Human Pose as Compositional Tokens

TL;DR

This paper introduces Pose as Compositional Tokens (PCT), a structured representation of human pose that models joint dependencies, enabling more accurate and occlusion-robust pose estimation through a classification approach.

Contribution

It proposes a novel compositional token-based pose representation and formulates pose estimation as a classification task, improving accuracy and robustness over existing methods.

Findings

Achieves comparable or better pose estimation accuracy.

Performs well under occlusion conditions.

Uses a classification-based approach for pose estimation.

Abstract

Human pose is typically represented by a coordinate vector of body joints or their heatmap embeddings. While easy for data processing, unrealistic pose estimates are admitted due to the lack of dependency modeling between the body joints. In this paper, we present a structured representation, named Pose as Compositional Tokens (PCT), to explore the joint dependency. It represents a pose by M discrete tokens with each characterizing a sub-structure with several interdependent joints. The compositional design enables it to achieve a small reconstruction error at a low cost. Then we cast pose estimation as a classification task. In particular, we learn a classifier to predict the categories of the M tokens from an image. A pre-learned decoder network is used to recover the pose from the tokens without further post-processing. We show that it achieves better or comparable pose estimation results as the existing methods in general scenarios, yet continues to work well when occlusion occurs, which is ubiquitous in practice. The code and models are publicly available at https://github.com/Gengzigang/PCT.