mlsanityData to 14 Oct 2026

"Teaching Independent Parts Separately" (TIPSy-GAN) : Improving Accuracy and Stability in Unsupervised Adversarial 2D to 3D Pose Estimation

arXiv 2205.05980 ↗First posted 12 May 2022cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

We present TIPSy-GAN, a new approach to improve the accuracy and stability in unsupervised adversarial 2D to 3D human pose estimation. In our work we demonstrate that the human kinematic skeleton should not be assumed as a single spatially codependent structure; in fact, we posit when a full 2D pose is provided during training, there is an inherent bias learned where the 3D coordinate of a keypoint is spatially codependent on the 2D coordinates of all other keypoints. To investigate our hypothesis we follow previous adversarial approaches but train two generators on spatially independent parts of the kinematic skeleton, the torso and the legs. We find that improving the self-consistency cycle is key to lowering the evaluation error and therefore introduce new consistency constraints during training. A TIPSy model is produced via knowledge distillation from these generators which can predict the 3D ordinates for the entire 2D pose with improved results. Furthermore, we address an unanswered question in prior work of how long to train in a truly unsupervised scenario. We show that for two independent generators training adversarially has improved stability than that of a solo generator which collapses. TIPSy decreases the average error by 17% when compared to that of a baseline solo generator on the Human3.6M dataset. TIPSy improves upon other unsupervised approaches while also performing strongly against supervised and weakly-supervised approaches during evaluation on both the Human3.6M and MPI-INF-3DHP datasets.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 39 of its 52 references that are in the map. First posted on arXiv 12 May 2022.

Builds on

39 of 52 references.

References · 39

  1. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  2. 2015Distilling the Knowledge in a Neural Networkcited by 10,324
  3. 2016Layer Normalizationcited by 5,669
  4. 2022Generative Adversarial Networkscited by 5,627
  5. 2017Improved Training of Wasserstein GANscited by 2,235
  6. 2015Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networkscited by 2,054
  7. 2018OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fieldscited by 1,162
  8. 2016Stacked Hourglass Networks for Human Pose Estimationcited by 915
  9. 2017End-to-end Recovery of Human Shape and Posecited by 885
  10. 2016Monocular 3D Human Pose Estimation In The Wild Using Improved CNN Supervisioncited by 494
  11. 2017Towards Principled Methods for Training Generative Adversarial Networkscited by 464
  12. 20183D human pose estimation in video with temporal convolutions and semi-supervised trainingcited by 403
27 more
  1. 2017A simple yet effective baseline for 3d human pose estimationcited by 401
  2. 2018How Does Batch Normalization Help Optimization?cited by 357
  3. 2016Coarse-to-Fine Volumetric Prediction for Single-Image 3D Human Posecited by 221
  4. 2017VNect: Real-time 3D Human Pose Estimation with a Single RGB Cameracited by 129
  5. 2017Compositional Human Pose Regressioncited by 124
  6. 2017Towards 3D Human Pose Estimation in the Wild: a Weakly-supervised Approachcited by 115
  7. 20163D Human Pose Estimation = 2D Pose Estimation + Matchingcited by 113
  8. 2018Ordinal Depth Supervision for 3D Human Pose Estimationcited by 107
  9. 2019Unsupervised 3D Pose Estimation with Geometric Self-Supervisioncited by 95
  10. 2020SRNet: Improving Generalization in 3D Human Pose Estimation with a Split-and-Recombine Approachcited by 94
  11. 20183D Human Pose Estimation in the Wild by Adversarial Learningcited by 86
  12. 2019RepNet: Weakly Supervised Training of an Adversarial Reprojection Network for 3D Human Pose Estimationcited by 85
  13. 2017Lifting from the Deep: Convolutional 3D Pose Estimation from a Single Imagecited by 73
  14. 2018Can 3D Pose be Learned from 2D Projections Alone?cited by 44
  15. 2016Structured Prediction of 3D Human Pose with Deep Neural Networkscited by 43
  16. 2015Maximum-Margin Structured Learning with Deep Networks for 3D Human Pose Estimationcited by 37
  17. 2019Multi-task Deep Learning for Real-Time 3D Human Pose Estimation and Action Recognitioncited by 35
  18. 2020Feature Importance Ranking for Deep Learningcited by 31
  19. 2018OriNet: A Fully Convolutional Network for 3D Human Pose Estimationcited by 28
  20. 2020Kinematic-Structure-Preserved Representation for Unsupervised 3D Human Pose Estimationcited by 26
  21. 2020DOPE: Distillation Of Part Experts for whole-body 3D pose estimation in the wildcited by 25
  22. 2018Unsupervised Adversarial Learning of 3D Human Pose from 2D Joint Locationscited by 18
  23. 2019Distill Knowledge from NRSfM for Weakly Supervised 3D Pose Learningcited by 13
  24. 20183D Human Pose Estimation with Relational Networkscited by 10
  25. 2020Invariant Teacher and Equivariant Student for Unsupervised 3D Human Pose Estimationcited by 10
  26. 2017A Kinematic Chain Space for Monocular Motion Capturecited by 8
  27. 2018It's all Relative: Monocular 3D Human Pose Estimation from Weakly Supervised Datacited by 5

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2021When Human Pose Estimation Meets Robustness: Adversarial Algorithms and Benchmarks
  2. 2024GUNet: A Graph Convolutional Network United Diffusion Model for Stable and Diversity Pose Generation
  3. 2025ARMO: Autoregressive Rigging for Multi-Category Objects
  4. 2019Learning to Reconstruct 3D Human Pose and Shape via Model-fitting in the Loop
  5. 2025PoseBH: Prototypical Multi-Dataset Training Beyond Human Pose Estimation
  6. 2026VARPose: Flexible 2D Pose Densification via Visual Autoregressive Modeling for Enhanced 3D Lifting
  7. 2025HiPART: Hierarchical Pose AutoRegressive Transformer for Occluded 3D Human Pose Estimation
  8. 2025FastDDHPose: Towards Unified, Efficient, and Disentangled 3D Human Pose Estimation