Time Coherent Full-Body Poses Estimated Using Only Five Inertial Sensors: Deep versus Shallow Learning
| Autor: | Matteo Giuberti, Nina Rudigkeit, Peter H. Veltink, Bert-Jan van Beijnum, Mannes Poel, Frank J. Wouda |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
030506 rehabilitation
Computer science Movement Acceleration Posture 02 engineering and technology Biosensing Techniques lcsh:Chemical technology human movement pose estimation Biochemistry Motion capture Article Analytical Chemistry Machine Learning 03 medical and health sciences Inertial measurement unit 0202 electrical engineering electronic engineering information engineering Humans lcsh:TP1-1185 Electrical and Electronic Engineering Instrumentation Pose Gait Monitoring Physiologic Human Body Artificial neural network business.industry Deep learning time coherence deep learning 020207 software engineering Rigid body neural networks Atomic and Molecular Physics and Optics reduced sensor set Artificial intelligence Neural Networks Computer inertial motion capture 0305 other medical science business LSTM Algorithm Algorithms |
| Zdroj: | Sensors (Basel, Switzerland) Sensors, Vol 19, Iss 17, p 3716 (2019) Sensors (Switzerland), 19:3716. MDPI Sensors Volume 19 Issue 17 |
| ISSN: | 1424-8220 |
| Popis: | Full-body motion capture typically requires sensors/markers to be placed on each rigid body segment, which results in long setup times and is obtrusive. The number of sensors/markers can be reduced using deep learning or offline methods. However, this requires large training datasets and/or sufficient computational resources. Therefore, we investigate the following research question: &ldquo What is the performance of a shallow approach, compared to a deep learning one, for estimating time coherent full-body poses using only five inertial sensors?&rdquo We propose to incorporate past/future inertial sensor information into a stacked input vector, which is fed to a shallow neural network for estimating full-body poses. Shallow and deep learning approaches are compared using the same input vector configurations. Additionally, the inclusion of acceleration input is evaluated. The results show that a shallow learning approach can estimate full-body poses with a similar accuracy (~6 cm) to that of a deep learning approach (~7 cm). However, the jerk errors are smaller using the deep learning approach, which can be the effect of explicit recurrent modelling. Furthermore, it is shown that the delay using a shallow learning approach (72 ms) is smaller than that of a deep learning approach (117 ms). |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
| Nepřihlášeným uživatelům se plný text nezobrazuje | K zobrazení výsledku je třeba se přihlásit. |