Filters








1,894 Hits in 6.5 sec

Compressive Video Classification In A Low-Dimensional Manifold With Learned Distance Metric

Jean-Luc Starck, Grigorios Tsagkatakis, P. Tsakalides, George Tzagkarakis
2012 Zenodo  
Publication in the conference proceedings of EUSIPCO, Bucharest, Romania, 2012  ...  A distance metric learning approach followed the initial dimensionality reduction in the compressed domain, to learn the relative distances between the training samples in a supervised way by exploiting  ...  We overcome the above drawbacks by employing a distance metric learning (DML) combined with a manifold learning (ML) approach.  ... 
doi:10.5281/zenodo.52363 fatcat:zvnof4sz6zfg3efqu6xl2be4ne

Multiple Manifolds Metric Learning with Application to Image Set Classification [article]

Rui Wang, Xiao-Jun Wu, Kai-Xuan Chen, Josef Kittler
2018 arXiv   pre-print
A metric Learning method has been devised to embed these kernel spaces into a lower dimensional common subspace for classification.  ...  In image set classification, a considerable advance has been made by modeling the original image sets by second order statistics or linear subspace, which typically lie on the Riemannian manifold.  ...  Riemannian distance metric [11, 13, 10, 12] , and then learn a map to transform it to a low dimensional one.  ... 
arXiv:1805.11918v1 fatcat:yrfxbnvhunaczcsmbdhojol6zq

Projection Metric Learning on Grassmann Manifold with Application to Video based Face Recognition

Zhiwu Huang, Ruiping Wang, Shiguang Shan, Xilin Chen
2015 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)  
To overcome such limitations, we propose a novel method to learn the Projection Metric directly on Grassmann manifold rather than in Hilbert space.  ...  In video based face recognition, great success has been made by representing videos as linear subspaces, which typically lie in a special type of non-Euclidean space known as Grassmann manifold.  ...  a low-dimensional SPD manifold.  ... 
doi:10.1109/cvpr.2015.7298609 dblp:conf/cvpr/HuangWSC15 fatcat:xcnbavicpbfiva3rvhy6ppmuh4

Grassmannian Discriminant Maps (GDM) for Manifold Dimensionality Reduction with Application to Image Set Classification [article]

Rui Wang, Xiao-Jun Wu, Kai-Xuan Chen, Josef Kittler
2022 arXiv   pre-print
In image set classification, a considerable progress has been made by representing original image sets on Grassmann manifolds.  ...  and metric learning on Grassmann manifold to improve performance.  ...  Huang et al. in [1] proposed a Grassmann manifold dimensionality reduction method, which jointly performs the process of mapping and metric learning, and generates a low dimensional, more discriminative  ... 
arXiv:1806.10830v2 fatcat:odvxpnxpc5gy5h3mzkyjpnsrre

Viewpoint Manifolds for Action Recognition

Richard Souvenir, Kyle Parrigan
2009 EURASIP Journal on Image and Video Processing  
In this paper, we present a framework for learning a compact representation of primitive actions (e.g., walk, punch, kick, sit) that can be used for video obtained from a single camera for simultaneous  ...  Using our method, which models the low-dimensional structure of these actions relative to viewpoint, we show recognition rates on a publicly available dataset previously only achieved using multiple simultaneous  ...  In [23] , a framework is presented for selecting image distance metrics for use with manifold learning.  ... 
doi:10.1155/2009/738702 fatcat:h4anyadp3be6rojbcbur4oy4pa

Hybrid Euclidean-and-Riemannian Metric Learning for Image Set Classification [chapter]

Zhiwu Huang, Ruiping Wang, Shiguang Shan, Xilin Chen
2015 Lecture Notes in Computer Science  
spaces into high dimensional Hilbert spaces and then jointly learn hybrid metrics with discriminant constraint.  ...  We propose a novel hybrid metric learning approach to combine multiple heterogenous statistics for robust image set classification.  ...  It has 1,910 video clips of 47 subjects collected from YouTube. Most clips contains hundreds of frames, which are often low resolution and highly compressed with noise and low quality.  ... 
doi:10.1007/978-3-319-16811-1_37 fatcat:ihjhwmxo5rbglfwknfvbovj2jq

Probability Distribution-based Dimensionality Reduction on Riemannian Manifold of SPD Matrices

Jieyi Ren, Xiao-jun Wu
2020 IEEE Access  
Most video sequences are low resolution and recorded at a high compression ratio with different numbers of frames (vary from 8 to 400).  ...  [2] and framebased video classification [15] .  ...  In the future, we intend to extend our probability distribution-based approach to the unsupervised scenario.  ... 
doi:10.1109/access.2020.3017234 fatcat:w2brc7hqbfdqtgei63iseojz2a

A Compact and Discriminative Face Track Descriptor

Omkar M. Parkhi, Karen Simonyan, Andrea Vedaldi, Andrew Zisserman
2014 2014 IEEE Conference on Computer Vision and Pattern Recognition  
Our goal is to learn a compact, discriminative vector representation of a face track, suitable for the face recognition tasks of verification and classification.  ...  Second, the descriptor is compact due to discriminative dimensionality reduction, and it can be further compressed using binarization.  ...  Binary compression While the low-rank metric learning method of Sect. 3.2 is already capable of achieving a very good compression factor, for large scale applications the goal is to further decrease the  ... 
doi:10.1109/cvpr.2014.219 dblp:conf/cvpr/ParkhiSVZ14 fatcat:kefzq3qrv5f7bkjmebvxfwyktu

Comparative Evaluation of Action Recognition Methods via Riemannian Manifolds, Fisher Vectors and GMMs: Ideal and Challenging Conditions [chapter]

Johanna Carvajal, Arnold Wiliem, Chris McCool, Brian Lovell, Conrad Sanderson
2016 Lecture Notes in Computer Science  
Despite recent advancements for handling manifolds, manifold based techniques obtain the lowest performance and their kernel representations are more unstable in the presence of challenging conditions.  ...  Moreover, FV best deals with moderate scale and translation changes.  ...  The space of m-dimensional LS in R n can be viewed as a special case of Riemannian manifolds, known as Grassmann manifolds [47] .  ... 
doi:10.1007/978-3-319-42996-0_8 fatcat:yxeuj3sxnzf6rghujwi5l7mwiu

Event monitoring via local motion abnormality detection in non-linear subspace

Ioannis Tziakos, Andrea Cavallaro, Li-Qun Xu
2010 Neurocomputing  
We use motion vectors extracted over a Region of Interest (ROI) as features and a non-linear, graph-based manifold learning algorithm coupled with a supervised novelty classifier to label segments of a  ...  video sequence.  ...  Using this metric, Isomap can achieve adequate results to learn (unfold) manifolds that have a non-linear global structure.  ... 
doi:10.1016/j.neucom.2009.10.028 fatcat:if6u5a3ag5ffvaegl2vdvxhzkm

A Framework for Short Video Recognition Based on Motion Estimation and Feature Curves on SPD Manifolds

Xiaohe Liu, Shuyu Liu, Zhengming Ma
2022 Applied Sciences  
This paper presents a new short video recognition algorithm framework that transforms a short video into a family of feature curves on symmetric positive definite (SPD) manifold as the basis of recognition  ...  A significant feature of short video is that there are few switches of scenes in short video, and the target (e.g., the face of the key person in the short video) often runs through the short video.  ...  DALG DALG aims to learn a mapping that transforms a high-dimensional Lie group (LG) into a more discriminative, low-dimensional one.  ... 
doi:10.3390/app12094669 fatcat:52hnmqfw6zgkjezdiraeytouq4

Learning the viewpoint manifold for action recognition

Richard Souvenir, Justin Babbs
2008 2008 IEEE Conference on Computer Vision and Pattern Recognition  
In this paper, we present a framework for learning a compact representation of primitive actions (e.g., walk, punch, kick, sit) that can be used for video obtained from a single camera for simultaneous  ...  Using our method, which models the low-dimensional structure of these actions relative to viewpoint, we show recognition rates on a publicly available data set previously only acheieved using multiple  ...  learning how the appearance of an action varies as the viewpoint changes by learning a low dimensional representation of action primitives using manifold learning.  ... 
doi:10.1109/cvpr.2008.4587552 dblp:conf/cvpr/SouvenirB08 fatcat:dvxc4hgrgffzxauhlpukba665y

Face recognition on large-scale video in the wild with hybrid Euclidean-and-Riemannian metric learning

Zhiwu Huang, Ruiping Wang, Shiguang Shan, Xilin Chen
2015 Pattern Recognition  
With a LogDet divergence based objective function, the hybrid kernels are then fused by our hybrid metric learning framework, which can efficiently perform the fusing procedure on large-scale videos.  ...  In this paper, we propose a novel Hybrid Euclidean-and-Riemannian Metric Learning (HERML) method to fuse multiple statistics of image set.  ...  Most clips contain hundreds of frames, which are often of low resolution and highly compressed with noise and low quality.  ... 
doi:10.1016/j.patcog.2015.03.011 fatcat:sqr2cvioajdczbfmno3mnfonha

Deep Manifold Learning of Symmetric Positive Definite Matrices with Application to Face Recognition

Zhen Dong, Su Jia, Chi Zhang, Mingtao Pei, Yuwei Wu
2017 PROCEEDINGS OF THE THIRTIETH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE AND THE TWENTY-EIGHTH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE  
In this paper, we aim to construct a deep neural network which embeds high dimensional symmetric positive definite (SPD) matrices into a more discriminative low dimensional SPD manifold.  ...  Finally, we complete the construction of the deep neural network for SPD manifold learning by stacking the two layers.  ...  First, learning discriminative representations in new learned low dimensional SPD space brings low computational cost.  ... 
doi:10.1609/aaai.v31i1.11232 fatcat:bfd24zkkrve27e2gw3td7vxixm

Some new directions in graph-based semi-supervised learning

Xiaojin Zhu, Andrew B. Goldberg, Tushar Khot
2009 2009 IEEE International Conference on Multimedia and Expo  
(ii) online semi-supervised learning that learns incrementally with low computation and memory needs; and (iii) learning spectrally sparse but non-smooth labels with compressive sensing.  ...  is restricted to live on a single manifold; (2) learning must happen in batch mode; and (3) the target label is assumed smooth on the manifold.  ...  In particular, Belkin et al. generalize graph-based learning to the manifold setting, where X is assumed to be a low dimensional manifold in R D , the labels y change smoothly on the manifold, and the  ... 
doi:10.1109/icme.2009.5202789 dblp:conf/icmcs/ZhuGK09 fatcat:2v5ss5iowfewhbb7w4a5ysiqti
« Previous Showing results 1 — 15 out of 1,894 results