Structure-Aware Human-Action Generation

TL;DR

This paper introduces a structure-aware approach for generating long-range human actions by combining graph convolutional networks with self-attention to dynamically sparsify action graphs, improving the quality of generated sequences.

Contribution

The paper proposes a novel GCN variant that uses self-attention to adaptively sparsify action graphs, effectively capturing structure information in long-range skeleton-based human action generation.

Findings

Outperforms existing methods on standard datasets

Effectively captures structure information in action sequences

Demonstrates superior long-range action generation quality

Abstract

Generating long-range skeleton-based human actions has been a challenging problem since small deviations of one frame can cause a malformed action sequence. Most existing methods borrow ideas from video generation, which naively treat skeleton nodes/joints as pixels of images without considering the rich inter-frame and intra-frame structure information, leading to potential distorted actions. Graph convolutional networks (GCNs) is a promising way to leverage structure information to learn structure representations. However, directly adopting GCNs to tackle such continuous action sequences both in spatial and temporal spaces is challenging as the action graph could be huge. To overcome this issue, we propose a variant of GCNs to leverage the powerful self-attention mechanism to adaptively sparsify a complete action graph in the temporal space. Our method could dynamically attend to important past frames and construct a sparse graph to apply in the GCN framework, well-capturing the structure information in action sequences. Extensive experimental results demonstrate the superiority of our method on two standard human action datasets compared with existing methods.