# Unsupervised Multi-Person 3D Human Pose Estimation From 2D Poses Alone

arXiv: https://arxiv.org/abs/2309.14865 · HTML: https://mlsanity.com/paper/2309.14865/

## Key facts

- Cited by 1 papers in the map (data through 14 Oct 2026).
- Builds on 8 of its 19 references that are in the map.
- First posted on arXiv 26 Sep 2023.

## Builds on (most cited first, 8 of 8)

- [Expressive Body Capture: 3D Hands, Face, and Body from a Single Image](https://mlsanity.com/paper/1904.05866/) (2019)
- [Monocular 3D Human Pose Estimation In The Wild Using Improved CNN Supervision](https://mlsanity.com/paper/1611.09813/) (2016)
- [CLIFF: Carrying Location Information in Full Frames into Human Pose and Shape Estimation](https://mlsanity.com/paper/2208.00571/) (2022)
- [Coherent Reconstruction of Multiple Humans from a Single Image](https://mlsanity.com/paper/2006.08586/) (2020)
- [Unsupervised 3D Pose Estimation with Geometric Self-Supervision](https://mlsanity.com/paper/1904.04812/) (2019)
- [Can 3D Pose be Learned from 2D Projections Alone?](https://mlsanity.com/paper/1808.07182/) (2018)
- [ElePose: Unsupervised 3D Human Pose Estimation by Predicting Camera Elevation and Learning Normalizing Flows on 2D Poses](https://mlsanity.com/paper/2112.07088/) (2021)
- [LInKs "Lifting Independent Keypoints" -- Partial Pose Lifting for Occlusion Handling with Improved Accuracy in 2D-3D Human Pose Estimation](https://mlsanity.com/paper/2309.07243/) (2023)

## Built on by (most cited first, 1 of 1)

- [Reconstructing People, Places, and Cameras](https://mlsanity.com/paper/2412.17806/) (2024)

## Abstract (arXiv)

Current unsupervised 2D-3D human pose estimation (HPE) methods do not work in multi-person scenarios due to perspective ambiguity in monocular images. Therefore, we present one of the first studies investigating the feasibility of unsupervised multi-person 2D-3D HPE from just 2D poses alone, focusing on reconstructing human interactions. To address the issue of perspective ambiguity, we expand upon prior work by predicting the cameras' elevation angle relative to the subjects' pelvis. This allows us to rotate the predicted poses to be level with the ground plane, while obtaining an estimate for the vertical offset in 3D between individuals. Our method involves independently lifting each subject's 2D pose to 3D, before combining them in a shared 3D coordinate system. The poses are then rotated and offset by the predicted elevation angle before being scaled. This by itself enables us to retrieve an accurate 3D reconstruction of their poses. We present our results on the CHI3D dataset, introducing its use for unsupervised 2D-3D pose estimation with three new quantitative metrics, and establishing a benchmark for future research.