mlsanityData to 7 Oct 2026

HuManiFlow: Ancestor-Conditioned Normalising Flows on SO(3) Manifolds for Human Pose and Shape Distribution Estimation

arXiv 2305.06968 ↗First posted 11 May 2023cs.CVcited by 23

Not on the map. Why?

Abstract (arXiv)

Monocular 3D human pose and shape estimation is an ill-posed problem since multiple 3D solutions can explain a 2D image of a subject. Recent approaches predict a probability distribution over plausible 3D pose and shape parameters conditioned on the image. We show that these approaches exhibit a trade-off between three key properties: (i) accuracy - the likelihood of the ground-truth 3D solution under the predicted distribution, (ii) sample-input consistency - the extent to which 3D samples from the predicted distribution match the visible 2D image evidence, and (iii) sample diversity - the range of plausible 3D solutions modelled by the predicted distribution. Our method, HuManiFlow, predicts simultaneously accurate, consistent and diverse distributions. We use the human kinematic tree to factorise full body pose into ancestor-conditioned per-body-part pose distributions in an autoregressive manner. Per-body-part distributions are implemented using normalising flows that respect the manifold structure of SO(3), the Lie group of per-body-part poses. We show that ill-posed, but ubiquitous, 3D point estimate losses reduce sample diversity, and employ only probabilistic training losses. Code is available at: https://github.com/akashsengupta1997/HuManiFlow.

Key facts. Cited by 23 papers in the map (data through 7 Oct 2026). Builds on 46 of its 68 references that are in the map. First posted on arXiv 11 May 2023.

Builds on

46 of 68 references.

References · 46

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  3. 2014Microsoft COCO: Common Objects in Contextcited by 17,498
  4. 2013Auto-Encoding Variational Bayescited by 3,980
  5. 2015Variational Inference with Normalizing Flowscited by 2,161
  6. 2016Density estimation using Real NVPcited by 2,029
  7. 2019Expressive Body Capture: 3D Hands, Face, and Body from a Single Imagecited by 1,600
  8. 2015LSUN: Construction of a Large-scale Image Dataset using Deep Learning with Humans in the Loopcited by 1,350
  9. 2019Deep High-Resolution Representation Learning for Human Pose Estimationcited by 1,339
  10. 2019Normalizing Flows for Probabilistic Modeling and Inferencecited by 1,315
  11. 2018On the Continuity of Rotation Representations in Neural Networkscited by 1,186
  12. 2015Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)cited by 1,061
34 more
  1. 2014NICE: Non-linear Independent Components Estimationcited by 1,052
  2. 2017End-to-end Recovery of Human Shape and Posecited by 883
  3. 2019PIFu: Pixel-Aligned Implicit Function for High-Resolution Clothed Human Digitizationcited by 818
  4. 2020Accelerating 3D Deep Learning with PyTorch3Dcited by 781
  5. 2016Keep it SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Imagecited by 692
  6. 2019Learning to Reconstruct 3D Human Pose and Shape via Model-fitting in the Loopcited by 625
  7. 2018DensePose: Dense Human Pose Estimation In The Wildcited by 573
  8. 2020PIFuHD: Multi-Level Pixel-Aligned Implicit Function for High-Resolution 3D Human Digitizationcited by 521
  9. 2017Learning from Synthetic Humanscited by 358
  10. 2021PARE: Part Attention Regressor for 3D Human Body Estimationcited by 334
  11. 2020HybrIK: A Hybrid Analytical-Neural Inverse Kinematics Solution for 3D Human Pose and Shape Estimationcited by 304
  12. 2020I2L-MeshNet: Image-to-Lixel Prediction Network for Accurate 3D Human Pose and Mesh Estimation from a Single RGB Imagecited by 274
  13. 2019Convolutional Mesh Regression for Single-Image Human Shape Reconstructioncited by 270
  14. 2020Pose2Mesh: Graph Convolutional Network for 3D Human Pose and Mesh Recovery from a 2D Human Posecited by 248
  15. 2019Multi-Garment Net: Learning to Dress 3D People from Imagescited by 244
  16. 2017Unite the People: Closing the Loop Between 3D and 2D Human Representationscited by 205
  17. 2019Learning Likelihoods with Conditional Normalizing Flowscited by 170
  18. 2021Probabilistic Modeling for Human Mesh Recoverycited by 152
  19. 2018BodyNet: Volumetric Inference of 3D Human Body Shapescited by 141
  20. 2020Exemplar Fine-Tuning for 3D Human Model Fitting Towards In-the-Wild 3D Human Pose Estimationcited by 113
  21. 2016Learning in an Uncertain World: Representing Ambiguity Through Multiple Hypothesescited by 96
  22. 2019Overcoming Limitations of Mixture Density Networks: A Sampling and Fitting Framework for Multimodal Future Predictioncited by 96
  23. 2019Generating Multiple Hypotheses for 3D Human Pose Estimation with Mixture Density Networkcited by 92
  24. 2021Probabilistic Monocular 3D Human Pose Estimation with Normalizing Flowscited by 84
  25. 2020Synthetic Training for Accurate 3D Human Pose and Shape Estimation in the Wildcited by 82
  26. 2020Hierarchical Kinematic Human Mesh Recoverycited by 71
  27. 2020Neural Descent for Visual 3D Human Pose and Shapecited by 60
  28. 2022FLAG: Flow-based 3D Avatar Generation from Sparse Observationscited by 54
  29. 2021Hierarchical Kinematic Probability Distributions for 3D Human Shape and Pose Estimation from Images in the Wildcited by 48
  30. 2021Probabilistic 3D Human Shape and Pose Estimation from Multiple Unconstrained Images in the Wildcited by 40
  31. 2020Invertible Generative Modeling using Linear Rational Splinescited by 39
  32. 2020Probabilistic orientation estimation with matrix Fisher distributionscited by 36
  33. 2020GraphMDN: Leveraging graph structure and deep learning to solve inverse problemscited by 26
  34. 2021Probabilistic Estimation of 3D Human Shape and Pose with a Semantic Local Parametric Modelcited by 5

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 4; 2024: 10; 2025: 5; 2026: 4.

23 papers.

Sort:

By area:

  1. 2024TokenHMR: Advancing Human Mesh Recovery with a Tokenized Pose Representationcited by 93
  2. 2024Neural Localizer Fields for Continuous 3D Human Pose and Shape Estimationcited by 57
  3. 2024PhysPT: Physics-aware Pretrained Transformer for Estimating Human Dynamics from Monocular Videoscited by 27
  4. 2023POCO: 3D Pose and Shape Estimation with Confidencecited by 20
  5. 2023VQ-HPS: Human Pose and Shape Estimation in a Vector-Quantized Latent Spacecited by 13
  6. 2024DiffHuman: Probabilistic Photorealistic 3D Reconstruction of Humanscited by 13
  7. 2024MEGA: Masked Generative Autoencoder for Human Mesh Recoverycited by 13
  8. 2024Robust 3D Shape Reconstruction in Zero-Shot from a Single Image in the Wildcited by 7
  9. 2024Normalizing Flows on the Product Space of SO(3) Manifolds for Probabilistic Human Pose Modelingcited by 7
  10. 2025ADHMR: Aligning Diffusion-based Human Mesh Recovery via Direct Preference Optimizationcited by 6
  11. 2024CondiMen: Conditional Multi-Person Mesh Recoverycited by 5
  12. 2024Utilizing Uncertainty in 2D Pose Detectors for Probabilistic 3D Human Mesh Recoverycited by 4
11 more
  1. 2025FastHMR: Accelerating Human Mesh Recovery via Token and Layer Merging with Diffusion Decodingcited by 4
  2. 2025Kineo: Calibration-Free Metric Motion Capture From Sparse RGB Camerascited by 4
  3. 2023Spatial-Related Sensors Matters: 3D Human Motion Reconstruction Assisted with Textual Semanticscited by 2
  4. 2025PHD: Personalized 3D Human Body Fitting with Point Diffusioncited by 2
  5. 2026VLM-Guided Group Preference Alignment for Diffusion-based Human Mesh Recoverycited by 2
  6. 2023SHARE: Single-view Human Adversarial REconstructioncited by 0
  7. 2024ProGraph: Temporally-alignable Probability Guided Graph Topological Modeling for 3D Human Reconstructioncited by 0
  8. 2025LieHMR: Autoregressive Human Mesh Recovery with SO(3) Diffusioncited by 0
  9. 2026Monocular Models are Strong Learners for Multi-View Human Mesh Recoverycited by 0
  10. 2026CQF-HMR: Continuous Quaternion Flows for Probabilistic 3D Human Mesh Recovery from a Single Imagecited by 0
  11. 2026MHE-Former: Multi-Hypothesis Transformers via Entropy Maximization for 3D Mesh Recoverycited by 0