Semi-Local Affine Parts for Object Recognition

By Svetlana Lazebnik, Cordelia Schmid, Jean Ponce. British Machine Vision Conference, 2004

304 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/6b59cea2f573ecae7f14adf47b984f1095a8c341

How researchers see this paper

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

In Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

3D generic object categorization, localization and pose estimation (2007, background): “This choice stems from the observation that larger regions of objects often carry more discriminative information in appearance and are more stable in their geometric relationships with other parts of the object [16].”
Foreground Focus: Unsupervised Learning from Partially Matching Images (2009, background): “2004; Winn and Jojic 2005; Chum and Zisserman 2007; Ling and Soatto 2007) and robust local feature representations (Lowe 2004; Agarwal and Triggs 2006; Lazebnik et al. 2004).”
Detecting Symmetry and Symmetric Constellations of Features (2006, background): “[13] also noted that clusters of features could be matched within an image to detect symmetries.”
A latent model of discriminative aspect (2009, background): “Usually, these multiple classes are compacted into a single model, assembled from local patches (which might have quite simple behavior), tied together by observation [25], with geometric reasoning [14], with statistical reasoning [18, 20], or with a combination [24].”
Automatic annotation of human actions in video (2009, background): “Our weakly supervised clustering is also related to the work on learning object models from weakly annotated images [4, 15].”
Unsupervised Learning of Invariant Feature Hierarchies with Applications to Object Recognition (2007, background): “This includes SIFT [14] features and their many derivatives, such as afﬁne-invariant patches [11].”
3DMatch: Learning Local Geometric Descriptors from RGB-D Reconstructions (2016, background): “While there is a wide range of low-level hand-crafted geometric feature descriptors that can be used for this task, they are mostly based on signatures derived from histograms over static geometric properties [19, 22, 28].”
Invariant Image-Based Species Classification of Butterflies and Reef Fish (2015, methodology): “The UIUC dataset [8] has images of 7 species and the images in Leeds butterfly dataset [17] belong to 10 species.”
Self-Supervised Representation Learning by Rotation Feature Decoupling (2019, background): “Many classical hand-crafted features like SIFT [34] and RIFT [29] for computer vision are insensitive to certain rotation transformations.”
Learning Models for Object Recognition from Natural Language Descriptions (2009, background): “To our knowledge, one previous work [16] has investigated visual recognition of butterfly species, by matching local invariant features to training images.”
Oriented Response Networks (2017, background): “SIFT descriptor [31] and its modification with affine-local regions [25] find the dominant orientation of a feature point, according to which statistics of local gradient directions of image intensities are accumulated to give a summarizing description of local image structures.”
Local Features and Kernels for Classification of Texture and Object Categories: A Comprehensive Study (2006, methodology): “, 2003) and affine deformations (Lazebnik et al., 2004).”
Object Recognition in the Geometric Era: A Retrospective (2006, background): “Aﬃne patches have also been exploited as parts in a new attack on the problem of generic object recognition [24, 44].”
Feature representation for statistical-learning-based object detection: A review (2015, unknown): “There are several generalizations that obtain feature representation in circular local regions, such as RIFT, GLOH, DAISY, C-HOG.”
Unsupervised Learning of Invariant Feature Hierarchies with Applications to Object Recognition (2008, background): “The feature vector Z indicates what feature is present in the input, while the transformation parameters U indicate where each feature is present in the input. patches [11].”
Large-scale Learning with SVM and Convolutional for Generic Object Categorization (2006, background): “The use of color, texture, and contours have been advocated in [16], the use of global appearance templates in [19, 18, 25], silhouettes and edge information [22, 29, 16, 5, 25], pose-invariant feature histograms [17, 6, 1], distinctive local features [26, 35, 33, 12], and image fragments [30].”