Object detection using a max-margin Hough transform

By Subhransu Maji, Jitendra Malik. 2009

331 citations recorded by the source corpus.

Subfields: Aerospace Engineering, 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 Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

ImageNet Large Scale Visual Recognition Challenge (2014, methodology): “The winning entry from NEC team (Lin et al., 2011) used SIFT (Lowe, 2004) and LBP (Ahonen et al., 2006) features with two non-linear coding representations (Zhou et al., 2010; Wang et al., 2010) and a stochastic SVM.”
Unsupervised object discovery and localization in the wild: Part-based matching with bottom-up region proposals (2015, background): “Like all its “probabilistic Hough transform” predecessors we know of [3, 31, 33], however, it lacks a proper probabilistic interpretation.”
Segmentation as selective search for object recognition (2011, methodology): “Maji and Malik [20] combine multiple of these relations to predict the object location using a Hough-transform, after which they randomly sample windows close to the Hough maximum.”
Learning to Rank 3D Features (2014, background): “Moji and Malik [21] presented a 2D part based object detector where the contribution of each local part to a given object location is learned by optimizing a classification error cost.”
Evolutionary Hough Games for coherent object detection (2012, unknown): “Due to its generality, the Hough concept has recently gained much attention in the field of part-based object detection as e.g. in [3] [4] [5].”
Randomized Max-Margin Compositions for Visual Recognition (2014, background): “Moreover, the individual contributions of compositions and their parts are evaluated in separate experiments that demonstrate their potential.”
Scalable multi-class object detection (2011, methodology): “[19, 31] build the codebook in a generative way but learn appropriate weights for them discriminatively using a max-margin framework.”
Deep Hough Voting for 3D Object Detection in Point Clouds (2019, methodology): “In [30] the votes were assigned with weights indicating their importance, which were learned using a max-margin framework.”
Category Independent Object Proposals (2010, background): “, [7, 8]), which are found by searching naively over position and scale or by voting from learned codewords [9, 10], distinctive keypoints [11, 12], or regions [13].”
Selective Search for Object Recognition (2013, unknown): “Maji and Malik [23] combine multiple of these relations to predict the object location using a Hough-transform, after which they randomly sample windows close to the Hough maximum.”
Depth-Encoded Hough Voting for Joint Object Detection and Shape Recovery (2010, methodology): “For example, parts may be either selected by an interest point detector [13,16], or densely sampled across many scales and locations [10]; and the quality of the part can be learned by estimating the probability [13] that the part is good or discriminatively trained using different types of classifi”
Efficient Classification for Additive Kernel SVMs (2013, background): “Methods like branch and bound [31], context [26], bottom-up regions [56], Hough transformation [37], [33], etc.”
Hough Transform and 3D SURF for Robust Three Dimensional Classification (2010, background): “[23] work, we normalize votes on the basis of how often they vote for the correct training shape centers (during training).”
Hough Regions for Joining Instance Localization and Segmentation (2012, methodology): “We assume that we are given any generalized Hough voting method like [17, 5, 18, 19], which provides our N voting elements Xi within the test image and”
Discriminative figure-centric models for joint action localization and recognition (2011, background): “Maji and Malik [16] develop a maxmargin formulation of the Hough Transform.”
A Hough transform-based voting framework for action recognition (2010, methodology): “Inspired by the success of Hough transform-based methods in object detection [2, 15, 16, 19, 26, 11, 25], we propose a new action detection system based on Hough transform voting.”
Poselets: Body part detectors trained using 3D human pose annotations (2009, methodology): “We use the Max Margin Hough Transform (M 2 HT) proposed by Maji and Malik [ 10 ] to learn the weights.”
Latent-Class Hough Forests for 3D Object Detection and Pose Estimation (2014, background): “sentations model the relationship between these local features, either explicitly [15, 21, 22, 24, 45] by their spatial relationships, or implicitly [26, 43, 50, 55, 66] by a spatial relationship to an anchor point such as the object centre (not necessarily a feature).”
Hough Forests for Object Detection, Tracking, and Action Recognition (2011, methodology): “Over the years, many adaptations of the implicit shape model have been proposed, focusing on improving voting and hypotheses generation [21], [23], [25], [26], [28].”