Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MoViD: View-Invariant 3D Human Pose Estimation via Motion-View Disentanglement

Yejia Liu, Hengle Jiang, Haoxian Liu, Runxi Huang, Xiaomin Ouyang

Abstract

3D human pose estimation is a key enabling technology for applications such as healthcare monitoring, human-robot collaboration, and immersive gaming, but real-world deployment remains challenged by viewpoint variations. Existing methods struggle to generalize to unseen camera viewpoints, require large amounts of training data, and suffer from high inference latency. We propose MoViD, a viewpoint-invariant 3D human pose estimation framework that disentangles viewpoint information from motion features. The key idea is to extract viewpoint information from intermediate pose features and leverage it to enhance both the robustness and efficiency of pose estimation. MoViD introduces a view estimator that models key joint relationships to predict viewpoint information, and an orthogonal projection module to disentangle motion and view features, further enhanced through physics-grounded contrastive alignment across views. For real-time edge deployment, MoViD employs a frame-by-frame inference pipeline with a view-aware strategy that adaptively activates flip refinement based on the estimated viewpoint. Evaluations on nine public datasets and newly collected multiview UAV and gait analysis datasets show that MoViD reduces pose estimation error by over 24.2\% compared to state-of-the-art methods, maintains robust performance under severe occlusions with 60\% less training data, and achieves real-time inference at 15 FPS on NVIDIA edge devices.

MoViD: View-Invariant 3D Human Pose Estimation via Motion-View Disentanglement

Abstract

3D human pose estimation is a key enabling technology for applications such as healthcare monitoring, human-robot collaboration, and immersive gaming, but real-world deployment remains challenged by viewpoint variations. Existing methods struggle to generalize to unseen camera viewpoints, require large amounts of training data, and suffer from high inference latency. We propose MoViD, a viewpoint-invariant 3D human pose estimation framework that disentangles viewpoint information from motion features. The key idea is to extract viewpoint information from intermediate pose features and leverage it to enhance both the robustness and efficiency of pose estimation. MoViD introduces a view estimator that models key joint relationships to predict viewpoint information, and an orthogonal projection module to disentangle motion and view features, further enhanced through physics-grounded contrastive alignment across views. For real-time edge deployment, MoViD employs a frame-by-frame inference pipeline with a view-aware strategy that adaptively activates flip refinement based on the estimated viewpoint. Evaluations on nine public datasets and newly collected multiview UAV and gait analysis datasets show that MoViD reduces pose estimation error by over 24.2\% compared to state-of-the-art methods, maintains robust performance under severe occlusions with 60\% less training data, and achieves real-time inference at 15 FPS on NVIDIA edge devices.

Paper Structure

This paper contains 50 sections, 8 equations, 17 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. A Motivation Study
  4. Impact of Viewpoint Variations
  5. Potential of Viewpoint Estimation
  6. Embedded viewpoint information in pose features.
  7. Incorporating viewpoint information.
  8. Summary
  9. System Overview
  10. Applications and Challenges
  11. System Architecture
  12. Design of MoViD
  13. View Estimator
  14. Geometric feature extraction.
  15. View embedding extraction.
  16. ...and 35 more sections

Figures (17)

  • Figure 1: Comparison of the traditional framework for 3D human pose estimation and our proposed approach MoViD. The key idea of MoViD is to extract and leverage viewpoint information for view-invariant pose estimation.
  • Figure 2: Pose estimation errors across different views.
  • Figure 3: The intermediate pose features exhibit clear clustering patterns by viewpoint.
  • Figure 4: Incorporating view information ("View-aware") reduces pose estimation errors across views.
  • Figure 5: Typical application scenarios of MoViD: Activity recognition across diverse deployment settings and human-robot interaction during navigation.
  • ...and 12 more figures