PLRF-NMS : A Piecewise Linear Rational Function in Non-Maximum Suppression
Persson, Ivar LU ; Ardö, Håkan LU and Nilsson, Mikael LU
(2025)
15622.
p.340-350
- Abstract
- Activation functions are fundamental components in neural networks, enabling non-linear transformations essential for tasks like signal processing, control systems, image analysis, economics, and robotics. They play a crucial role in facilitating processes such as noise reduction, segmentation, and decision-making across various applications. Splines offer an alternative approach to traditional activation functions (e.g., ReLU or Sigmoid), providing flexibility and adaptability to enhance function approximation. In this work a specific spline, the Piecewise Linear Fractional Function (PLRF), is introduced and proposed as a re-scoring mechanism for soft Non-Maximum Suppression (NMS) in object detection pipelines. The PLRF is parametrized... (More)
- Activation functions are fundamental components in neural networks, enabling non-linear transformations essential for tasks like signal processing, control systems, image analysis, economics, and robotics. They play a crucial role in facilitating processes such as noise reduction, segmentation, and decision-making across various applications. Splines offer an alternative approach to traditional activation functions (e.g., ReLU or Sigmoid), providing flexibility and adaptability to enhance function approximation. In this work a specific spline, the Piecewise Linear Fractional Function (PLRF), is introduced and proposed as a re-scoring mechanism for soft Non-Maximum Suppression (NMS) in object detection pipelines. The PLRF is parametrized with up to four hyperparameters within the range (0, 1) and the paper presents two black-box optimization techniques, GridFib and HybridNM, to refine hyperparameters. Experimental results on two different datasets indicate that the PLRF achieves higher scores compared to Greedy-NMS and Soft-NMS methods. Furthermore, the number of function evaluations needed with the proposed optimization methods reduces computational evaluations needed relative to the Bayesian optimization technique commonly used in this context. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/90a7eddb-74c9-421f-a3c8-2ad81d4998c0
- author
- Persson, Ivar
LU
; Ardö, Håkan
LU
and Nilsson, Mikael
LU
- organization
- publishing date
- 2025-10-25
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- host publication
- Computer Analysis of Images and Patterns : 21st International Conference, CAIP 2025, Las Palmas de Gran Canaria, Spain, September 22–25, 2025, Proceedings, Part II - 21st International Conference, CAIP 2025, Las Palmas de Gran Canaria, Spain, September 22–25, 2025, Proceedings, Part II
- volume
- 15622
- pages
- 11 pages
- publisher
- Springer Nature
- external identifiers
-
- scopus:105024561285
- DOI
- 10.1007/978-3-032-05060-1_29
- language
- English
- LU publication?
- yes
- id
- 90a7eddb-74c9-421f-a3c8-2ad81d4998c0
- date added to LUP
- 2025-10-03 13:48:25
- date last changed
- 2026-01-23 10:11:30
@inproceedings{90a7eddb-74c9-421f-a3c8-2ad81d4998c0,
abstract = {{Activation functions are fundamental components in neural networks, enabling non-linear transformations essential for tasks like signal processing, control systems, image analysis, economics, and robotics. They play a crucial role in facilitating processes such as noise reduction, segmentation, and decision-making across various applications. Splines offer an alternative approach to traditional activation functions (e.g., ReLU or Sigmoid), providing flexibility and adaptability to enhance function approximation. In this work a specific spline, the Piecewise Linear Fractional Function (PLRF), is introduced and proposed as a re-scoring mechanism for soft Non-Maximum Suppression (NMS) in object detection pipelines. The PLRF is parametrized with up to four hyperparameters within the range (0, 1) and the paper presents two black-box optimization techniques, GridFib and HybridNM, to refine hyperparameters. Experimental results on two different datasets indicate that the PLRF achieves higher scores compared to Greedy-NMS and Soft-NMS methods. Furthermore, the number of function evaluations needed with the proposed optimization methods reduces computational evaluations needed relative to the Bayesian optimization technique commonly used in this context.}},
author = {{Persson, Ivar and Ardö, Håkan and Nilsson, Mikael}},
booktitle = {{Computer Analysis of Images and Patterns : 21st International Conference, CAIP 2025, Las Palmas de Gran Canaria, Spain, September 22–25, 2025, Proceedings, Part II}},
language = {{eng}},
month = {{10}},
pages = {{340--350}},
publisher = {{Springer Nature}},
title = {{PLRF-NMS : A Piecewise Linear Rational Function in Non-Maximum Suppression}},
url = {{http://dx.doi.org/10.1007/978-3-032-05060-1_29}},
doi = {{10.1007/978-3-032-05060-1_29}},
volume = {{15622}},
year = {{2025}},
}