3D human pose from silhouettes by relevance vector regression

By Ankur Agarwal, Bill Triggs. Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2004. CVPR 2004., 2004

495 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/3845d9e62540b8e2406343f801e026b562299ae0

How researchers see this paper

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

Citations in the corpus

Advances in View-Invariant Human Motion Analysis: A Review (2010, methodology): “A method was proposed by Agarwal and Triggs [69], [70] that recovered a 3-”
Exemplar-based Action Recognition in Video (2009, background): “[1, 7, 20]) or approaches based on pose estimation (e.”
Structured Prediction of 3D Human Pose with Deep Neural Networks (2016, background): “The benefits are twofold: We can leverage the power of learned features that have proven more effective than handcrafted ones such as HOG [1], and our framework relies on a direct and efficient regression between the two spaces, thus avoiding the computational burden of the state-of-the-art approach”
Fusing 2D Uncertainty and 3D Cues for Monocular Body Pose Estimation (2016, background): “Over the years, two main classes of approaches have been proposed: Discriminative ones, that directly regress 3D pose from image data [2, 7, 31, 50, 61], and generative ones that search the pose space for a plausible skeleton configuration that aligns with the image data [20, 53, 62].”
Real time motion capture using a single time-of-flight camera (2010, background): “Several learning based approaches [1, 17, 14] attempt to directly map image structures, silhouettes or features computed from them directly to poses.”
Learning to Fuse 2D and 3D Image Cues for Monocular Body Pose Estimation (2016, background): “Over the years, two main classes of approaches have been proposed: Discriminative ones that directly regress 3D pose from image data [1, 8, 34, 46, 56, 68] and generative ones that search the pose space for a plausible body configuration that aligns with the image data [21, 60, 69].”
3D Human Pose Estimation = 2D Pose Estimation + Matching (2016, methodology): “Other methods also warp 3D exemplars to 2D image descriptors, often based on shape-context [1, 20] or silhouette features [5].”
Progressive search space reduction for human pose estimation (2008, background): “Indeed 2D human segmentation is often the ﬁrst step in determining 3D human pose from individual frames [2].”
Iterated Second-Order Label Sensitive Pooling for 3D Human Pose Estimation (2014, background): “Discriminative regression methods[1, 31, 28, 20, 17] focus on prediction from image descriptors, but depend on the existence of a sufficiently large and diverse training set, which until recently has not been available.”
A Simple Yet Effective Baseline for 3d Human Pose Estimation (2017, methodology): “Early methods achieved this invariance through features such as silhouettes [1], shape context [28], SIFT descriptors [6] or edge direction histograms [40].”
HumanEva: Synchronized Video and Motion Capture Dataset and Baseline Algorithm for Evaluation of Articulated Human Motion (2010, methodology): “A variety of statistical (Agarwal and Triggs 2004a, 2004b; Balan et al. 2005; Deutscher and Reid 2005; Hua et al. 2005; Sigal et al. 2004; Sigal and Black 2006; Sminchisescu et al. 2005) as well as deterministic methods (Mori et al. 2004; Taylor 2000; Shakhnarovich et al. 2003) have been developed…”
Cascaded pose regression (2010, background): “Although for certain taks in computer vision regression has been successful [30, 1], its applicability to more general pose estimation remains unclear.”
A review on vision techniques applied to Human Behaviour Analysis for Ambient-Assisted Living (2012, methodology): “Furthermore, techniques that are examplebased are presented too; these use either a representation of the mapping from 2D silhouette sequences in image space to skeletal motion in 3D pose space [98] or direct lookup of silhouette sequences for recognition [1, 40, 104, 108].”
HumanEva: Synchronized Video and Motion Capture Dataset for Evaluation of Articulated Human Motion (2006, methodology): “A variety of statistical [1, 2, 3, 6, 12, 36, 37, 38] as well as deterministic methods [21, 41, 35] have been developed for tracking people from single [1, 2, 7, 12, 15, 20, 21, 25, 26, 27, 29, 37] as well as multiple [3, 6, 9, 36] views.”
3D Human pose estimation: A review of the literature and analysis of covariates (2016, background): “Illustrative examples can be found in the early publications of Agarwal and Triggs [64, 154] and in the work of Ning et al. [40].”
Semiparametric latent factor models (2005, background): “In this case the response variables are the joint angles of the human body [1].”
The Pyramid Match Kernel: Efficient Learning with Sets of Features (2007, background): “Many vision researchers have addressed the difficult problem of articulated pose estimation; recent approaches have attempted to directly learn the relationship between images and 3-D pose (Agarwal and Triggs, 2004; Shakhnarovich et al., 2003; Grauman et al., 2003).”
Seeing Invisible Poses: Estimating 3D Body Pose from Egocentric Video (2016, background): “, [5, 8, 9, 6]), including the recent DeepPose work using convolutional neural networks [5].”
Human3.6M: Large Scale Datasets and Predictive Methods for 3D Human Sensing in Natural Environments (2014, methodology): “Quantitative evaluations were pursued later using graphic renderings of synthetic models [29], [30], [12].”