A review of 3D human pose estimation algorithms for markerless motion capture

TL;DR

This paper reviews recent 3D human pose estimation algorithms for markerless motion capture, analyzing their accuracy, speed, and robustness to guide future research in the field.

Contribution

It provides a comprehensive taxonomy and comparison of recent methods, highlighting key metrics, benchmarks, and structural differences.

Findings

Average error per joint is around 20 mm in current methods.

Classification based on accuracy, speed, and robustness.

Guidelines for future research directions.

Abstract

Human pose estimation is a very active research field, stimulated by its important applications in robotics, entertainment or health and sports sciences, among others. Advances in convolutional networks triggered noticeable improvements in 2D pose estimation, leading modern 3D markerless motion capture techniques to an average error per joint of 20 mm. However, with the proliferation of methods, it is becoming increasingly difficult to make an informed choice. Here, we review the leading human pose estimation methods of the past five years, focusing on metrics, benchmarks and method structures. We propose a taxonomy based on accuracy, speed and robustness that we use to classify de methods and derive directions for future research.