TY - JOUR

T1 - Sparse Inertial Poser

T2 - Automatic 3D Human Pose Estimation from Sparse IMUs

AU - von Marcard, Timo

AU - Rosenhahn, Bodo

AU - Black, Michael J.

AU - Pons-Moll, Gerard

N1 - Funding Information: This work is partly funded by the DFG-Project RO 2497/11-1. Authors gratefully acknowledge the sup-port. We thank Timo Bolkart, Laura Sevilla, Sergi Pujades, NaureenMahmood, Melanie Feldhofer and Osman Ulusoy for proofreading,Bastian Wandt and Aron Sommer for help with motion recordings,Talha Zaman for voice recordings, Alejandra Quiros for providingthe bodies from words and Senya Polikovsky, Andrea Keller andJorge Marquez for technical support.

PY - 2017/5/23

Y1 - 2017/5/23

N2 - We address the problem of making human motion capture in the wild more practical by using a small set of inertial sensors attached to the body. Since the problem is heavily under-constrained, previous methods either use a large number of sensors, which is intrusive, or they require additional video input. We take a different approach and constrain the problem by: (i) making use of a realistic statistical body model that includes anthropometric constraints and (ii) using a joint optimization framework to fit the model to orientation and acceleration measurements over multiple frames. The resulting tracker Sparse Inertial Poser (SIP) enables motion capture using only 6 sensors (attached to the wrists, lower legs, back and head) and works for arbitrary human motions. Experiments on the recently released TNT15 dataset show that, using the same number of sensors, SIP achieves higher accuracy than the dataset baseline without using any video data. We further demonstrate the effectiveness of SIP on newly recorded challenging motions in outdoor scenarios such as climbing or jumping over a wall.

AB - We address the problem of making human motion capture in the wild more practical by using a small set of inertial sensors attached to the body. Since the problem is heavily under-constrained, previous methods either use a large number of sensors, which is intrusive, or they require additional video input. We take a different approach and constrain the problem by: (i) making use of a realistic statistical body model that includes anthropometric constraints and (ii) using a joint optimization framework to fit the model to orientation and acceleration measurements over multiple frames. The resulting tracker Sparse Inertial Poser (SIP) enables motion capture using only 6 sensors (attached to the wrists, lower legs, back and head) and works for arbitrary human motions. Experiments on the recently released TNT15 dataset show that, using the same number of sensors, SIP achieves higher accuracy than the dataset baseline without using any video data. We further demonstrate the effectiveness of SIP on newly recorded challenging motions in outdoor scenarios such as climbing or jumping over a wall.

KW - Categories and Subject Descriptors (according to ACM CCS)

KW - I.3.3 [Computer Graphics]: Three-Dimensional Graphics and Realism—Animation

UR - http://www.scopus.com/inward/record.url?scp=85019702578&partnerID=8YFLogxK

U2 - 10.48550/arXiv.1703.08014

DO - 10.48550/arXiv.1703.08014

M3 - Article

AN - SCOPUS:85019702578

VL - 36

SP - 349

EP - 360

JO - Computer graphics forum

JF - Computer graphics forum

SN - 0167-7055

IS - 2

ER -