Runtime Monitoring for Out-of-Distribution Detection in Object
Detection Neural Networks

D 2023

Runtime Monitoring for Out-of-Distribution Detection in Object Detection Neural Networks

HASHEMI, Vahid; Jan KŘETÍNSKÝ; Sabine RIEDER a Jessica SCHMIDT

Základní údaje

Originální název

Runtime Monitoring for Out-of-Distribution Detection in Object Detection Neural Networks

Autoři

HASHEMI, Vahid; Jan KŘETÍNSKÝ; Sabine RIEDER a Jessica SCHMIDT

Vydání

Lübeck, FORMAL METHODS, FM 2023, od s. 622-634, 13 s. 2023

Nakladatel

SPRINGER INTERNATIONAL PUBLISHING AG

Další údaje

Jazyk

angličtina

Typ výsledku

Stať ve sborníku

Obor

10201 Computer sciences, information science, bioinformatics

Stát vydavatele

Německo

Utajení

není předmětem státního či obchodního tajemství

Forma vydání

tištěná verze "print"

Impakt faktor

Impact factor: 0.402 v roce 2005

Označené pro přenos do RIV

Organizační jednotka

Fakulta informatiky

ISBN

978-3-031-27480-0

ISSN

Klíčová slova anglicky

Runtime monitoring; Neural networks; Out-of-distribution detection; Object detection

Příznaky

Mezinárodní význam, Recenzováno

Změněno: 5. 2. 2026 09:26, Sabine Rieder, M.Sc.

Anotace

V originále

Runtime monitoring provides a more realistic and applicable alternative to verification in the setting of real neural networks used in industry. It is particularly useful for detecting out-of-distribution (OOD) inputs, for which the network was not trained and can yield erroneous results. We extend a runtime-monitoring approach previously proposed for classification networks to perception systems capable of identification and localization of multiple objects. Furthermore, we analyze its adequacy experimentally on different kinds of OOD settings, documenting the overall efficacy of our approach.

Návaznosti

MUNI/A/1081/2022, interní kód MU

Název: Modelování, analýza a verifikace (2023)

Investor: Masarykova univerzita, Modelování, analýza a verifikace (2023)

MUNI/A/1433/2022, interní kód MU

Název: Zapojení studentů Fakulty informatiky do mezinárodní vědecké komunity 23

Investor: Masarykova univerzita, Zapojení studentů Fakulty informatiky do mezinárodní vědecké komunity 23

Citovat

HASHEMI, Vahid; Jan KŘETÍNSKÝ; Sabine RIEDER a Jessica SCHMIDT. Runtime Monitoring for Out-of-Distribution Detection in Object Detection Neural Networks. In 978-3-031-27481-7. FORMAL METHODS, FM 2023. Lübeck: SPRINGER INTERNATIONAL PUBLISHING AG, 2023, s. 622-634. ISBN 978-3-031-27480-0. Dostupné z: https://doi.org/10.1007/978-3-031-27481-7_36.

@inproceedings{2392228,
   author = {Hashemi, Vahid and Křetínský, Jan and Rieder, Sabine and Schmidt, Jessica},
   address = {Lübeck},
   booktitle = {FORMAL METHODS, FM 2023},
   doi = {https://doi.org/10.1007/978-3-031-27481-7_36},
   editor = {978-3-031-27481-7},
   keywords = {Runtime monitoring; Neural networks; Out-of-distribution detection; Object detection},
   howpublished = {tištěná verze "print"},
   language = {eng},
   location = {Lübeck},
   isbn = {978-3-031-27480-0},
   pages = {622-634},
   publisher = {SPRINGER INTERNATIONAL PUBLISHING AG},
   title = {Runtime Monitoring for Out-of-Distribution Detection in Object Detection Neural Networks},
   year = {2023}
}

TY  - CONF
ID  - 2392228
AU  - Hashemi, Vahid - Křetínský, Jan - Rieder, Sabine - Schmidt, Jessica
PY  - 2023
TI  - Runtime Monitoring for Out-of-Distribution Detection in Object Detection Neural Networks
PB  - SPRINGER INTERNATIONAL PUBLISHING AG
CY  - Lübeck
SN  - 9783031274800
KW  - Runtime monitoring
KW  - Neural networks
KW  - Out-of-distribution detection
KW  - Object detection
N2  - Runtime monitoring provides a more realistic and applicable alternative to verification in the setting of real neural networks used in industry. It is particularly useful for detecting out-of-distribution (OOD) inputs, for which the network was not trained and can yield erroneous results. We extend a runtime-monitoring approach previously proposed for classification networks to perception systems capable of identification and localization of multiple objects. Furthermore, we analyze its adequacy experimentally on different kinds of OOD settings, documenting the overall efficacy of our approach.
ER  -

HASHEMI, Vahid; Jan KŘETÍNSKÝ; Sabine RIEDER a Jessica SCHMIDT. Runtime Monitoring for Out-of-Distribution Detection in Object Detection Neural Networks. In 978-3-031-27481-7. \textit{FORMAL METHODS, FM 2023}. Lübeck: SPRINGER INTERNATIONAL PUBLISHING AG, 2023, s.~622-634. ISBN~978-3-031-27480-0. Dostupné z: https://doi.org/10.1007/978-3-031-27481-7\_{}36.

Přehled o publikaci