Out-of-Distribution Detection for Adaptive Computer Vision
(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:
https://lup.lub.lu.se/record/a0e476b9-cbc6-483c-9aed-58b39efb1c3e
- author
- Kristoffersson Lind, Simon LU ; Triebel, Rudolph ; Nardi, Luigi LU and Krueger, Volker LU
- organization
- publishing date
- 2023
- 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}}, }