Spatio-Temporal Multi-Subgraph GCN for 3D Human Motion Prediction

TL;DR

This paper introduces STMS-GCN, a novel graph convolutional network that effectively models complex spatio-temporal dependencies in 3D human motion prediction by decoupling and cross-referencing spatial and temporal features.

Contribution

The paper proposes a new multi-subgraph GCN architecture that decouples spatial and temporal modeling and employs cross-domain transfer and multiple subgraphs for richer motion representation.

Findings

Outperforms existing methods on standard benchmarks.

Effectively captures complex spatio-temporal dependencies.

Enhances motion prediction accuracy through multi-scale feature transfer.

Abstract

Human motion prediction (HMP) involves forecasting future human motion based on historical data. Graph Convolutional Networks (GCNs) have garnered widespread attention in this field for their proficiency in capturing relationships among joints in human motion. However, existing GCN-based methods tend to focus on either temporal-domain or spatial-domain features, or they combine spatio-temporal features without fully leveraging the complementarity and cross-dependency of these two features. In this paper, we propose the Spatial-Temporal Multi-Subgraph Graph Convolutional Network (STMS-GCN) to capture complex spatio-temporal dependencies in human motion. Specifically, we decouple the modeling of temporal and spatial dependencies, enabling cross-domain knowledge transfer at multiple scales through a spatio-temporal information consistency constraint mechanism. Besides, we utilize multiple subgraphs to extract richer motion information and enhance the learning associations of diverse subgraphs through a homogeneous information constraint mechanism. Extensive experiments on the standard HMP benchmarks demonstrate the superiority of our method.