CoChat
Sign Up
Cited by 1,086Industrial Vision Systems and Defect DetectionAnomaly Detection Techniques and ApplicationsInfrastructure Maintenance and Monitoring

DRÆM – A discriminatively trained reconstruction embedding for surface anomaly detection

Zavrtanik, Vitjan, Kristan, Matej, Skocaj, Danijel·2021·2021 IEEE/CVF International Conference on Computer Vision (ICCV)
DOI 10.1109/iccv48922.2021.00822
Open this paper in CoChat — free account

Abstract

Visual surface anomaly detection aims to detect local image regions that significantly deviate from normal appearance. Recent surface anomaly detection methods rely on generative models to accurately reconstruct the normal areas and to fail on anomalies. These methods are trained only on anomaly-free images, and often require hand-crafted post-processing steps to localize the anomalies, which prohibits optimizing the feature extraction for maximal detection capability. In addition to reconstructive approach, we cast surface anomaly detection primarily as a discriminative problem and propose a discriminatively trained reconstruction anomaly embedding model (DRÆM). The proposed method learns a joint representation of an anomalous image and its anomaly-free reconstruction, while simultaneously learning a decision boundary between normal and anomalous examples. The method enables direct anomaly localization without the need for additional complicated post-processing of the network output and can be trained using simple and general anomaly simulations. On the challenging MVTec anomaly detection dataset, DRÆM outperforms the current state-of-the-art unsupervised methods by a large margin and even de-livers detection performance close to the fully-supervised methods on the widely used DAGM surface-defect detection dataset, while substantially outperforming them in localization accuracy. Code at github.com/VitjanZ/DRAEM.

Integrity signals

Registry

Crossref · 10.1109/iccv48922.2021.00822

Cites

29 of 29 references verified0 citation errors · 0 not checkable

Statistics

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

Claims the paper makes

This paper’s full text isn’t openly available, 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?

  • DRAEM uses U-Net to outperform other methods on pixel-level anomaly segmentation.

    “DRAEM [91] also uses U-Net to outperform other methods on pixel-level anomaly segmentation.”

    From Deep Industrial Image Anomaly Detection: A Survey (2024) · cites this paper as [91]

  • Many unsupervised anomaly detection works use data augmentation to synthesize anomaly images and significantly improve model performance.

    “There are many unsupervised anomaly detection works that use data augmentation to synthetic anomaly images and significantly improve model performance.”

    From Deep Industrial Image Anomaly Detection: A Survey (2024) · cites this paper as [91]

  • CutPaste, DRAEM, and MemSeg are representative methods of unsupervised anomaly detection using data augmentation to synthesize anomaly images.

    “For examples, CutPaste [31] , DRAEM [91] , MemSeg [18] are representative methods.”

    From Deep Industrial Image Anomaly Detection: A Survey (2024) · cites this paper as [91]

  • DRAEM ceases to be effective under the unified case.

    “However, DRAEM [53] ceases to be effective under the unified case.”

    From A Unified Model for Multi-Class Anomaly Detection (2022) · cites this paper as [53]

  • Existing unsupervised anomaly detection methods train separate models for different object classes.

    “Despite the rapid advance of unsupervised anomaly detection, existing methods require to train separate models for different objects.”

    From A Unified Model for Multi-Class Anomaly Detection (2022) · cites this paper as [53]

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Zavrtanik, V., Kristan, M., & Skocaj, D.
Affiliations
  • University of Ljubljana (SI)
  • University of Ljubljana,Faculty of Computer and Information Science
ORCID
Kristan, Matej
Venue
2021 IEEE/CVF International Conference on Computer Vision (ICCV)
Publisher
IEEE
Published
2021-10
Pages
8310-8319
DOI
10.1109/iccv48922.2021.00822
OpenAlex
W4214694907
Cited by
1,086
References
29
Citations by year
2026: 298 · 2025: 342 · 2024: 293 · 2023: 122 · 2022: 30 · 2021: 1
FWCI
153.09
Type
conference paper
Language
en
Open access
closed
Licence
Locations
IEEE/CVF International Conference on Computer Vision (ICCV) (publishedVersion)
Topics
  • Physical Sciences › Engineering › Industrial and Manufacturing Engineering › Industrial Vision Systems and Defect Detection
  • Physical Sciences › Computer Science › Artificial Intelligence › Anomaly Detection Techniques and Applications
  • Physical Sciences › Engineering › Civil and Structural Engineering › Infrastructure Maintenance and Monitoring
Keywords
surface anomaly detection, image anomaly detection, anomaly localization, unsupervised anomaly detection, surface defect detection, decision boundary learning

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