Image Classification Using Super-Vector Coding of Local Image Descriptors

By Xi Zhou, Kai Yu, Tong Zhang, Thomas S. Huang. 2010

581 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

ImageNet Large Scale Visual Recognition Challenge (2014, methodology): “linear coding representations (Zhou et al., 2010; Wang et al., 2010) and a stochastic SVM.”
Understanding deep image representations by inverting them (2014, background): “Most image understanding and computer vision methods build on image representations such as textons [17], histogram of oriented gradients (SIFT [20] and HOG [4]), bag of visual words [3][27], sparse [37] and local coding [34], super vector coding [40], VLAD [10], Fisher Vectors [23], and, lately, de”
Understanding Image Representations by Measuring Their Equivariance and Equivalence (2014, background): “2010), super vector coding (Zhou et al. 2010), VLAD (Jégou et al.”
Visualizing Deep Convolutional Neural Networks Using Natural Pre-images (2015, background): “2010), super vector coding (Zhou et al. 2010), VLAD (Jégou et al.”
Bag of visual words and fusion methods for action recognition: Comprehensive study and good practice (2014, neutral): “Super Vector Coding (SVC) [65] is a simple version of LTC.”
Deep learning for visual understanding (2017, unknown): “Autoencoder-based Methods [ Sparse Autoencoder [28] Denoising Autoencoder [29] Contractive Autoencoder [30] Sparse Coding-based Methods  Sparse Coding SPM [31] Laplacian Sparse Coding [32] Local Coordinate Coding [33] Super-Vector Coding [34]”
Deep convolutional filter banks for texture recognition and segmentation (2014, methodology): “Texton distributions were the early versions of “bag-of-words” representations, a dominant approach in recognition in the early 2000s, since then improved by new pooling schemes such as soft-assignment [26, 39, 45] and Fisher Vectors (FVs) [31].”
High-dimensional signature compression for large-scale image classification (2011, background): “PASCAL VOC 2009 winners mention in [10] (slide 27) that their GMM-based [39] and sparse-coding-based [37] representations have respectively 655K and 2M dimensions.”
Image Reconstruction from Bag-of-Visual-Words (2014, methodology): “Actually, BoVW has been extended to state-of-the-art hand-engineered features, such as VLAD [41], Super Vector [10], and Fisher Vector [9].”
Evaluating knowledge transfer and zero-shot learning in a large-scale setting (2011, methodology): “BoW: bag of visual words, SPM: spatial pyramid matching [15], LLC: locality-constrained linear coding [30], Fisher vector [19], SVC: Super-Vector Coding [ 31 ], Lbp: local binary patterns, SGD: stochastic gradient decent [3], ASGD: averaging SGD [16].”
Author manuscript, published in "International Journal of Computer Vision (2013)" International Journal of Computer Vision manuscript No. (will be inserted by the editor) Image Classification with the Fisher Vector: Theory and Practice (2013, unknown): “We see in Figure 4 that there is an increase in performance from 0-order (BoV) to the combination of 0-order and 1st-order statistics (similar to the statistics used in the SV(Zhou et al, 2010)), and even further when the 1st-order and 2nd-order statistics are combined.”
A codebook-free and annotation-free approach for fine-grained image categorization (2012, methodology): “Some recent work [28] [33] uses sparse coding to obtain more accurate encodings of image patches, and have demonstrated performance improvement on many image classification tasks.”
Large-Scale Weakly Supervised Object Localization via Latent Category Learning (2015, methodology): “After the encoding, average pooling [46] is used on the encoding of all the T regions to construct the BoW image representation, as shown in Fig.”
Large Scale Visual Recognition (2012, background): “One line of work is concerned with developing effective feature representations [95, 96, 97, 98] and achieving state of the art performances.”
All About VLAD (2013, methodology): “Note, VLAD is similar to other descriptors that record residuals such as Fisher vectors [11] and super-vector coding [20].”
Selective Search for Object Recognition (2013, unknown): “In image classification, features of this size are already used [25, 37].”