Local Features and Kernels for Classication of Texture and Object Categories: A Comprehensive Study

By Jianguo Zhang, Svetlana Lazebnik, Cordelia Schmid. 2006

529 citations recorded by the source corpus.

Subfields: 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 Computer Vision and Pattern Recognition

Citations in the corpus

Sharp analysis of low-rank kernel matrix approximations (2012, neutral): “Kernel methods, such as the support vector machine or kernel ridge regression, are now widely used in many areas of science and engineering, such as computer vision or bioinformatics (see, e.g., Sch¨olkopf et al., 2004; Zhang et al., 2007).”
Approximating CNNs with Bag-of-local-Features models works surprisingly well on ImageNet (2019, methodology): “Many successful image classiﬁcation models have been based on this pipeline (Csurka et al., 2004; Jurie & Triggs, 2005; Zhang et al., 2007; Lazebnik et al., 2006), see O’Hara & Draper (2011) for an up-to-date overview.”
Modeling Spatial-Temporal Clues in a Hybrid Deep Learning Framework for Video Classification (2015, methodology): “For example, in [53], Zhang et al. computed non-linear kernels for each feature separately, and then fused the kernels for model training.”
Sliced Wasserstein Kernels for Probability Distributions (2015, methodology): “…beneﬁts of using the transport distances out-lined above, and given the ﬂexibility and power of kernel-based methods [16], several methods using transport-related distances in constructing kernel matrices have been described with applications in computer vision, and EEG data classiﬁcation [41, 10].”
Affordance Prediction via Learned Object Attributes (2011, background): “The multi-channel χ 2 kernel has been shown to work well for the classiﬁcation of image data, where multiple, unrelated, features are extracted and combined [31, 32].”
Weakly Supervised Learning of Interactions between Humans and Objects (2012, unknown): “Moreover, it takes further advantage of using a robust human detector, such as the one in sec.”
Improving the Fisher Kernel for Large-Scale Image Classification (2010, background): “…detectors, descriptors and spatial pyramids (where a combination is often referred to as a “channel”), to train one classiﬁer per channel and then to combine the output of the classiﬁers [5–7 , An important limitation of such approaches is their scalability to large quantities of training images.”
Locally-Transferred Fisher Vectors for Texture Classification (2017, methodology): “While the earlier studies have mainly used the bag-of-words (BOW) model and its variations [14, 11, 34, 13, 30, 19, 18], encoding via Fisher vectors (FV) has become the dominant approach in texture classification [21, 6, 25, 7].”
Weakly Supervised Localization and Learning with Generic Knowledge (2012, unknown): “Many approaches are based on a bag-of-word model for the entire image [Dork´o and Schmid, 2005, Zhang et al, 2007].”
Learning person-object interactions for action recognition in still images (2011, methodology): “We build on the locally orderless statistical representations based on spatial pyramids [28] and bag-of-features models [9, 16, 34], which have demonstrated excellent performance on a range of scene [28], object [22, 36, 41] and action [11] recognition tasks.”