GloPro: Globally-Consistent Uncertainty-Aware 3D Human Pose Estimation & Tracking in the Wild

TL;DR

GloPro is a novel framework that predicts uncertainty distributions for 3D human body meshes, including shape, pose, and root pose, improving accuracy and consistency in real-time human tracking in the wild.

Contribution

It introduces the first method to jointly estimate uncertainty for shape, pose, and root pose in 3D human mesh prediction, integrating visual cues with a motion model.

Findings

Outperforms state-of-the-art in trajectory accuracy

Provides consistent uncertainty estimates

Operates in real-time

Abstract

An accurate and uncertainty-aware 3D human body pose estimation is key to enabling truly safe but efficient human-robot interactions. Current uncertainty-aware methods in 3D human pose estimation are limited to predicting the uncertainty of the body posture, while effectively neglecting the body shape and root pose. In this work, we present GloPro, which to the best of our knowledge the first framework to predict an uncertainty distribution of a 3D body mesh including its shape, pose, and root pose, by efficiently fusing visual clues with a learned motion model. We demonstrate that it vastly outperforms state-of-the-art methods in terms of human trajectory accuracy in a world coordinate system (even in the presence of severe occlusions), yields consistent uncertainty distributions, and can run in real-time.