Deep Inertial Pose: A deep learning approach for human pose estimation

TL;DR

This paper explores neural network methods for human pose estimation using inertial sensors, demonstrating that deep learning can achieve accuracy comparable to traditional biomechanical models and filters.

Contribution

It compares various neural network architectures for inertial pose estimation and identifies the most effective hybrid LSTM-Madgwick approach, including an ablation study on key factors.

Findings

Hybrid LSTM-Madgwick achieved 7.96 quaternion angle error with high-end sensors

Neural networks can approximate complex biomechanical models for pose estimation

Data augmentation and output choices significantly impact accuracy

Abstract

Inertial-based Motion capture system has been attracting growing attention due to its wearability and unsconstrained use. However, accurate human joint estimation demands several complex and expertise demanding steps, which leads to expensive software such as the state-of-the-art MVN Awinda from Xsens Technologies. This work aims to study the use of Neural Networks to abstract the complex biomechanical models and analytical mathematics required for pose estimation. Thus, it presents a comparison of different Neural Network architectures and methodologies to understand how accurately these methods can estimate human pose, using both low cost(MPU9250) and high end (Mtw Awinda) Magnetic, Angular Rate, and Gravity (MARG) sensors. The most efficient method was the Hybrid LSTM-Madgwick detached, which achieved an Quaternion Angle distance error of 7.96, using Mtw Awinda data. Also, an ablation study was conducted to study the impact of data augmentation, output representation, window size, loss function and magnetometer data on the pose estimation error. This work indicates that Neural Networks can be trained to estimate human pose, with results comparable to the state-of-the-art fusion filters.