Neural Human Pose Prior

TL;DR

This paper presents a neural prior for human body poses using normalizing flows, enabling flexible and stable modeling of 6D rotations, which improves human motion capture and reconstruction.

Contribution

It introduces a novel, data-driven neural prior with RealNVP for 6D rotations, addressing manifold constraints and ensuring framework-agnostic, reproducible training.

Findings

Effective pose prior modeling demonstrated through qualitative evaluations.

Quantitative improvements in pose estimation accuracy.

Ablation studies highlight the impact of the proposed approach.

Abstract

We introduce a principled, data-driven approach for modeling a neural prior over human body poses using normalizing flows. Unlike heuristic or low-expressivity alternatives, our method leverages RealNVP to learn a flexible density over poses represented in the 6D rotation format. We address the challenge of modeling distributions on the manifold of valid 6D rotations by inverting the Gram-Schmidt process during training, enabling stable learning while preserving downstream compatibility with rotation-based frameworks. Our architecture and training pipeline are framework-agnostic and easily reproducible. We demonstrate the effectiveness of the learned prior through both qualitative and quantitative evaluations, and we analyze its impact via ablation studies. This work provides a sound probabilistic foundation for integrating pose priors into human motion capture and reconstruction pipelines.