Classification using intersection kernel support vector machines is efficient

By Subhransu Maji, Alexander C. Berg, Jitendra Malik. 2008

1,111 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Efficiency & Compression

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Efficiency & Compression

Citations in the corpus

Histogram of Oriented Principal Components for Cross-View Action Recognition (2014, methodology): “For classification, we use an SVM classifier with the histogram intersection kernel [54].”
HOPC: Histogram of Oriented Principal Components of 3D Pointclouds for Action Recognition (2014, methodology): “For classification, we use SVM with the histogram intersection kernel [16].”
DeepID-Net: multi-stage and deformable deep convolutional neural networks for object detection (2014, methodology): “The widely used classification approaches include various boosting classifiers [14, 15, 56], linear SVM [9], histogram intersection kernel SVM [34], latent SVM [20], multiple kernel SVM [53], structural SVM [65], and probabilistic models [3, 36].”
From BoW to CNN: Two Decades of Texture Representation for Texture Classification (2018, methodology): “Fisher Vector (FV) Based Methods Require a small codebook; Very high dimension; Combines with efﬁcient linear SVM. • FV [178] GMM-based; Encodes higher order statistics; Efﬁcient to compute.”
Measuring the Objectness of Image Windows (2011, methodology): “Excellent approximations to the histogram intersection kernel [39] and the 2 kernel [51] have been”
Traffic sign recognition — How far are we from the solution? (2013, methodology): “4) Support Vector Machines Classifiers: Support Vector Machines (SVM) are a very popular technique for out of the box classification [22], [23].”
Efficient additive kernels via explicit feature maps (2010, methodology): “Compared to Maji et al. [ 15 ], our technique also has an edge on the testing efficiency.”
Unsupervised Feature Learning Via Spectral Clustering of Multidimensional Patches for Remotely Sensed Scene Classification (2015, methodology): “We perform the classification experiment with HIK SVM [52] once to obtain the final classification accuracy.”
Harmony potentials for joint classification and segmentation (2010, methodology): “We use a different SVM classifier with intersection kernel [16] for each label to obtain the classification scores.”
Optimizing 1-Nearest Prototype Classifiers (2013, methodology): “For example in [11] an approximation of the popular intersection kernel was described.”
Temporal Localization of Actions with Actoms (2013, neutral): “For efficiency reasons, we choose to use the intersection kernel [66].”
Spatial pyramid matching (2009, positive): “In fact, just as in Grauman and Darrell (2005), the computational complexity of the kernel is linear in the number of features (more recently, Maji et al. (2008) have shown that the histogram intersection kernel is amenable to further optimizations, leading to extremely fast support vector…”
High-dimensional signature compression for large-scale image classification (2011, background): “In [19], Maji et al. showed that the runtime cost of an intersection kernel (IK) SVM could be made independent of the number of support vectors.”
BING: Binarized normed gradients for objectness estimation at 300fps (2014, neutral): “Also, some other efficient classifiers [17] and approximate kernels [43, 51] have been proposed.”
Exploiting Feature and Class Relationships in Video Categorization with Regularized Deep Neural Networks (2015, background): “Several recent studies investigated this issue by proposing efficient classification methods [19], [20] or parallel computing strategies [21], [22].”
Building a better probabilistic model of images by factorization (2011, methodology): “Using a modified version of LIBSVM [4] that incorporates the fast approximate intersection kernel [20], we solve the SVM optimization problem, inputting the model’s latent representation for each digit.”
Multiscale Dynamic Graph Convolutional Network for Hyperspectral Image Classification (2019, methodology): “As a traditional yet important remote sensing technique, hyperspectral image classification has been intensively investigated and many related methods have been proposed, such as Bayesian methods [37], random forest [38], and kernel methods [39].”