Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Out-of-Distribution Detection for Adaptive Computer Vision

Kristoffersson Lind, Simon LU ; Triebel, Rudolph ; Nardi, Luigi LU and Krueger, Volker LU orcid (2023) 23nd Scandinavian Conference on Image Analysis, SCIA 2023 In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 13886 LNCS. p.311-325
Abstract

It is well known that computer vision can be unreliable when faced with previously unseen imaging conditions. This paper proposes a method to adapt camera parameters according to a normalizing flow-based out-of-distibution detector. A small-scale study is conducted which shows that adapting camera parameters according to this out-of-distibution detector leads to an average increase of 3 to 4% points in mAP, mAR and F1 performance metrics of a YOLOv4 object detector. As a secondary result, this paper also shows that it is possible to train a normalizing flow model for out-of-distribution detection on the COCO dataset, which is larger and more diverse than most benchmarks for out-of-distibution detectors.

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
Autonomous Systems, Normalizing Flows, Object Detection, Out-of-Distribution Detection
host publication
Image Analysis - 23rd Scandinavian Conference, SCIA 2023, Proceedings
series title
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
editor
Gade, Rikke ; Felsberg, Michael and Kämäräinen, Joni-Kristian
volume
13886 LNCS
pages
15 pages
publisher
Springer Science and Business Media B.V.
conference name
23nd Scandinavian Conference on Image Analysis, SCIA 2023
conference location
Lapland, Finland
conference dates
2023-04-18 - 2023-04-21
external identifiers
  • scopus:85161436449
ISSN
1611-3349
0302-9743
ISBN
9783031314377
DOI
10.1007/978-3-031-31438-4_21
language
English
LU publication?
yes
id
a0e476b9-cbc6-483c-9aed-58b39efb1c3e
date added to LUP
2023-08-22 14:36:49
date last changed
2024-04-20 01:17:04
@inproceedings{a0e476b9-cbc6-483c-9aed-58b39efb1c3e,
  abstract     = {{<p>It is well known that computer vision can be unreliable when faced with previously unseen imaging conditions. This paper proposes a method to adapt camera parameters according to a normalizing flow-based out-of-distibution detector. A small-scale study is conducted which shows that adapting camera parameters according to this out-of-distibution detector leads to an average increase of 3 to 4% points in mAP, mAR and F1 performance metrics of a YOLOv4 object detector. As a secondary result, this paper also shows that it is possible to train a normalizing flow model for out-of-distribution detection on the COCO dataset, which is larger and more diverse than most benchmarks for out-of-distibution detectors.</p>}},
  author       = {{Kristoffersson Lind, Simon and Triebel, Rudolph and Nardi, Luigi and Krueger, Volker}},
  booktitle    = {{Image Analysis - 23rd Scandinavian Conference, SCIA 2023, Proceedings}},
  editor       = {{Gade, Rikke and Felsberg, Michael and Kämäräinen, Joni-Kristian}},
  isbn         = {{9783031314377}},
  issn         = {{1611-3349}},
  keywords     = {{Autonomous Systems; Normalizing Flows; Object Detection; Out-of-Distribution Detection}},
  language     = {{eng}},
  pages        = {{311--325}},
  publisher    = {{Springer Science and Business Media B.V.}},
  series       = {{Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}},
  title        = {{Out-of-Distribution Detection for Adaptive Computer Vision}},
  url          = {{http://dx.doi.org/10.1007/978-3-031-31438-4_21}},
  doi          = {{10.1007/978-3-031-31438-4_21}},
  volume       = {{13886 LNCS}},
  year         = {{2023}},
}