Zur Hauptnavigation wechseln Zur Suche wechseln Zum Hauptinhalt wechseln

Fully Convolutional Cross-Scale-Flows for Image-based Defect Detection

  • Marco Rudolph*
  • , Tom Wehrbein
  • , Bodo Rosenhahn
  • , Bastian Wandt
  • *Korrespondierende*r Autor*in für diese Arbeit

Publikation: Beitrag in Buch/Bericht/Sammelwerk/KonferenzbandAufsatz in KonferenzbandForschungPeer-Review

Abstract

In industrial manufacturing processes, errors frequently occur at unpredictable times and in unknown manifestations. We tackle the problem of automatic defect detection without requiring any image samples of defective parts. Recent works model the distribution of defect-free image data, using either strong statistical priors or overly simplified data representations. In contrast, our approach handles fine-grained representations incorporating the global and local image context while flexibly estimating the density. To this end, we propose a novel fully convolutional cross-scale normalizing flow (CS-Flow) that jointly processes multiple feature maps of different scales. Using normalizing flows to assign meaningful likelihoods to input samples allows for efficient defect detection on image-level. Moreover, due to the preserved spatial arrangement the latent space of the normalizing flow is interpretable which enables to localize defective regions in the image. Our work sets a new state-of-the-art in image-level defect detection on the benchmark datasets Magnetic Tile Defects and MVTec AD showing a 100% AUROC on 4 out of 15 classes.

OriginalspracheEnglisch
Titel des SammelwerksProceedings - 2022 IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2022
Herausgeber (Verlag)Institute of Electrical and Electronics Engineers Inc.
Seiten1829-1838
Seitenumfang10
ISBN (elektronisch)9781665409155
ISBN (Print)978-1-6654-0916-2
DOIs
PublikationsstatusVeröffentlicht - 2022
Veranstaltung22nd IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2022 - Waikoloa, USA / Vereinigte Staaten
Dauer: 3 Jan. 20228 Jan. 2022

Publikationsreihe

NameProceedings - 2022 IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2022
ISSN (Print)2472-6737
ISSN (elektronisch)2642-9381

Konferenz

Konferenz22nd IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2022
Land/GebietUSA / Vereinigte Staaten
OrtWaikoloa
Zeitraum3 Jan. 20228 Jan. 2022

ASJC Scopus Sachgebiete

  • Maschinelles Sehen und Mustererkennung
  • Angewandte Informatik

Dieses zitieren