Motion-DVAE: Unsupervised learning for fast human motion denoising

TL;DR

Motion-DVAE is an unsupervised, real-time human motion denoising method that models short-term dependencies using a dynamical VAE, combining generative and temporal modeling for improved efficiency.

Contribution

It introduces Motion-DVAE, a novel motion prior that captures short-term dependencies and enables fast, unsupervised denoising for real-time 3D human pose estimation.

Findings

Achieves competitive accuracy with state-of-the-art methods.

Operates significantly faster than existing approaches.

Unifies regression and optimization-based denoising in a single framework.

Abstract

Pose and motion priors are crucial for recovering realistic and accurate human motion from noisy observations. Substantial progress has been made on pose and shape estimation from images, and recent works showed impressive results using priors to refine frame-wise predictions. However, a lot of motion priors only model transitions between consecutive poses and are used in time-consuming optimization procedures, which is problematic for many applications requiring real-time motion capture. We introduce Motion-DVAE, a motion prior to capture the short-term dependencies of human motion. As part of the dynamical variational autoencoder (DVAE) models family, Motion-DVAE combines the generative capability of VAE models and the temporal modeling of recurrent architectures. Together with Motion-DVAE, we introduce an unsupervised learned denoising method unifying regression- and optimization-based approaches in a single framework for real-time 3D human pose estimation. Experiments show that the proposed approach reaches competitive performance with state-of-the-art methods while being much faster.