Recurrent Network Models for Human Dynamics

TL;DR

The paper introduces the ERD model, a recurrent neural network architecture that improves human pose recognition and prediction in videos and motion capture data, outperforming previous models in accuracy and robustness.

Contribution

The ERD model extends LSTM architectures with nonlinear encoder and decoder networks, enabling joint learning of representations and dynamics for human motion tasks.

Findings

ERD outperforms per frame detectors in pose labeling.

ERD accurately predicts joint displacements over 400ms.

ERD synthesizes novel motions and maintains long-term stability.

Abstract

We propose the Encoder-Recurrent-Decoder (ERD) model for recognition and prediction of human body pose in videos and motion capture. The ERD model is a recurrent neural network that incorporates nonlinear encoder and decoder networks before and after recurrent layers. We test instantiations of ERD architectures in the tasks of motion capture (mocap) generation, body pose labeling and body pose forecasting in videos. Our model handles mocap training data across multiple subjects and activity domains, and synthesizes novel motions while avoid drifting for long periods of time. For human pose labeling, ERD outperforms a per frame body part detector by resolving left-right body part confusions. For video pose forecasting, ERD predicts body joint displacements across a temporal horizon of 400ms and outperforms a first order motion model based on optical flow. ERDs extend previous Long Short Term Memory (LSTM) models in the literature to jointly learn representations and their dynamics. Our experiments show such representation learning is crucial for both labeling and prediction in space-time. We find this is a distinguishing feature between the spatio-temporal visual domain in comparison to 1D text, speech or handwriting, where straightforward hard coded representations have shown excellent results when directly combined with recurrent units.