Investigating Pose Representations and Motion Contexts Modeling for 3D Motion Prediction

TL;DR

This paper studies how different skeletal pose representations affect 3D human motion prediction and introduces a novel RNN architecture, AHMR, that captures local and global motion contexts, improving short-term and long-term prediction accuracy.

Contribution

It provides an in-depth analysis of pose representations and proposes a new RNN model with geometric loss functions for enhanced motion prediction.

Findings

Outperforms state-of-the-art methods in short-term prediction

Achieves significantly better long-term motion retention over 50 seconds

Effective across various articulated objects like humans, fish, and mice

Abstract

Predicting human motion from historical pose sequence is crucial for a machine to succeed in intelligent interactions with humans. One aspect that has been obviated so far, is the fact that how we represent the skeletal pose has a critical impact on the prediction results. Yet there is no effort that investigates across different pose representation schemes. We conduct an indepth study on various pose representations with a focus on their effects on the motion prediction task. Moreover, recent approaches build upon off-the-shelf RNN units for motion prediction. These approaches process input pose sequence sequentially and inherently have difficulties in capturing long-term dependencies. In this paper, we propose a novel RNN architecture termed AHMR (Attentive Hierarchical Motion Recurrent network) for motion prediction which simultaneously models local motion contexts and a global context. We further explore a geodesic loss and a forward kinematics loss for the motion prediction task, which have more geometric significance than the widely employed L2 loss. Interestingly, we applied our method to a range of articulate objects including human, fish, and mouse. Empirical results show that our approach outperforms the state-of-the-art methods in short-term prediction and achieves much enhanced long-term prediction proficiency, such as retaining natural human-like motions over 50 seconds predictions. Our codes are released.