Discriminative learned dictionaries for local image analysis

By Julien Mairal, Francis R. Bach, Jean Ponce, Guillermo Sapiro, Andrew Zisserman. 2008

818 citations recorded by the source corpus.

Subfields: Computational Mechanics, Computer Vision and Pattern Recognition

How researchers see this paper

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

In Computational Mechanics

In Computer Vision and Pattern Recognition

Citations in the corpus

Pedestrian Detection with Unsupervised Multi-stage Feature Learning (2012, background): “Following [30] many authors have used sparse dictionary learning to represent images [27, 1, 19].”
Fast Inference in Sparse Coding Algorithms with Applications to Object Recognition (2010, background): “However, application of these methods to vision problems ha s been limited due to prohibitive cost of calculating sparse representations for a given image [6] .”
Online Learning for Matrix Factorization and Sparse Coding (2009, neutral): “It has been shown by Bach et al. (2008) that using the ℓ 1 -norm only in such problems lead to trivial solutions when k is large enough.”
Moving Poselets: A Discriminative and Interpretable Skeletal Motion Representation for Action Recognition (2015, background): “[4] automatically learned a set of weights on the LDSs extracted from a larger set of body parts, and showed that most of the weights from smaller body parts are zero, suggesting that smaller parts might be redundant in representing body-level actions.”
What is the best multi-stage architecture for object recognition? (2009, background): “The main problem with the unsupervised approach is that the filters are learned independently of the task, although a few authors have proposed methods that combine unsupervised and supervised criteria to alleviate the problem [21, 27, 4].”
Sparse Dictionary Learning for Edit Propagation of High-Resolution Images (2014, methodology): “Sparse representations have also been applied to face recognition [26], image background modeling [3], and image classification [20].”
Robust Visual Tracking using 1 Minimization (2009, background): “Other applications include background subtraction [6], media recovery [15], texture segmentation [24], and lighting estimation [26], etc.”
Sparse subspace clustering (2009, neutral): “[19] uses a sparse representation to learn a dictionary for object recognition.”
Deep learning of feature representation with multiple instance learning for medical image analysis (2014, background): “Unsupervised feature learning [12, 13, 14, 15] is based on unlabeled data.”
Learning invariant features through topographic filter maps (2009, background): “…proposed in the literature and in this work we focus on the following convex formulation: This particular formulation has been extensively studied [19, 4, 2, 14, 17, 24, 16, 9], and it has also been extended to the case when the dictionary D is learned, thus adapting to the statistics of the input.”
Learning hierarchical sparse features for RGB-(D) object recognition (2014, background): “…in various applications such as image denoising (Elad and Aharon, 2006b), face image compression (Bryt and Elad, 2008), texture segmentation (Mairal et al., 2008a), and contour detection (Ren and Bo, 2012 Here, N is the size of the training data, the notation (cid:7) A (cid:7) F denotes the…”
Learning to Sense Sparse Signals: Simultaneous Sensing Matrix and Sparsifying Dictionary Optimization (2009, methodology): “We start by reviewing prior related work, following by the presentation of our proposed algorithm and the first results showing the computational and reconstruction advantages of this new approach.”
From BoW to CNN: Two Decades of Texture Representation for Texture Classification (2018, methodology): “…the class speciﬁc codebooks were optimized for reconstruction, but signiﬁcant improvements have been shown by optimizing for discriminative power instead [45, 139, 140], an approach which is, however, associated with high computational cost, especially when the number of texture classes C is large.”
Pixel-level image fusion with simultaneous orthogonal matching pursuit (2012, background): “Applications that benefit from the sparsity and overcompleteness include compression, regularization in inverse problems [29], image and video compression [30,31], computer vision tasks and so on [32].”
Submitted to Ieee Transactions on Pattern Analysis and Machine Intelligence 1 Multiscale Centerline Detection (neutral): “However, attention has recently shifted to classification based methods [2], [11], [12], [29], [31], [41], [47].”