ReactionMamba: Generating Short & Long Human Reaction Sequences

TL;DR

ReactionMamba is a new framework that efficiently generates realistic long and short 3D human reaction motions, including complex activities like dance and martial arts, with improved speed and diversity.

Contribution

It introduces a novel combination of motion VAE and Mamba-based models for long sequence generation of human reactions, outperforming previous methods in realism and speed.

Findings

Competitive realism and diversity in generated motions

Superior inference speed over prior methods

Effective long-sequence generation for complex activities

Abstract

We present ReactionMamba, a novel framework for generating long 3D human reaction motions. Reaction-Mamba integrates a motion VAE for efficient motion encoding with Mamba-based state-space models to decode temporally consistent reactions. This design enables ReactionMamba to generate both short sequences of simple motions and long sequences of complex motions, such as dance and martial arts. We evaluate ReactionMamba on three datasets--NTU120-AS, Lindy Hop, and InterX--and demonstrate competitive performance in terms of realism, diversity, and long-sequence generation compared to previous methods, including InterFormer, ReMoS, and Ready-to-React, while achieving substantial improvements in inference speed.