An Affine Invariant Interest Point Detector

By Krystian Mikolajczyk, Cordelia Schmid. European Conference on Computer Vision, 2002

1,701 citations recorded by the source corpus.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/9c7a96155f10f152cae0866102c061cdf6da02e8

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

One-shot learning of object categories (2006, background): “Individual objects are easier to handle, therefore, more progress has been made on efficient recognition [27], lighting-invariant [27], [28], and viewpoint-invariant [22], [31] representations and recognition.”
Robust wide-baseline stereo from maximally stable extremal regions (2004, methodology): “In [11], the approach was combined with Baumberg’s iteration to obtain an affine-invariant detector.”
Description of interest regions with local binary patterns (2009, background): “The number of correct matches is determined with the overlap error [30].”
Keypoint recognition using randomized trees (2006, methodology): “Our method does away with having to compute ad hoc descriptors, rectify patches, and search for nearest-neighbors, which are time-consuming steps common in earlier methods [28], [29], [3], [19], [17], [27], [15], [21].”
A Comparison of Affine Region Detectors (2005, neutral): “The detectors are: (i) the ‘Harris-Aﬃne’ detector [24, 27, 34]; (ii) the ‘Hessian-Aﬃne’ detector [24, 27]; (iii) the ‘maximally stable extremal region’ detector (or MSER, for short) [21, 22]; (iv) an edge-based region detector [45, 48] (referred to as EBR); (v) an intensity extrema-based region…”
A SIFT descriptor with global context (2005, background): “There has been a lot of other work in detecting, describing , and matching feature points [1] [12] [15] [19] [20].”
Multi-view Matching for Unordered Image Sets, or "How Do I Organize My Holiday Snaps?" (2002, background): “By contrast Schmid [20] and Baumberg [2] both learn a distance metric in invariant space from training data, which has the disadvantage of tuning the metric to the domain of training data.”
Unifying Deep Local and Global Features for Image Search (2020, neutral): “Early systems [32,28,40] worked by searching for query local descriptors against a large database of local descriptors, followed by geometrically verifying database images with sufficient number of correspondences.”
Class-Based Feature Matching Across Unrestricted Transformations (2008, background): “First, in several popular schemes of object classification [6], [7], indexing, and retrieval [3], objects are represented by their constituent parts.”
From BoW to CNN: Two Decades of Texture Representation for Texture Classification (2018, methodology): “In [110] two types of complementary region detectors, the Harris afﬁne detector of Mikolajczyk and Schmid [153] and the Laplacian blob detector of G ˚ arding and Lindeberg [64], were used to detect afﬁne covariant regions, meaning that the region content is afﬁne invariant.”
Matching with PROSAC - progressive sample consensus (2005, methodology): “Tentative correspondences were obtained matching MSER [6] and affineinvariant [7] regions using SIFT descriptors.”
Finding People in Images and Videos (2006, methodology): “Some point detectors such as DoGs or Harris-Laplace also provide additional local scale and/or dominant orientation information.”
Efficient representation of local geometry for large scale object retrieval (2009, neutral): “Figure 1).”
Creating efficient codebooks for visual recognition (2005, positive): “[5, 7, 15, 19]) have proposed ‘bag of features’ methods based on normalized patches or SIFT descriptors [19] over Difference of Gaussian, Harris-scale or Harris-afﬁne keypoints [22], vector quantized us-ing k-means.”
A sparse texture representation using local affine regions (2005, methodology): “In this work, we use two types of detectors: the Harris-a(cid:14)ne detector of Mikolajczyk and Schmid [33] and the Laplacian blob detector of G(cid:23)arding and Lindeberg [9].”
A multi-view probabilistic model for 3D object classes (2009, methodology): “Inspired by the implicit shape model for object recognition [16], we accomplish the task in the following steps: • Extract features and propose candidate parts We use Hessian-Affine feature detector [21] , Maximally stable extremal regions (MSER) [19] detector, and canny edge detector [23] to detect”
3D Object Modeling and Recognition Using Local Affine-Invariant Image Descriptors and Multi-View Spatial Constraints (2006, positive): “Combining this idea with the afﬁne regions of Mikolajczyk and Schmid (2002) has allowed us to construct a normalized representation of local surface appearance that can be used to 7 The shoe, like the apple, is now long gone, preventing us from adding any more training images. select promising…”
SURF: Speeded Up Robust Features (2006, neutral): “[1–6]).”