mlsanityData to 7 Oct 2026

Learning from Abstract Images: on the Importance of Occlusion in a Minimalist Encoding of Human Poses

arXiv 2307.09893 ↗First posted 19 Jul 2023cs.CVcited by 0

Not on the map. Why?

Abstract (arXiv)

Existing 2D-to-3D pose lifting networks suffer from poor performance in cross-dataset benchmarks. Although the use of 2D keypoints joined by "stick-figure" limbs has shown promise as an intermediate step, stick-figures do not account for occlusion information that is often inherent in an image. In this paper, we propose a novel representation using opaque 3D limbs that preserves occlusion information while implicitly encoding joint locations. Crucially, when training on data with accurate three-dimensional keypoints and without part-maps, this representation allows training on abstract synthetic images, with occlusion, from as many synthetic viewpoints as desired. The result is a pose defined by limb angles rather than joint positions unicodex2013 because poses are, in the real world, independent of cameras unicodex2013 allowing us to predict poses that are completely independent of camera viewpoint. The result provides not only an improvement in same-dataset benchmarks, but a "quantum leap" in cross-dataset benchmarks.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 30 of its 41 references that are in the map. First posted on arXiv 19 Jul 2023.

Builds on

30 of 41 references.

References · 30

  1. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  2. 2019Deep High-Resolution Representation Learning for Human Pose Estimationcited by 1,339
  3. 2017A simple yet effective baseline for 3d human pose estimationcited by 401
  4. 20183D human pose estimation in video with temporal convolutions and semi-supervised trainingcited by 401
  5. 2017Learning from Synthetic Humanscited by 358
  6. 2017Learning to Estimate 3D Hand Pose from Single RGB Imagescited by 234
  7. 2016Coarse-to-Fine Volumetric Prediction for Single-Image 3D Human Posecited by 220
  8. 2019Semantic Graph Convolutional Networks for 3D Human Pose Regressioncited by 204
  9. 2017VNect: Real-time 3D Human Pose Estimation with a Single RGB Cameracited by 129
  10. 2017Compositional Human Pose Regressioncited by 124
  11. 20163D Human Pose Estimation = 2D Pose Estimation + Matchingcited by 112
  12. 2018Ordinal Depth Supervision for 3D Human Pose Estimationcited by 107
18 more
  1. 2019Unsupervised 3D Pose Estimation with Geometric Self-Supervisioncited by 95
  2. 2020SRNet: Improving Generalization in 3D Human Pose Estimation with a Split-and-Recombine Approachcited by 94
  3. 2019Generating Multiple Hypotheses for 3D Human Pose Estimation with Mixture Density Networkcited by 92
  4. 2021PoseAug: A Differentiable Pose Augmentation Framework for 3D Human Pose Estimationcited by 88
  5. 20183D Human Pose Estimation in the Wild by Adversarial Learningcited by 86
  6. 2019RepNet: Weakly Supervised Training of an Adversarial Reprojection Network for 3D Human Pose Estimationcited by 85
  7. 2019Monocular 3D Human Pose Estimation by Generation and Ordinal Rankingcited by 76
  8. 20163D Human Pose Estimation from a Single Image via Distance Matrix Regressioncited by 73
  9. 2019HEMlets Pose: Learning Part-Centric Heatmap Triplets for Accurate 3D Human Pose Estimationcited by 50
  10. 2019In the Wild Human Pose Estimation Using Explicit 2D Features and Intermediate 3D Representationscited by 48
  11. 2020Self-Supervised 3D Human Pose Estimation via Part Guided Novel Image Synthesiscited by 44
  12. 2016Structured Prediction of 3D Human Pose with Deep Neural Networkscited by 43
  13. 2016Deep Kinematic Pose Regressioncited by 40
  14. 2015Maximum-Margin Structured Learning with Deep Networks for 3D Human Pose Estimationcited by 37
  15. 2020Predicting Camera Viewpoint Improves Cross-dataset Generalization for 3D Human Pose Estimationcited by 36
  16. 2021ElePose: Unsupervised 3D Human Pose Estimation by Predicting Camera Elevation and Learning Normalizing Flows on 2D Posescited by 35
  17. 2018DRPose3D: Depth Ranking in 3D Human Pose Estimationcited by 9
  18. 2016Learning camera viewpoint using CNN to improve 3D body pose estimationcited by 6

Built on by

Not cited yet.