-linear Distance Metric Learning Furthermore, we adopt the non-linear distance metric module [33] to learn to compare the embedded features in few-shot classification.  ... 
arXiv:1907.03123v2 fatcat:swfxz3qjnbczdlkxzhe3e7d35q
« Previous Showing results 1 — 15 out of 105,502 results