ML Anthology

IGE-Net: Inverse Graphics Energy Networks for Human Pose Estimation and Single-View Reconstruction

Dominic Jack, Frederic Maire, Sareh Shirazi, Anders Eriksson
CVPR 2019
doi:10.1109/CVPR.2019.00724 /cvpr/2019/jack2019cvpr-igenet/

Abstract

Inferring 3D scene information from 2D observations is an open problem in computer vision. We propose using a deep-learning based energy minimization framework to learn a consistency measure between 2D observations and a proposed world model, and demonstrate that this framework can be trained end-to-end to produce consistent and realistic inferences. We evaluate the framework on human pose estimation and voxel-based object reconstruction benchmarks and show competitive results can be achieved with relatively shallow networks with drastically fewer learned parameters and floating point operations than conventional deep-learning approaches.

PDF CVPR Semantic Scholar

Cite

Text

Jack et al. "IGE-Net: Inverse Graphics Energy Networks for Human Pose Estimation and Single-View Reconstruction." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019. doi:10.1109/CVPR.2019.00724

Markdown

[Jack et al. "IGE-Net: Inverse Graphics Energy Networks for Human Pose Estimation and Single-View Reconstruction." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019.](https://mlanthology.org/cvpr/2019/jack2019cvpr-igenet/) doi:10.1109/CVPR.2019.00724

BibTeX

@inproceedings{jack2019cvpr-igenet,
  title     = {{IGE-Net: Inverse Graphics Energy Networks for Human Pose Estimation and Single-View Reconstruction}},
  author    = {Jack, Dominic and Maire, Frederic and Shirazi, Sareh and Eriksson, Anders},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year      = {2019},
  doi       = {10.1109/CVPR.2019.00724},
  url       = {https://mlanthology.org/cvpr/2019/jack2019cvpr-igenet/}
}