Compositional Human Pose Regression

TL;DR

This paper introduces a structure-aware regression method for human pose estimation that leverages bone-based representations and compositional loss functions, improving accuracy in 2D and 3D pose tasks.

Contribution

It proposes a novel reparameterized pose representation and a compositional loss that better exploits pose structure, advancing regression-based human pose estimation.

Findings

Significantly improves state-of-the-art on Human3.6M

Achieves competitive results on MPII

Effective for both 2D and 3D pose estimation

Abstract

Regression based methods are not performing as well as detection based methods for human pose estimation. A central problem is that the structural information in the pose is not well exploited in the previous regression methods. In this work, we propose a structure-aware regression approach. It adopts a reparameterized pose representation using bones instead of joints. It exploits the joint connection structure to define a compositional loss function that encodes the long range interactions in the pose. It is simple, effective, and general for both 2D and 3D pose estimation in a unified setting. Comprehensive evaluation validates the effectiveness of our approach. It significantly advances the state-of-the-art on Human3.6M and is competitive with state-of-the-art results on MPII.