Header logo is

Learning to Reconstruct 3D Human Pose and Shape via Model-fitting in the Loop

2019

Conference Paper

ps


Model-based human pose estimation is currently approached through two different paradigms. Optimization-based methods fit a parametric body model to 2D observations in an iterative manner, leading to accurate image-model alignments, but are often slow and sensitive to the initialization. In contrast, regression-based methods, that use a deep network to directly estimate the model parameters from pixels, tend to provide reasonable, but not pixel accurate, results while requiring huge amounts of supervision. In this work, instead of investigating which approach is better, our key insight is that the two paradigms can form a strong collaboration. A reasonable, directly regressed estimate from the network can initialize the iterative optimization making the fitting faster and more accurate. Similarly, a pixel accurate fit from iterative optimization can act as strong supervision for the network. This is the core of our proposed approach SPIN (SMPL oPtimization IN the loop). The deep network initializes an iterative optimization routine that fits the body model to 2D joints within the training loop, and the fitted estimate is subsequently used to supervise the network. Our approach is self-improving by nature, since better network estimates can lead the optimization to better solutions, while more accurate optimization fits provide better supervision for the network. We demonstrate the effectiveness of our approach in different settings, where 3D ground truth is scarce, or not available, and we consistently outperform the state-of-the-art model-based pose estimation approaches by significant margins.

Author(s): Kolotouros, Nikos and Pavlakos, Georgios and Black, Michael J. and Daniilidis, Kostas
Book Title: Proceedings International Conference on Computer Vision (ICCV)
Pages: 2252--2261
Year: 2019
Month: October
Day: 27
Publisher: IEEE

Department(s): Perceiving Systems
Research Project(s): Regressing Humans
Bibtex Type: Conference Paper (conference)
Paper Type: Conference

DOI: 10.1109/ICCV.2019.00234
Event Name: 2019 IEEE/CVF International Conference on Computer Vision (ICCV)
Event Place: Seoul, South Korea

ISBN: 978-1-7281-4803-8
Note: ISSN: 2380-7504
State: Published

Links: pdf
code
project

BibTex

@conference{SPIN:ICCV:2019,
  title = {Learning to Reconstruct {3D} Human Pose and Shape via Model-fitting in the Loop},
  author = {Kolotouros, Nikos and Pavlakos, Georgios and Black, Michael J. and Daniilidis, Kostas},
  booktitle = {Proceedings International Conference on Computer Vision (ICCV)},
  pages = {2252--2261},
  publisher = {IEEE},
  month = oct,
  year = {2019},
  note = {ISSN: 2380-7504},
  doi = {10.1109/ICCV.2019.00234},
  month_numeric = {10}
}