Attention Guided Anomaly Localization in Images
Open accessThe claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
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?
The method in [216] is only applicable to Isolation Forest.
“Only applicable to Isolation Forest”
From A Survey on Explainable Anomaly Detection (2024) · cites this paper as [216]
The Conclusion (p. 14) says the paper’s attention losses are applicable to different network architectures, contrary to the claimed restriction to Isolation Forest. The Introduction (p. 2) identifies the proposed method as a convolutional adversarial variational autoencoder with guided attention, not an Isolation Forest method. (LLM, conf=0.90)
Checked by a reasoning model — may be wrong.
Experimental — accuracy not yet measuredThe method in [216] has high explanation fidelity.
“High explanation fidelity”
From A Survey on Explainable Anomaly Detection (2024) · cites this paper as [216]
The Method (p. 6) describes attention maps used to localize anomalies, and Experimental Results (p. 11) report localization IoU and AuROC. Those results support the maps’ localization performance, but the paper does not evaluate whether they faithfully explain the model’s decisions; high explanation fidelity is therefore not established. (LLM, conf=0.76)
Experimental — accuracy not yet measuredBoth CutPaste [31] and CAVGA [42] treat anomaly detection as a classification problem, while GradCAM is used for pixel-level anomaly localization.
“Both CutPaste [31] and CAVGA [42] treat anomaly detection as a classification problem, while GradCAM is used for pixel-level anomaly localization.”
From Deep Industrial Image Anomaly Detection: A Survey (2024) · cites this paper as [42]
Existing surveys on XAD contain no more than 40 relevant works each.
“Specifically, each of the above surveys contains no more than 40 relevant works in the field.”
From A Survey on Explainable Anomaly Detection (2024) · cites this paper as [216]
The Related Works section (p. 4) compares anomaly-localization methods, but does not count works covered by XAD surveys. Searches for surveys, XAD, and the 40-work limit found no passage in our copy addressing the claim. (LLM, conf=0.76)
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
The citation lists haven’t been fetched yet
References 68 · Cited by 247
Research works printed but not linked: 8
- [3][3] Akcay, S., Atapour-Abarghouei, A., Breckon, T.P.: GANomaly: Semisupervised anomaly detection via adversarial training. In: Asian Conference on Computer Vision. pp. 622-637. Springer (2018) -
- [13][13] Dehaene, D., Frigo, O., Combrexelle, S., Eline, P.: Iterative energy-based projection on a normal data manifold for anomaly localization. International Conference on Learning Representations (2020) -
- [14][14] Dieng, A.B., Kim, Y., Rush, A.M., Blei, D.M.: Avoiding latent variable collapse with generative skip models. In: The 22nd International Conference on Artificial Intelligence and Statistics. pp. 2397-2405 (2019) -
- [17][17] Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., Bengio, Y.: Generative adversarial nets. In: Advances in neural information processing systems. pp. 2672-2680 (2014) -
- [24][24] Kiran, B., Thomas, D., Parakkal, R.: An overview of deep learning based methods for unsupervised and semi-supervised anomaly detection in videos. Journal of Imaging 4 (2), 36 (2018) -
- [42][42] Radford, A., Metz, L., Chintala, S.: Unsupervised representation learning with deep convolutional generative adversarial networks. In: International Conference on Learning Representations (2016) -
- [44][44] Ruff, L., Vandermeulen, R.A., G¨ ornitz, N., Binder, A., M¨ uller, E., M¨ uller, K.R., Kloft, M.: Deep semi-supervised anomaly detection. International Conference on Learning Representations (2020) -
- [46][46] Sabokrou, M., Khalooei, M., Fathy, M., Adeli, E.: Adversarially learned oneclass classifier for novelty detection. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. pp. 3379-3388 (2018) -
- Authors
- Venkataramanan, S., Peng, K.-C., Singh, R. V., & Mahalanobis, A.
- Affiliations
- University of Central Florida (US)
- Mitsubishi Electric (United States) (US)
- Siemens (Germany) (DE)
- Corresponding author
- Venkataramanan, Shashanka
- Venue
- Lecture Notes in Computer Science
- Publisher
- Springer International Publishing
- Published
- 2020
- Pages
- 485-503
- ISSN
- 0302-9743 (ISSN-L)
- 1611-3349
- OpenAlex
- W3012693338
- Cited by
- 247
- References
- 68
- Citations by year
- 2026: 10 · 2025: 35 · 2024: 57 · 2023: 61 · 2022: 40 · 2021: 38 · 2020: 6
- FWCI
- 13.85
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- Lecture notes in computer science (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- Defense Advanced Research Projects Agency
- 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
- anomaly localization, anomaly detection, attention guidance, unsupervised anomaly detection, weakly supervised anomaly detection, glaucoma detection, industrial inspection, surveillance
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
