Kernel Codebooks for Scene Categorization

By Jan van Gemert, Jan-Mark Geusebroek, Cor J. Veenman, Arnold M. W. Smeulders. 2008

239 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

Describing Textures in the Wild (2013, methodology): “1), soft quantization with a kernel in KCB [37], Localityconstrained Linear Coding (LLC) [41], or specialized quantization schemes, such as mapping LBPs to uniform patterns [25] (LBP; we use the rotation invariant multipleradii version of [24] for comparison purposes).”
From BoW to CNN: Two Decades of Texture Representation for Texture Classification (2018, methodology): “• Soft Voting [2, 196, 232] Assigns each feature to multiple codewords; Does not minimize reconstruction error.”
Learning Feature Representations with K-Means (2012, methodology): “When we are unhappy with performance, searching for better choices of these parameters can often be more beneficial than trying to modify our learning algorithm [11,18].”
Linear spatial pyramid matching using sparse coding for image classification (2009, methodology): “Following our own baseline, the Linear ScSPM algorithm again achieves much better performance than KSPM and KC [25].”
Locality-constrained Linear Coding for image classification (2010, unknown): “10 Gemert [8] - - - - - 64.”
A Review on Human Activity Recognition Using Vision-Based Method (2017, unknown): “(2008) [60], 586; [61], 761 Spase coding (SPC) Yang et al.”
Large-scale web video event classification by use of Fisher Vectors (2013, methodology): “Although modifications to BoW can help it capture more information, such as assigning different weights to codewords and soft assignment of codewords[16], GMM incorporates them naturally.”
A Comparative Study of Encoding, Pooling and Normalization Methods for Action Recognition (2012, methodology): “The soft-assignment encoding has two forms: assigning each feature to all codeword (SA-all) [15] and assigning each feature to k nearest neighborhood (SA-k) [1].”
Learning a discriminative dictionary for sparse coding via label consistent K-SVD (2011, methodology): “We evaluate our approach using spatial pyramid features and compare the results with K-SVD [1], D-KSVD [33], SRC [28] and other state-of-art approaches [32, 15, 11, 2, 14, 24, 8, 29, 27].”
k-Sparse Autoencoders (2013, background): “For example, while in k - means, it is natural to have a hard-assignment of the points to the nearest cluster in the encoding stage, it has been shown in (Van Gemert et al., 2008) that soft assignments result in better classiﬁcation performance.”
Understanding bag-of-words model: a statistical framework (2010, background): “Tuytelaars and Schmid (2007) propose a quantization method that discretizes a feature space by a regular lattice. van Gemert et al.(2008) use kernel density estimation to avoid the problem of 'codeword uncertainty' and 'codeword plausibility'.”
In defense of soft-assignment coding (2011, methodology): “In this paper, we follow the original softassignment coding scheme [6, 14] to define d(xi,bl) as squared Euclidean distance but other distances can certainly be used.”
Aggregating Continuous Word Embeddings for Information Retrieval (2013, methodology): “These improvements include the soft assignment of descriptors to visual words (Farquhar et al., 2005; Philbin et al., 2008; Gemert et al., 2008) or the use of more advanced coding techniques than vector quantization such as sparse coding (Yang et al., 2009) or locality-constrained linear coding…”
Discriminatively Trained Dense Surface Normal Estimation (2014, methodology): “Each feature was clustered into 512 visual words using k-means clustering, and for each pixel a soft assignment for 8 nearest cluster centres is calculated using distance-based exponential kernel weighting [49].”