Better Prior Knowledge Improves Human-Pose-Based Extrinsic Camera Calibration
Moliner, Olivier LU
; Huang, Sangxia
and Åström, Kalle
LU
(2021)
2020 25th International Conference on Pattern Recognition
In International Conference on Pattern Recognition
p.4758-4765
- Abstract
- Accurate extrinsic calibration of wide baseline multi-camera systems enables better understanding of 3D scenes for many applications and is of great practical importance. Classical Structure-from-Motion calibration methods require special calibration equipment so that accurate point correspondences can be detected between different views. In addition, an operator with some training is usually needed to ensure that data is collected in a way that leads to good calibration accuracy. This limits the ease of adoption of such technologies. Recently, methods have been proposed to use human pose estimation models to establish point correspondences, thus removing the need for any special equipment. The challenge with this approach is that human... (More)
- Accurate extrinsic calibration of wide baseline multi-camera systems enables better understanding of 3D scenes for many applications and is of great practical importance. Classical Structure-from-Motion calibration methods require special calibration equipment so that accurate point correspondences can be detected between different views. In addition, an operator with some training is usually needed to ensure that data is collected in a way that leads to good calibration accuracy. This limits the ease of adoption of such technologies. Recently, methods have been proposed to use human pose estimation models to establish point correspondences, thus removing the need for any special equipment. The challenge with this approach is that human pose estimation algorithms typically produce much less accurate feature points compared to classical patch-based methods. Another problem is that ambient human motion might not be optimal for calibration. We build upon prior works and introduce several novel ideas to improve the accuracy of human-pose-based extrinsic calibration. Our first contribution is a robust reprojection loss based on a better understanding of the sources of pose estimation error. Our second contribution is a 3D human pose likelihood model learned from motion capture data. We demonstrate significant improvements in calibration accuracy by evaluating our method on four publicly available datasets. (Less)
- Abstract (Swedish)
- Accurate extrinsic calibration of wide baseline multi-camera systems enables better understanding of 3D scenes for many applications and is of great practical importance. Classical Structure-from-Motion calibration methods require special calibration equipment so that accurate point correspondences can be detected between different views. In addition, an operator with some training is usually needed to ensure that data is collected in a way that leads to good calibration accuracy. This limits the ease of adoption of such technologies. Recently, methods have been proposed to use human pose estimation models to establish point correspondences, thus removing the need for any special equipment. The challenge with this approach is that human... (More)
- Accurate extrinsic calibration of wide baseline multi-camera systems enables better understanding of 3D scenes for many applications and is of great practical importance. Classical Structure-from-Motion calibration methods require special calibration equipment so that accurate point correspondences can be detected between different views. In addition, an operator with some training is usually needed to ensure that data is collected in a way that leads to good calibration accuracy. This limits the ease of adoption of such technologies. Recently, methods have been proposed to use human pose estimation models to establish point correspondences, thus removing the need for any special equipment. The challenge with this approach is that human pose estimation algorithms typically produce much less accurate feature points compared to classical patch-based methods. Another problem is that ambient human motion might not be optimal for calibration. We build upon prior works and introduce several novel ideas to improve the accuracy of human-pose-based extrinsic calibration. Our first contribution is a robust reprojection loss based on a better understanding of the sources of pose estimation error. Our second contribution is a 3D human pose likelihood model learned from motion capture data. We demonstrate significant improvements in calibration accuracy by evaluating our method on four publicly available datasets. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/a353c156-7910-483a-960c-abf85cf70ed4
- author
- Moliner, Olivier
LU
; Huang, Sangxia
and Åström, Kalle
LU
- organization
- publishing date
- 2021-05-05
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- host publication
- 2020 25th International Conference on Pattern Recognition (ICPR)
- series title
- International Conference on Pattern Recognition
- pages
- 8 pages
- publisher
- IEEE - Institute of Electrical and Electronics Engineers Inc.
- conference name
- 2020 25th International Conference on Pattern Recognition
- conference location
- Milan, Italy
- conference dates
- 2021-01-10 - 2021-01-15
- external identifiers
-
- scopus:85110456147
- ISSN
- 1051-4651
- ISBN
- 978-1-7281-8808-9
- DOI
- 10.1109/ICPR48806.2021.9411927
- project
- WASP: Wallenberg AI, Autonomous Systems and Software Program at Lund University
- language
- English
- LU publication?
- yes
- id
- a353c156-7910-483a-960c-abf85cf70ed4
- date added to LUP
- 2021-05-27 11:13:32
- date last changed
- 2023-09-12 20:25:44
@inproceedings{a353c156-7910-483a-960c-abf85cf70ed4,
abstract = {{Accurate extrinsic calibration of wide baseline multi-camera systems enables better understanding of 3D scenes for many applications and is of great practical importance. Classical Structure-from-Motion calibration methods require special calibration equipment so that accurate point correspondences can be detected between different views. In addition, an operator with some training is usually needed to ensure that data is collected in a way that leads to good calibration accuracy. This limits the ease of adoption of such technologies. Recently, methods have been proposed to use human pose estimation models to establish point correspondences, thus removing the need for any special equipment. The challenge with this approach is that human pose estimation algorithms typically produce much less accurate feature points compared to classical patch-based methods. Another problem is that ambient human motion might not be optimal for calibration. We build upon prior works and introduce several novel ideas to improve the accuracy of human-pose-based extrinsic calibration. Our first contribution is a robust reprojection loss based on a better understanding of the sources of pose estimation error. Our second contribution is a 3D human pose likelihood model learned from motion capture data. We demonstrate significant improvements in calibration accuracy by evaluating our method on four publicly available datasets.}},
author = {{Moliner, Olivier and Huang, Sangxia and Åström, Kalle}},
booktitle = {{2020 25th International Conference on Pattern Recognition (ICPR)}},
isbn = {{978-1-7281-8808-9}},
issn = {{1051-4651}},
language = {{eng}},
month = {{05}},
pages = {{4758--4765}},
publisher = {{IEEE - Institute of Electrical and Electronics Engineers Inc.}},
series = {{International Conference on Pattern Recognition}},
title = {{Better Prior Knowledge Improves Human-Pose-Based Extrinsic Camera Calibration}},
url = {{http://dx.doi.org/10.1109/ICPR48806.2021.9411927}},
doi = {{10.1109/ICPR48806.2021.9411927}},
year = {{2021}},
}