Temporal Dynamics Decoupling with Inverse Processing for Enhancing Human Motion Prediction

TL;DR

This paper introduces TD2IP, a novel human motion prediction method that decouples reconstruction and prediction tasks using separate decoders and employs inverse processing to leverage bidirectional temporal correlations, improving motion understanding.

Contribution

The paper proposes a new approach, TD2IP, that separates reconstruction and prediction decoding and uses inverse processing to better model temporal dependencies in human motion prediction.

Findings

TD2IP improves prediction accuracy across multiple datasets.

Decoupling decoding tasks reduces conflicts and enhances motion modeling.

Inverse processing leverages bidirectional temporal information effectively.

Abstract

Exploring the bridge between historical and future motion behaviors remains a central challenge in human motion prediction. While most existing methods incorporate a reconstruction task as an auxiliary task into the decoder, thereby improving the modeling of spatio-temporal dependencies, they overlook the potential conflicts between reconstruction and prediction tasks. In this paper, we propose a novel approach: Temporal Decoupling Decoding with Inverse Processing (\textbf{$TD^2IP$}). Our method strategically separates reconstruction and prediction decoding processes, employing distinct decoders to decode the shared motion features into historical or future sequences. Additionally, inverse processing reverses motion information in the temporal dimension and reintroduces it into the model, leveraging the bidirectional temporal correlation of human motion behaviors. By alleviating the conflicts between reconstruction and prediction tasks and enhancing the association of historical and future information, \textbf{$TD^2IP$} fosters a deeper understanding of motion patterns. Extensive experiments demonstrate the adaptability of our method within existing methods.