QPoser: Quantized Explicit Pose Prior Modeling for Controllable Pose Generation

TL;DR

QPoser introduces a controllable explicit pose prior model that guarantees correctness and expressiveness, enabling detailed and conditional human pose generation with improved accuracy over existing methods.

Contribution

The paper proposes QPoser, a novel pose prior model combining multi-head vector quantized autoencoder and global-local feature integration for enhanced controllability, correctness, and expressiveness.

Findings

Outperforms state-of-the-art in pose representation accuracy

Enables detailed conditional pose generation

Maintains physical plausibility of generated poses

Abstract

Explicit pose prior models compress human poses into latent representations for using in pose-related downstream tasks. A desirable explicit pose prior model should satisfy three desirable abilities: 1) correctness, i.e. ensuring to generate physically possible poses; 2) expressiveness, i.e. ensuring to preserve details in generation; 3) controllability, meaning that generation from reference poses and explicit instructions should be convenient. Existing explicit pose prior models fail to achieve all of three properties, in special controllability. To break this situation, we propose QPoser, a highly controllable explicit pose prior model which guarantees correctness and expressiveness. In QPoser, a multi-head vector quantized autoencoder (MS-VQVAE) is proposed for obtaining expressive and distributed pose representations. Furthermore, a global-local feature integration mechanism (GLIF-AE) is utilized to disentangle the latent representation and integrate full-body information into local-joint features. Experimental results show that QPoser significantly outperforms state-of-the-art approaches in representing expressive and correct poses, meanwhile is easily to be used for detailed conditional generation from reference poses and prompting instructions.