ManiPose: Manifold-Constrained Multi-Hypothesis 3D Human Pose Estimation

TL;DR

ManiPose introduces a manifold-constrained multi-hypothesis approach for monocular 3D human pose estimation, effectively addressing depth ambiguity and pose consistency issues, outperforming existing methods on real-world datasets.

Contribution

The paper presents a novel multi-hypothesis model that constrains outputs to the human pose manifold, improving pose consistency without using generative models.

Findings

Outperforms state-of-the-art in pose consistency metrics

Maintains competitive MPJPE performance

Addresses depth ambiguity effectively

Abstract

We propose ManiPose, a manifold-constrained multi-hypothesis model for human-pose 2D-to-3D lifting. We provide theoretical and empirical evidence that, due to the depth ambiguity inherent to monocular 3D human pose estimation, traditional regression models suffer from pose-topology consistency issues, which standard evaluation metrics (MPJPE, P-MPJPE and PCK) fail to assess. ManiPose addresses depth ambiguity by proposing multiple candidate 3D poses for each 2D input, each with its estimated plausibility. Unlike previous multi-hypothesis approaches, ManiPose forgoes generative models, greatly facilitating its training and usage. By constraining the outputs to lie on the human pose manifold, ManiPose guarantees the consistency of all hypothetical poses, in contrast to previous works. We showcase the performance of ManiPose on real-world datasets, where it outperforms state-of-the-art models in pose consistency by a large margin while being very competitive on the MPJPE metric.