Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

DTSPL-BEV: Decomposable Tiled Soccer Player Localization

Persson, Ivar LU ; Chaman Motlagh, Abolfazl LU orcid and Nilsson, Mikael LU orcid (2026) p.483-490
Abstract
We present an extension of the SPL-BEV (Soccer Player Localization for Bird’s-Eye-View) method, introducing Decomposable Tiled SPL-BEV (DTSPL-BEV), a variant designed to enable efficient inference on embedded devices. We show specialized tiling algorithms that enable inference on a given devices with some known memory limitations, while also ensuring the same detection results as running inference on the entire image. The tiling algorithms optimise tile configuration to minimises computational cost in terms of FLOPs, given memory constraints. We also introduce several potential network architecture changes that enable performance on par with SPL-BEV while reducing the computational cost. The code is available at... (More)
We present an extension of the SPL-BEV (Soccer Player Localization for Bird’s-Eye-View) method, introducing Decomposable Tiled SPL-BEV (DTSPL-BEV), a variant designed to enable efficient inference on embedded devices. We show specialized tiling algorithms that enable inference on a given devices with some known memory limitations, while also ensuring the same detection results as running inference on the entire image. The tiling algorithms optimise tile configuration to minimises computational cost in terms of FLOPs, given memory constraints. We also introduce several potential network architecture changes that enable performance on par with SPL-BEV while reducing the computational cost. The code is available at https://github.com/IvarPersson/SPL-BEV. (Less)
Please use this url to cite or link to this publication:
author
; and
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
keywords
Object Detection, Bird-Eye-View (BEV), Decomposable Methods, Sports Analytics
host publication
Proceedings of the 15th International Conference on Pattern Recognition Applications and Methods (ICPRAM 2026)
pages
8 pages
ISBN
978-989-758-797-9
DOI
10.5220/0014468500004067
language
English
LU publication?
yes
id
68a9979e-bc59-4c10-96d8-bd70ee6fb143
date added to LUP
2026-04-16 20:35:06
date last changed
2026-04-21 11:50:55
@inproceedings{68a9979e-bc59-4c10-96d8-bd70ee6fb143,
  abstract     = {{We present an extension of the SPL-BEV (Soccer Player Localization for Bird’s-Eye-View) method, introducing Decomposable Tiled SPL-BEV (DTSPL-BEV), a variant designed to enable efficient inference on embedded devices. We show specialized tiling algorithms that enable inference on a given devices with some known memory limitations, while also ensuring the same detection results as running inference on the entire image. The tiling algorithms optimise tile configuration to minimises computational cost in terms of FLOPs, given memory constraints. We also introduce several potential network architecture changes that enable performance on par with SPL-BEV while reducing the computational cost. The code is available at https://github.com/IvarPersson/SPL-BEV.}},
  author       = {{Persson, Ivar and Chaman Motlagh, Abolfazl and Nilsson, Mikael}},
  booktitle    = {{Proceedings of the 15th International Conference on Pattern Recognition Applications and Methods (ICPRAM 2026)}},
  isbn         = {{978-989-758-797-9}},
  keywords     = {{Object Detection; Bird-Eye-View (BEV); Decomposable Methods; Sports Analytics}},
  language     = {{eng}},
  pages        = {{483--490}},
  title        = {{DTSPL-BEV: Decomposable Tiled Soccer Player Localization}},
  url          = {{http://dx.doi.org/10.5220/0014468500004067}},
  doi          = {{10.5220/0014468500004067}},
  year         = {{2026}},
}