Covariance discriminative learning: A natural and efficient approach to image set classification

By Ruiping Wang, Huimin Guo, Larry S. Davis, Qionghai Dai. 2012

474 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

When Kernel Methods Meet Feature Learning: Log-Covariance Network for Action Recognition From Skeletal Data (2017, methodology): “Comparative analysis In a broad experimental validation of Log-COV-Net, we considered the sparse coding with dictionary learning for SPD matrices of [13], the covariance discriminative learning framework of [39], and the dimensionality reduction technique of [12] for SPD matrices.”
Attention-Aware Deep Reinforcement Learning for Video Face Recognition (2017, methodology): “In previous works, many models have been used for manifold modeling such as affine hull [6], SPD models [18, 20], Grassmann manifolds [21] and n-order statistics [27, 38].”
Scene-Level Geographic Image Classification Based on a Covariance Descriptor Using Supervised Collaborative Kernel Coding (2016, methodology): “Here, we adopt the idea of Ruiping Wang [22] and compute the distance of two covds C1 and C2 using log-Euclidean distance [25,26]: d(C1, C2) = ||logm(C1)− logm(C2)||F (2) where logm is the logarithm computation of the matrix and || · ||F denotes the Frobenius norm.”
Bregman Divergences for Infinite Dimensional Covariance Matrices (2014, methodology): “We also provide the results of the PLS-based Covariance Discriminant Learning (CDL) technique of [25], which can”
Beyond Covariance: Feature Representation with Nonlinear Kernel Matrices (2015, methodology): “, λ = 10) to the matrix as in the literature [27].”
Distribution Consistency Based Covariance Metric Networks for Few-Shot Learning (2019, methodology): “However, unlike the covariance representation of a general image set (Wang et al. 2012), there are very few images (e.”
A Riemannian Network for SPD Matrix Learning (2016, methodology): “The comparing state-of-the-art SPD matrix learning methods are Covariance Discriminative Learning (CDL) (Wang et al. 2012), Log-Euclidean Metric Learning (LEML) (Huang et al. 2015b) and SPD Manifold Learning (SPDML) (Harandi, Salzmann, and Hartley 2014) that uses afﬁne-invariant metric (AIM)…”
From Manifold to Manifold: Geometry-Aware Dimensionality Reduction for SPD Matrices (2014, methodology): “In addition to these methods, we also provide the results of the PLS-based Covariance Discriminant Learning (CDL) technique of [22], as well as of the state-of-the-art baselines of each specific dataset.”
Projection Metric Learning on Grassmann Manifold with Application to Video based Face Recognition (2015, methodology): “On this dataset, there are several state-of-the-art methods [39, 37, 38, 29, 17, 20] with other kinds of set modeling or other kinds of classifiers.”
Learning Expressionlets on Spatio-temporal Manifold for Dynamic Facial Expression Recognition (2014, methodology): “To characterize the correlations and variations among low-level features within each mode, the expressionlet is represented as the covariance matrix of the feature set in a statistic manner, which also makes it robust to local misalignment [25, 11, 29].”
A Comprehensive Look at Coding Techniques on Riemannian Manifolds (2018, methodology): “Another example of a structured descriptor is the region covariance descriptor (RCovD) successfully used in human detection [36], texture classification [9], human head pose estimation [35], and face recognition [18], [19], [38].”
Log-Euclidean Metric Learning on Symmetric Positive Definite Manifold with Application to Image Set Classification (2015, methodology): “For verification task on YTF, since RSR/LEK apply sparse representation based classifier, and CDL-PLS adopts a regression based classifier, they fail to work on this dataset and thus we do not report their results on this dataset.”
Log-Hilbert-Schmidt metric between positive definite operators on Hilbert spaces (2014, methodology): “(30)”
A Novel Space-Time Representation on the Positive Semidefinite Cone for Facial Expression Recognition (2017, methodology): “Speciﬁcally, given two matrices G 1 and G 2 in S + (2 , n ) : (1) as proposed in [42], we used d P n that was deﬁned in Eq.”
Quality Aware Network for Set to Set Recognition (2017, methodology): “The 2nd order statistics is used in [32, 43] in assuming that samples follow Gaussian distribution.”
Approximate Log-Hilbert-Schmidt Distances between Covariance Operators for Image Classification (2016, result): “5) CDL [32] 41.”
Multi-manifold deep metric learning for image set classification (2015, methodology): “For the YTC dataset, we employed the five fold cross validation strategy by following the same setting in [11, 26, 27, 34, 35].”
Deep Reconstruction Models for Image Set Classification (2015, methodology): “Image set classification methods commonly model the appearance variability information within images of a set on a geometric surface such as an image set modeled by a subspace [4], [15], a combination of subspaces [5], [16], a point on the Grassmannian [8] or Lie Group [10] of Riemannian manifold.”
Online action recognition using covariance of shape and motion (2014, methodology): “One of the possible extensions of this work could be deriving a more suitable classifier for our problem using either the Riemannian kernel [28] or the Riemannian pseudo-kernel [20].”