Validation of deep learning-based markerless 3D pose estimation

Veronika Kosourikhina; Diarmuid Kavanagh; Michael J. Richardson; David M. Kaplan

doi:10.1371/journal.pone.0276258

Validation of deep learning-based markerless 3D pose estimation

Veronika Kosourikhina, Diarmuid Kavanagh, Michael J. Richardson, David M. Kaplan

Research output: Contribution to journal › Article › peer-review

13 Citations (Scopus)

54 Downloads (Pure)

Abstract

Deep learning-based approaches to markerless 3D pose estimation are being adopted by researchers in psychology and neuroscience at an unprecedented rate. Yet many of these tools remain unvalidated. Here, we report on the validation of one increasingly popular tool (DeepLabCut) against simultaneous measurements obtained from a reference measurement system (Fastrak) with well-known performance characteristics. Our results confirm close (mm range) agreement between the two, indicating that under specific circumstances deep learning-based approaches can match more traditional motion tracking methods. Although more work needs to be done to determine their specific performance characteristics and limitations, this study should help build confidence within the research community using these new tools.

Original language	English
Article number	e0276258
Pages (from-to)	1-11
Number of pages	11
Journal	PLoS ONE
Volume	17
Issue number	10
DOIs	https://doi.org/10.1371/journal.pone.0276258
Publication status	Published - Oct 2022

Bibliographical note

Copyright the Author(s) 2022. Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.

Access to Document

10.1371/journal.pone.0276258Licence: CC BY

Publisher version (open access)Final published version, 1.12 MBLicence: CC BY

Cite this

@article{752e7277c8da439f972cb769ad84e364,

title = "Validation of deep learning-based markerless 3D pose estimation",

abstract = "Deep learning-based approaches to markerless 3D pose estimation are being adopted by researchers in psychology and neuroscience at an unprecedented rate. Yet many of these tools remain unvalidated. Here, we report on the validation of one increasingly popular tool (DeepLabCut) against simultaneous measurements obtained from a reference measurement system (Fastrak) with well-known performance characteristics. Our results confirm close (mm range) agreement between the two, indicating that under specific circumstances deep learning-based approaches can match more traditional motion tracking methods. Although more work needs to be done to determine their specific performance characteristics and limitations, this study should help build confidence within the research community using these new tools.",

author = "Veronika Kosourikhina and Diarmuid Kavanagh and Richardson, \{Michael J.\} and Kaplan, \{David M.\}",

note = "Copyright the Author(s) 2022. Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.",

year = "2022",

month = oct,

doi = "10.1371/journal.pone.0276258",

language = "English",

volume = "17",

pages = "1--11",

journal = "PLoS ONE",

issn = "1932-6203",

publisher = "Public Library of Science",

number = "10",

}

TY - JOUR

T1 - Validation of deep learning-based markerless 3D pose estimation

AU - Kosourikhina, Veronika

AU - Kavanagh, Diarmuid

AU - Richardson, Michael J.

AU - Kaplan, David M.

N1 - Copyright the Author(s) 2022. Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.

PY - 2022/10

Y1 - 2022/10

N2 - Deep learning-based approaches to markerless 3D pose estimation are being adopted by researchers in psychology and neuroscience at an unprecedented rate. Yet many of these tools remain unvalidated. Here, we report on the validation of one increasingly popular tool (DeepLabCut) against simultaneous measurements obtained from a reference measurement system (Fastrak) with well-known performance characteristics. Our results confirm close (mm range) agreement between the two, indicating that under specific circumstances deep learning-based approaches can match more traditional motion tracking methods. Although more work needs to be done to determine their specific performance characteristics and limitations, this study should help build confidence within the research community using these new tools.

AB - Deep learning-based approaches to markerless 3D pose estimation are being adopted by researchers in psychology and neuroscience at an unprecedented rate. Yet many of these tools remain unvalidated. Here, we report on the validation of one increasingly popular tool (DeepLabCut) against simultaneous measurements obtained from a reference measurement system (Fastrak) with well-known performance characteristics. Our results confirm close (mm range) agreement between the two, indicating that under specific circumstances deep learning-based approaches can match more traditional motion tracking methods. Although more work needs to be done to determine their specific performance characteristics and limitations, this study should help build confidence within the research community using these new tools.

UR - https://www.scopus.com/pages/publications/85140282800

U2 - 10.1371/journal.pone.0276258

DO - 10.1371/journal.pone.0276258

M3 - Article

C2 - 36264853

AN - SCOPUS:85140282800

SN - 1932-6203

VL - 17

SP - 1

EP - 11

JO - PLoS ONE

JF - PLoS ONE

IS - 10

M1 - e0276258

ER -

Validation of deep learning-based markerless 3D pose estimation

Abstract

Bibliographical note

Access to Document

Other files and links

Fingerprint

Cite this