CoChat
Sign Up
Cited by 990Anomaly Detection Techniques and ApplicationsIndustrial Vision Systems and Defect DetectionAdvanced Neural Network Applications

Uninformed Students: Student-Teacher Anomaly Detection With Discriminative Latent Embeddings

Open access
Bergmann, Paul, Fauser, Michael, Sattlegger, David, Steger, Carsten·2020·2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
DOI 10.1109/cvpr42600.2020.00424
Open this paper in CoChat — free accountOpen PDF

Abstract

We introduce a powerful student-teacher framework for the challenging problem of unsupervised anomaly detection and pixel-precise anomaly segmentation in high-resolution images. Student networks are trained to regress the output of a descriptive teacher network that was pretrained on a large dataset of patches from natural images. This circumvents the need for prior data annotation. Anomalies are detected when the outputs of the student networks differ from that of the teacher network. This happens when they fail to generalize outside the manifold of anomaly-free training data. The intrinsic uncertainty in the student networks is used as an additional scoring function that indicates anomalies. We compare our method to a large number of existing deep learning based methods for unsupervised anomaly detection. Our experiments demonstrate improvements over state-of-the-art methods on a number of real-world datasets, including the recently introduced MVTec Anomaly Detection dataset that was specifically designed to benchmark anomaly segmentation algorithms.

Integrity signals

Registry

Crossref · 10.1109/cvpr42600.2020.00424

Cites

5 of 5 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

We couldn’t retrieve this paper’s full text, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Bergmann et al. developed a student-teacher learning scheme with knowledge distillation for image anomaly detection and localization, achieving state-of-the-art results with a single network.

    “Recently P.Bergmann et.al [8] developed a novel network and training scheme for both image anomaly detection and localization. The approach uses a student-teacher learning scheme and knowledge distillation for achieving stateof-the-art results and a single network for both classification and pixel level anomaly localization.”

    From VT-ADL: A Vision Transformer Network for Image Anomaly Detection and Localization (2021) · cites this paper as [8]

  • On MVTec, the proposed method performs at par with the most recent state-of-the-art algorithms and outperforms them in 7 product categories.

    “Our proposed methods performed at par with the most recent state of the art algorithms (results taken from [8]), and even outperformed them in 7 product categories.”

    From VT-ADL: A Vision Transformer Network for Image Anomaly Detection and Localization (2021) · cites this paper as [8]

  • The Bergmann et al. approach uses a complex training scheme with a high number of student networks, demanding high computing resources for industrial applications.

    “The work achieves good results, but it uses a complex training scheme with high number of student networks, which again demands high computing resources for the industrial applications.”

    From VT-ADL: A Vision Transformer Network for Image Anomaly Detection and Localization (2021) · cites this paper as [8]

  • Feature-based approaches commonly employ networks pre-trained on ImageNet to map original images into a more distinguishable feature space, with low training costs and outstanding performance.

    “Feature-based approaches [17,18,19] commonly employ networks pre-trained on ImageNet [20] to map original images into a more distinguishable feature space. These methods are typically characterized by low training costs and outstanding performance.”

    From An Unsupervised Method for Industrial Image Anomaly Detection with Vision Transformer-Based Autoencoder (2024) · cites this paper as [17]

  • Bergmann et al. [10] are the first to use teacher-student architecture for anomaly detection.

    “Bergmann et al. [10] are the first to use teacher-student architecture for anomaly detection.”

    From Deep Industrial Image Anomaly Detection: A Survey (2024) · cites this paper as [10]

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Authors
Bergmann, P., Fauser, M., Sattlegger, D., & Steger, C.
Affiliations
MVTec Software GmbH
ORCID
Venue
2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
Publisher
IEEE
Published
2020-06
Pages
4182-4191
ISSN
1063-6919 (ISSN-L)
DOI
10.1109/cvpr42600.2020.00424
OpenAlex
W3034314048
Cited by
990
References
5
Citations by year
2026: 141 · 2025: 235 · 2024: 252 · 2023: 174 · 2022: 105 · 2021: 72 · 2020: 10 · 2019: 1
FWCI
34.48
Type
conference paper
Language
en
Open access
Licence
Locations
  • IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • arXiv (Cornell University)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Anomaly Detection Techniques and Applications
  • Physical Sciences › Engineering › Industrial and Manufacturing Engineering › Industrial Vision Systems and Defect Detection
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
Keywords
unsupervised anomaly detection, anomaly segmentation, student-teacher networks, knowledge distillation, latent space embedding, uncertainty estimation, high-resolution imagery

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review