A Non-parametric Bayesian Network Prior of Human Pose

By Andreas M. Lehrmann, Peter Gehler, Sebastian Nowozin. 2013

50 citations recorded by the source corpus.

Subfields: Artificial Intelligence, 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 Artificial Intelligence

In Computer Vision and Pattern Recognition

Citations in the corpus

Space-Time Representation of People Based on 3D Skeletal Data: A Review (2016, background): “[199] Vector of Joints Conc Lowlv Hand X X X”
Monocular 3D Human Pose Estimation in the Wild Using Improved CNN Supervision (2016, methodology): “For the joint set we consider, the kinematic relationships we propose are sufficient, as it puts at least one reference joint for each joint in the relatively low entropy torso [34].”
On Human Motion Prediction Using Recurrent Neural Networks (2017, background): “Another example of baselines outperforming recent work in the ﬁeld of pose models can be found in [25], where a Gaussian pose prior outperforms the more complicated GPLVM.”
Bayesian Image Based 3D Pose Estimation (2016, methodology): “As in [23], our approach is able to generalize over the observed data so as to generate new poses never seen before.”
Leveraging Hierarchical Parametric Networks for Skeletal Joints Based Action Segmentation and Recognition (2014, background): “Though we believe further carefully fine-tuned parameters would lead to more competitive results, in order not to “creeping overfitting”, as XXXXXXXXXXMethod Data Set ChaLearn Gesture MSR Action3D EigenJoints+NBNN [30] 0.593 0.720 GMM+HMM [21] 0.408 0.704 NN+DTW [29] 0.599 - DBN+HMM (this work)…”
Synthesizing Training Images for Boosting Human 3D Pose Estimation (2016, methodology): “We use continuous dynamic time warping [35] for computing the dense correspondences M(p) between the contours of the two human body parts P = {pi} and those of the imaged clothing articles.”
Monocular 3D Human Pose Estimation Using Transfer Learning and Improved CNN Supervision (2016, methodology): “For the joint set we consider, this suffices as it puts at least one reference joint for each joint in the relatively low entropy torso [38].”
Body Parts Dependent Joint Regressors for Human Pose Estimation in Still Images (2014, background): “The particular choice of regressors enables us to train our model also on large datasets in reasonable time, which is an important feature for real-world applications and a limitation of many other approaches.”