Matching shapes

By Serge J. Belongie, Jitendra Malik, Jan Puzicha. Proceedings Eighth IEEE International Conference on Computer Vision. ICCV 2001, 2001

461 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/500db68171e4a961d7fa87b8020b3a3e62133caf

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

Shape Matching and Object Recognition Using Shape Contexts (2002, background): “Parts of this work have appeared in [3, 2] We wish”
Gaussian Processes for Object Categorization (2010, methodology): “To describe each region or patch, we can choose from an array of descriptors designed to capture local texture while maintaining some invariance to small shifts and rotations, such as SIFT (Lowe 2004), shape context (Belongie et al. 2001), or geometric blur (Berg and Malik 2001).”
Estimating Human Body Configurations Using Shape Context Matching (2002, methodology): “The shape context was introduced in [3] to play such a role in shape matching.”
Scale-invariant contour completion using conditional random fields (2005, methodology): “Potential completions are generated by building a constrained Delaunay triangulation (CDT) over the set of contours found by a local edge detector.”
Fast Exact Hyper-graph Matching with Dynamic Programming for Spatio-temporal Data (2014, unknown): “Popular examples of appearance information are SIFT [35], histograms of gradient and optical flow [25], shape contexts [3] etc.”
Shape Matching and Object Recognition (2006, background): “Work based on shape contexts is indeed aimed at first finding correspondences [Belongie et al., 2001][Mori et al.”
Contextual Part Analogies in 3D Objects (2010, methodology): “Nevertheless, partial matching techniques (Belongie et al. 2001; Funkhouser and Shilane 2006; Gal and Cohen-Or 2006; Johnson and Hebert 1999; Novotni et al. 2005; Zhang et al. 2008) are based more on feature correspondence but less on segmentation results.”
Analyzing appearance and contour based methods for object categorization (2003, methodology): “We pick out a method based on the Shape Context proposed by Belongie [25], which has achieved excellent results, for example for handwritten digit recognition.”
Histograms of oriented gradients for human detection (2005, unknown): “Our circular block (C-HOG) descriptors are reminiscent of Shape Contexts [1] except that, crucially, each spatial cell contains a stack of gradient-weighted orientation cells instead of a single orientation-independent edge-presence count.”
Finding People in Images and Videos (2006, methodology): “They are reminiscent of edge orientation histograms [Freeman and Roth 1995, Freeman et al. 1996], SIFT descriptors [Lowe 2004] and shape contexts [Belongie et al. 2001], but they are computed on a dense grid of uniformly spaced cells and they use overlapping descriptors for improved performance.”
Human detection using partial least squares analysis (2009, methodology): “Their work uses HOG as low-level features, which were shown to outperform features such as wavelets [16], PCA-SIFT [11] and shape contexts [2].”
Fingerprint Matching Using an Orientation-Based Minutia Descriptor (2003, unknown): “[16].”
Review of shape representation and description techniques (2004, background): “3 [10], (a) is a character shape, (b) is the edge image of (a), (c) is the set of vectors at point p, (d) is the histogram or context extracted from the vectors, (e) is the context of the shape.”
Bundle Methods for Regularized Risk Minimization (2010, methodology): “…in Caetano et al. (2008) that usedBMRM with L2 regularization on the CMU house data set.20 For this data set, there are 30 hand-labeled landmark points in each image and the features for those points are the 60-dimensional Shape Context fea ures (Belongie et al., 2001; Caetano et al., 2008).”
Shape matching and object recognition using low distortion correspondences (2005, methodology): “Work based on shape contexts is indeed aimed at first finding correspondences [3, 22] and is close to the spirit of this work.”