Active Learning for Bayesian 3D Hand Pose Estimation

TL;DR

This paper introduces a Bayesian deep learning framework for 3D hand pose estimation that effectively models uncertainties, improves accuracy over baselines, and employs active learning to reduce data requirements, with a novel acquisition function.

Contribution

It presents a Bayesian approximation for 3D hand pose estimation, analyzes uncertainties, and introduces a new acquisition function for active learning to minimize data needs.

Findings

Outperforms baseline estimators on benchmarks.

Effective uncertainty modeling improves pose estimation.

Active learning reduces data requirements significantly.

Abstract

We propose a Bayesian approximation to a deep learning architecture for 3D hand pose estimation. Through this framework, we explore and analyse the two types of uncertainties that are influenced either by data or by the learning capability. Furthermore, we draw comparisons against the standard estimator over three popular benchmarks. The first contribution lies in outperforming the baseline while in the second part we address the active learning application. We also show that with a newly proposed acquisition function, our Bayesian 3D hand pose estimator obtains lowest errors with the least amount of data. The underlying code is publicly available at https://github.com/razvancaramalau/al_bhpe.