Dynamical binary latent variable models for 3D human pose tracking

By Graham W. Taylor, Leonid Sigal, David J. Fleet, Geoffrey E. Hinton. 2010

161 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, 3D Vision & Pose

How researchers see this paper

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

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In 3D Vision & Pose

Citations in the corpus

Embed to Control: A Locally Linear Latent Dynamics Model for Control from Raw Images (2015, background): “Second, learning relations between pairs of images – although without control – has received considerable attention from the community during the last years [31, 32].”
Structural-RNN: Deep Learning on Spatio-Temporal Graphs (2015, methodology): “Several approaches based on Gaussian processes [59, 63], Restricted Boltzmann Machines (RBMs) [57, 56, 51], and RNNs [14] have been proposed to model human motion.”
Learning Where to Attend with Deep Architectures for Image Tracking (2011, background): “Recently, a dynamic RBM state-space model was proposed in Taylor et al. (2010).”
Learning Human Pose Estimation Features with Convolutional Networks (2013, background): “Perhaps most relevant to our work is Taylor et al.’s work on tracking people in video [45], augmenting a particle ﬁlter with a structured prior over human pose and dynamics based on learning representations.”
Hierarchical Convolutional Deep Learning in Computer Vision (2013, methodology): “The TRBM has been used to generate and denoise sequences [123, 128], as well as a prior in multi-view person tracking [127].”
A Non-parametric Bayesian Network Prior of Human Pose (2013, unknown): “An interesting recent model based on a large number of latent binary variables is the implicit mixture of conditional restricted Boltzmann machines (imCRBM) [23]; both estimation and inference are again approximate.”
Kimera: From SLAM to spatial perception with 3D dynamic scene graphs (2021, methodology): “Taylor et al. (2010) combines a learned motion model with particle filtering to predict 3D human poses.”
Recurrent Network Models for Human Dynamics (2015, methodology): “Dynamical models based on Restricted Boltzmann Machines (RBMs) have been proposed for synthesis and infilling of motion data [34, 29, 33, 35].”
Coupled Action Recognition and Pose Estimation from Multiple Views (2012, background): “Therefore, many state-of-the-art pose estimation systems use action-specific priors to simplify the pose estimation problem, e.g. (Geiger et al. 2009; Li et al. 2010; Taylor et al. 2010; Lee and Elgammal 2010; Chen et al. 2009).”
Prior Knowledge Improves Decoding of Finger Flexion from Electrocorticographic Signals (2011, methodology): “This issue has been addressed in other domains using different algorithms (Viola and Jones, 2004; Taylor et al., 2006, 2010; Tong et al., 2007; Tong and Ji, 2008; Yao et al., 2009) that can incorporate such information.”
Human Instance Segmentation from Video using Detector-based Conditional Random Fields (2011, background): “The task of human instance segmentation has been studied extensively at the level of human tracking [21], pose estimation [7] and parts detection [12].”
Posebits for Monocular Human Pose Estimation (2014, background): “Our work is partially related to [2, 7, 23] in that we also define an intermediate pose representation.”
Bootstrapping bilinear models of Simple Vehicles (2015, background): “D(Y;X;U; ∆) is the set of all discrete-time dynamical systems with sampling interval ∆.”
3D Human pose estimation: A review of the literature and analysis of covariates (2016, background): “Latent variable models: Latent variables are often used in the literature [123, 124], because it is often difficult to obtain accurate estimates of part labels because of possible occlusions.”
3D Human Pose Estimation from Monocular Images with Deep Convolutional Neural Network (2014, methodology): “Finally, [24] uses an implicit mixture of conditional restricted Boltzmann machines to model the motion of 3D human poses (i.”