# Weakly Supervised Generative Network for Multiple 3D Human Pose Hypotheses

arXiv: https://arxiv.org/abs/2008.05770 · HTML: https://mlsanity.com/paper/2008.05770/

## Key facts

- Cited by 25 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 Aug 2020.

## Built on by (most cited first, 25 of 25)

- [MHFormer: Multi-Hypothesis Transformer for 3D Human Pose Estimation](https://mlsanity.com/paper/2111.12707/) (2021)
- [Probabilistic Monocular 3D Human Pose Estimation with Normalizing Flows](https://mlsanity.com/paper/2107.13788/) (2021)
- [Diffusion-Based 3D Human Pose Estimation with Multi-Hypothesis Aggregation](https://mlsanity.com/paper/2303.11579/) (2023)
- [DiffPose: Multi-hypothesis Human Pose Estimation using Diffusion models](https://mlsanity.com/paper/2211.16487/) (2022)
- [GFPose: Learning 3D Human Pose Prior with Gradient Fields](https://mlsanity.com/paper/2212.08641/) (2022)
- [DiffuPose: Monocular 3D Human Pose Estimation via Denoising Diffusion Probabilistic Model](https://mlsanity.com/paper/2212.02796/) (2022)
- [FinePOSE: Fine-Grained Prompt-Driven 3D Human Pose Estimation via Diffusion Models](https://mlsanity.com/paper/2405.05216/) (2024)
- [Regular Splitting Graph Network for 3D Human Pose Estimation](https://mlsanity.com/paper/2305.05785/) (2023)
- [Higher-Order Implicit Fairing Networks for 3D Human Pose Estimation](https://mlsanity.com/paper/2111.00950/) (2021)
- [MHR-Net: Multiple-Hypothesis Reconstruction of Non-Rigid Shapes from 2D Views](https://mlsanity.com/paper/2207.09086/) (2022)
- [Multi-hypothesis 3D human pose estimation metrics favor miscalibrated distributions](https://mlsanity.com/paper/2210.11179/) (2022)
- [ManiPose: Manifold-Constrained Multi-Hypothesis 3D Human Pose Estimation](https://mlsanity.com/paper/2312.06386/) (2023)
- [CHAMP: Conformalized 3D Human Multi-Hypothesis Pose Estimators](https://mlsanity.com/paper/2407.06141/) (2024)
- [Learning Temporal 3D Human Pose Estimation with Pseudo-Labels](https://mlsanity.com/paper/2110.07578/) (2021)
- [FMPose3D: monocular 3D pose estimation via flow matching](https://mlsanity.com/paper/2602.05755/) (2026)
- [Flow Matching for Probabilistic Monocular 3D Human Pose Estimation](https://mlsanity.com/paper/2601.16763/) (2026)
- [3D Human Pose Estimation Based on 2D-3D Consistency with Synchronized Adversarial Training](https://mlsanity.com/paper/2106.04274/) (2021)
- [Diversity-Promoting Human Motion Interpolation via Conditional Variational Auto-Encoder](https://mlsanity.com/paper/2111.06762/) (2021)
- [Iterative Graph Filtering Network for 3D Human Pose Estimation](https://mlsanity.com/paper/2307.16074/) (2023)
- [Diffusion-based Pose Refinement and Muti-hypothesis Generation for 3D Human Pose Estimaiton](https://mlsanity.com/paper/2401.04921/) (2024)
- [Multi-hop graph transformer network for 3D human pose estimation](https://mlsanity.com/paper/2405.03055/) (2024)
- [Probablistic Restoration with Adaptive Noise Sampling for 3D Human Pose Estimation](https://mlsanity.com/paper/2405.02114/) (2024)
- [Adaptive graph Kolmogorov-Arnold network for 3D human pose estimation](https://mlsanity.com/paper/2511.08809/) (2025)
- [Flexible Geometric Guidance for Probabilistic Human Pose Estimation with Diffusion Models](https://mlsanity.com/paper/2602.03126/) (2026)
- [MHE-Former: Multi-Hypothesis Transformers via Entropy Maximization for 3D Mesh Recovery](https://mlsanity.com/paper/2609.10743/) (2026)

## Abstract (arXiv)

3D human pose estimation from a single image is an inverse problem due to the inherent ambiguity of the missing depth. Several previous works addressed the inverse problem by generating multiple hypotheses. However, these works are strongly supervised and require ground truth 2D-to-3D correspondences which can be difficult to obtain. In this paper, we propose a weakly supervised deep generative network to address the inverse problem and circumvent the need for ground truth 2D-to-3D correspondences. To this end, we design our network to model a proposal distribution which we use to approximate the unknown multi-modal target posterior distribution. We achieve the approximation by minimizing the KL divergence between the proposal and target distributions, and this leads to a 2D reprojection error and a prior loss term that can be weakly supervised. Furthermore, we determine the most probable solution as the conditional mode of the samples using the mean-shift algorithm. We evaluate our method on three benchmark datasets -- Human3.6M, MPII and MPI-INF-3DHP. Experimental results show that our approach is capable of generating multiple feasible hypotheses and achieves state-of-the-art results compared to existing weakly supervised approaches. Our source code is available at the project website.