Temporally optimized inverse kinematics for 6DOF human pose estimation
(2022) ESB 2022- Abstract
- The recent emergence of deep learning and computer vision-based 3D human pose estimation presents opportunities for a form of markerless motion-capture. State-of-the-art approaches have achieved remarkable accuracy in predicting global and relative joint positions, however, many potential applications require information on joint orientations, e.g., in the fields of biomechanics. Furthermore, methods that do include joint orientations are incompatible for applications with predefined incongruent kinematic chains, i.e., chains with differing limb lengths and proportions. Therefore, we propose a temporal inverse kinematics (IK) optimization technique to infer joint orientations in a user customizable kinematic chain from a position based... (More)
- The recent emergence of deep learning and computer vision-based 3D human pose estimation presents opportunities for a form of markerless motion-capture. State-of-the-art approaches have achieved remarkable accuracy in predicting global and relative joint positions, however, many potential applications require information on joint orientations, e.g., in the fields of biomechanics. Furthermore, methods that do include joint orientations are incompatible for applications with predefined incongruent kinematic chains, i.e., chains with differing limb lengths and proportions. Therefore, we propose a temporal inverse kinematics (IK) optimization technique to infer joint orientations in a user customizable kinematic chain from a position based 3D pose input. This technique may be particularly useful for sports/injury biomechanics, and telehealth applications. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/c2fb8ddd-ee7b-4b8a-a79d-4927759f8301
- author
- Gildea, Kevin LU ; Mercadal-Baudart, Clara ; Blythman, Richard and Simms, Ciaran
- publishing date
- 2022
- type
- Contribution to conference
- publication status
- published
- subject
- keywords
- Human pose estimation, Computer vision, Inverse kinematics, Motion capture, Biomechanics
- pages
- 1 pages
- conference name
- ESB 2022
- conference location
- Porto, Portugal
- conference dates
- 2022-06-26 - 2022-06-29
- language
- English
- LU publication?
- no
- id
- c2fb8ddd-ee7b-4b8a-a79d-4927759f8301
- date added to LUP
- 2022-12-13 20:10:31
- date last changed
- 2023-06-09 11:51:26
@misc{c2fb8ddd-ee7b-4b8a-a79d-4927759f8301,
abstract = {{The recent emergence of deep learning and computer vision-based 3D human pose estimation presents opportunities for a form of markerless motion-capture. State-of-the-art approaches have achieved remarkable accuracy in predicting global and relative joint positions, however, many potential applications require information on joint orientations, e.g., in the fields of biomechanics. Furthermore, methods that do include joint orientations are incompatible for applications with predefined incongruent kinematic chains, i.e., chains with differing limb lengths and proportions. Therefore, we propose a temporal inverse kinematics (IK) optimization technique to infer joint orientations in a user customizable kinematic chain from a position based 3D pose input. This technique may be particularly useful for sports/injury biomechanics, and telehealth applications.}},
author = {{Gildea, Kevin and Mercadal-Baudart, Clara and Blythman, Richard and Simms, Ciaran}},
keywords = {{Human pose estimation; Computer vision; Inverse kinematics; Motion capture; Biomechanics}},
language = {{eng}},
title = {{Temporally optimized inverse kinematics for 6DOF human pose estimation}},
url = {{https://lup.lub.lu.se/search/files/150139635/AI_Biomechanics.pdf}},
year = {{2022}},
}