mkocabas/EpipolarPose
A PyTorch implementation of a CVPR 2019 paper that estimates 3D human pose from multi-view images using self-supervised learning via epipolar geometry.
Not currently ranked — collecting fresh signals.
star history
EpipolarPose is a self-supervised method for 3D human pose estimation that does not require 3D ground-truth data or camera extrinsics. During training, it estimates 2D poses from multi-view images and uses epipolar geometry to derive 3D pose and camera geometry, which then trains a 3D pose estimator. At test time, it takes a single RGB image as input to produce 3D pose output.
Frequently asked
- What is mkocabas/EpipolarPose?
- A PyTorch implementation of a CVPR 2019 paper that estimates 3D human pose from multi-view images using self-supervised learning via epipolar geometry.
- Is EpipolarPose open source?
- Yes — mkocabas/EpipolarPose is an open-source project tracked on heatdrop.
- What language is EpipolarPose written in?
- mkocabas/EpipolarPose is primarily written in Jupyter Notebook.
- How popular is EpipolarPose?
- mkocabas/EpipolarPose has 608 stars on GitHub.
- Where can I find EpipolarPose?
- mkocabas/EpipolarPose is on GitHub at https://github.com/mkocabas/EpipolarPose.