mlsanityData to 14 Oct 2026

Adversarial Parametric Pose Prior

arXiv 2112.04203 ↗First posted 8 Dec 2021cs.CVcited by 38

Not on the map. Why?

Abstract (arXiv)

The Skinned Multi-Person Linear (SMPL) model can represent a human body by mapping pose and shape parameters to body meshes. This has been shown to facilitate inferring 3D human pose and shape from images via different learning models. However, not all pose and shape parameter values yield physically-plausible or even realistic body meshes. In other words, SMPL is under-constrained and may thus lead to invalid results when used to reconstruct humans from images, either by directly optimizing its parameters, or by learning a mapping from the image to these parameters. In this paper, we therefore learn a prior that restricts the SMPL parameters to values that produce realistic poses via adversarial training. We show that our learned prior covers the diversity of the real-data distribution, facilitates optimization for 3D reconstruction from 2D keypoints, and yields better pose estimates when used for regression from images. We found that the prior based on spherical distribution gets the best results. Furthermore, in all these tasks, it outperforms the state-of-the-art VAE-based approach to constraining the SMPL parameters.

Key facts. Cited by 38 papers in the map (data through 14 Oct 2026). Builds on 23 of its 31 references that are in the map. First posted on arXiv 8 Dec 2021.

Builds on

23 of 31 references.

References · 23

  1. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  2. 2018A Style-Based Generator Architecture for Generative Adversarial Networkscited by 6,013
  3. 2019Expressive Body Capture: 3D Hands, Face, and Body from a Single Imagecited by 1,612
  4. 2019AMASS: Archive of Motion Capture as Surface Shapescited by 1,236
  5. 2017Deep Image Priorcited by 1,098
  6. 2017End-to-end Recovery of Human Shape and Posecited by 885
  7. 2016Keep it SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Imagecited by 695
  8. 2019Learning to Reconstruct 3D Human Pose and Shape via Model-fitting in the Loopcited by 627
  9. 2017Learning from Synthetic Humanscited by 359
  10. 2017Unite the People: Closing the Loop Between 3D and 2D Human Representationscited by 205
  11. 2018Neural Body Fitting: Unifying Deep Learning and Model-Based Human Pose and Shape Estimationcited by 185
  12. 2019NASA: Neural Articulated Shape Approximationcited by 149
11 more
  1. 2020Coherent Reconstruction of Multiple Humans from a Single Imagecited by 136
  2. 2020Exemplar Fine-Tuning for 3D Human Model Fitting Towards In-the-Wild 3D Human Pose Estimationcited by 113
  3. 2019Exploiting temporal context for 3D human pose estimation in the wildcited by 110
  4. 2020Contact and Human Dynamics from Monocular Videocited by 95
  5. 2020Weakly Supervised 3D Human Pose and Shape Reconstruction with Normalizing Flowscited by 94
  6. 2020Hierarchical Kinematic Human Mesh Recoverycited by 71
  7. 2020LIMP: Learning Latent Shape Representations with Metric Preservation Priorscited by 55
  8. 2021CAMPARI: Camera-Aware Decomposed Generative Neural Radiance Fieldscited by 55
  9. 2019TexturePose: Supervising Human Mesh Estimation with Texture Consistencycited by 53
  10. 2015Building Statistical Shape Spaces for 3D Human Modelingcited by 47
  11. 2021Everybody Is Unique: Towards Unbiased Human Mesh Recoverycited by 10

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2022: 8; 2023: 8; 2024: 15; 2025: 5; 2026: 2.

38 papers.

Sort:

By area:

  1. 2022CLIFF: Carrying Location Information in Full Frames into Human Pose and Shape Estimationcited by 238
  2. 2022Recovering 3D Human Mesh from Monocular Images: A Surveycited by 103
  3. 2022Pose-NDF: Modeling Human Pose Manifolds with Neural Distance Fieldscited by 85
  4. 2022DiffPose: Multi-hypothesis Human Pose Estimation using Diffusion modelscited by 65
  5. 2022GFPose: Learning 3D Human Pose Prior with Gradient Fieldscited by 57
  6. 2023Neural Sign Actors: A diffusion model for 3D sign language production from textcited by 49
  7. 2022DrapeNet: Garment Generation and Self-Supervised Drapingcited by 41
  8. 2024Score-Guided Diffusion for 3D Human Recoverycited by 39
  9. 2024NRDF: Neural Riemannian Distance Fields for Learning Articulated Pose Priorscited by 27
  10. 2022Learning 3D Scene Priors with 2D Supervisioncited by 24
  11. 2024Programmable Motion Generation for Open-Set Motion Control Taskscited by 20
  12. 2023Modiff: Action-Conditioned 3D Motion Generation with Denoising Diffusion Probabilistic Modelscited by 17
26 more
  1. 2025DPoser-X: Diffusion Model as Robust 3D Whole-body Human Pose Priorcited by 13
  2. 2024MetaCap: Meta-learning Priors from Multi-View Imagery for Sparse-view Human Performance Capture and Renderingcited by 11
  3. 2022Bringing motion taxonomies to continuous domains via GPLVM on hyperbolic manifoldscited by 10
  4. 2024SMPLer: Taming Transformers for Monocular 3D Human Shape and Pose Estimationcited by 9
  5. 2023Pose-to-Motion: Cross-Domain Motion Retargeting with Pose Priorcited by 8
  6. 2023BodySLAM++: Fast and Tightly-Coupled Visual-Inertial Camera and Human Motion Trackingcited by 7
  7. 2024Normalizing Flows on the Product Space of SO(3) Manifolds for Probabilistic Human Pose Modelingcited by 7
  8. 2023KBody: Towards general, robust, and aligned monocular whole-body estimationcited by 6
  9. 2025Geometric Neural Distance Fields for Learning Human Motion Priorscited by 6
  10. 2024Diversifying Human Pose in Synthetic Data for Aerial-view Human Detectioncited by 5
  11. 2024DiffSurf: A Transformer-based Diffusion Model for Generating and Reconstructing 3D Surfaces in Posecited by 3
  12. 2023PoseGen: Learning to Generate 3D Human Pose Dataset with NeRFcited by 2
  13. 2024Personalized 3D Human Pose and Shape Refinementcited by 2
  14. 2024Multi-modal Pose Diffuser: A Multimodal Generative Conditional Pose Priorcited by 2
  15. 2024Recovering Complete Actions for Cross-dataset Skeleton Action Recognitioncited by 2
  16. 2025MeshMamba: State Space Models for Articulated 3D Mesh Generation and Reconstructioncited by 2
  17. 2024Incorporating Test-Time Optimization into Training with Dual Networks for Human Mesh Recoverycited by 1
  18. 2024MoManifold: Learning to Measure 3D Human Motion via Decoupled Joint Acceleration Manifoldscited by 1
  19. 2026Anny-Fit: All-Age Human Mesh Recoverycited by 1
  20. 2023Noise-in, Bias-out: Balanced and Real-time MoCap Solvingcited by 0
  21. 2023QPoser: Quantized Explicit Pose Prior Modeling for Controllable Pose Generationcited by 0
  22. 2024Using Motion Cues to Supervise Single-Frame Body Pose and Shape Estimation in Low Data Regimescited by 0
  23. 2024Learning Human Motion from Monocular Videos via Cross-Modal Manifold Alignmentcited by 0
  24. 2025Neural Human Pose Priorcited by 0
  25. 2025LieHMR: Autoregressive Human Mesh Recovery with SO(3) Diffusioncited by 0
  26. 2026AvatarDynamizer: From Static to Dynamic Human Avatars via Generative Dynamic Texturescited by 0