Human Pose Estimation from Monocular Video


This paper addresses the challenge of 3D full-body human pose estimation from a monocular image sequence. Here, two cases are considered: (i) the image locations of the human joints are provided and (ii) the image locations of joints are unknown. In the former case, a novel approach is introduced that integrates a sparsity-driven 3D geometric prior and temporal smoothness. In the latter case, the former case is extended by treating the image locations of the joints as latent variables. A deep fully convolutional network is trained to predict the uncertainty maps of the 2D joint locations.  The 3D pose estimates are realized via an Expectation-Maximization algorithm over the entire sequence, where it is shown that the 2D joint location uncertainties can be conveniently marginalized out during inference. Empirical evaluation on the Human3.6M dataset shows that the proposed approaches achieve greater 3D pose estimation accuracy over  state-of-the-art baselines.  Further, the proposed approach outperforms a publicly available 2D pose estimation baseline on the challenging PennAction dataset.



Example results



Sparseness Meets Deepness: 3D Human Pose Estimation from Monocular Video.
X. Zhou, M. Zhu, S. Leonardos, K. Derpanis, K. Daniilidis.
CVPR 2016.

Supplementary material | Poster


Updated package that includes the whole pipeline for reconstructing 3D human poses from an image sequence
including the proposed reconstruction algorithm + the “Stacked Hourglass Network” for 2D pose detection.



Posted in My Projects