Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

Boosting Fine-Grained Visual Anomaly Detection with Coarse-Knowledge-Aware Adversarial Learning

T0 review · 4 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read This paper claims that a small, image-level-labeled anomaly set, used through energy-based adversarial learning, can align reconstructed features with normal features and thereby improve fine-grained anomaly detection and localization.

desk verdict A practical weak-supervision method with solid medical results, but the patch-level alignment theorem proves less than claimed. read the letter →

arxiv 2412.12850 v1 pith:VTYXMZAW submitted 2024-12-17 cs.CV cs.AIcs.LG

classification cs.CVcs.AIcs.LG
keywords visualanomalydetectionfine-graineddefectlocalizationcoarseknowledgeenergy-baseddiscriminatoradversariallearningfeaturereconstructionweaklysupervisedmedicalimageanalysis
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Reconstruction-based anomaly detectors learn to reproduce normal images and flag deviations, but neural autoencoders can also reconstruct defects, so error maps miss fine-grained anomalies. This paper claims that a small dataset of images known to be anomalous, labeled only at the image level, is enough to fix that weakness. The proposed method adds an energy-based discriminator that assigns low energy to normal features and high energy to anomalous and reconstructed features; the autoencoder is then pushed to output features that look normal even when the input is anomalous. A patch-level version treats every patch of an anomalous image as anomalous, and the paper proves that this noisy rule still aligns reconstructed patch features with normal patch features at equilibrium. On four medical and two industrial datasets the method improves both detection and localization compared with unsupervised reconstruction baselines and weakly supervised competitors.

What carries the argument

The load-bearing object is the energy-based discriminator, a network that maps a feature map or a spatial patch feature to a nonnegative energy value. It is trained with a hinge-style objective that pushes energy down on normal features and up on generated and anomalous features, while the autoencoder plays the generator and is rewarded for producing low-energy output. The patch-level variant applies the same per-layer discriminator at every spatial location, so each spatial feature vector acts as a patch feature; this is what allows localization without any patch-level labels. The proof machinery is a total-variation-distance argument: the optimal discriminator for each fixed generator is characterized via a Hahn decomposition bound, and substituting it into the generator loss shows the unique equilibrium places generated features exactly on the normal distribution.

What would settle it

Measure the empirical total variation distance between the reconstructed patch features of anomalous images and the patch features of normal images after training; the theorem predicts this distance can be driven to zero at equilibrium, so a statistically significant remaining gap would falsify the alignment claim. As a complementary check, train the same model with deliberately shuffled image-level anomaly labels: if detection and localization improve just as much, the coarse labels are not doing the work the paper claims.

Watch

Extended reading notes

Core claim

The paper's central claim is that coarse anomaly knowledge—knowing that an image is anomalous without knowing where—can be converted into a fine-grained localization signal for reconstruction-based detection. The conversion mechanism is distribution alignment: instead of merely enlarging reconstruction error on anomalous images, the autoencoder is trained so that its generated feature maps, whether computed from normal or anomalous inputs, follow the distribution of normal feature maps. The paper proves two equilibrium results for the proposed energy-based adversarial game. At the image level, there is a Nash equilibrium in which the generated feature distribution equals the normal feature distribution; at the patch level, even though an anomalous image is a mixture of normal and anomalous patches and the mixture ratio is unknown, there is again an equilibrium in which generated patch features match normal patch features. Consequently, at equilibrium the autoencoder 'repairs' anomalous regions into normal-looking features while leaving normal regions essentially unchanged, making the cosine-distance error map a reliable anomaly map.

Load-bearing premise

Everything depends on normal and anomalous image regions having completely separate learned features, on the adversarial training actually reaching the equilibrium whose existence is proved, and on elastically distorted normal images being a valid stand-in for real defects in the industrial experiments.

Editorial extensions

If this is right

  • With as little as 1% of training images labeled anomalous at image level, the method raises image-level AUC over the unsupervised reconstruction baseline on the medical datasets, for example by 3.51% on ISIC2018 and by 7.63% on Chest X-ray.
  • Because the patch-level discriminator works without pixel labels, the same coarse-labeled anomaly set improves pixel-level localization, not just image-level detection, on medical and industrial data.
  • The benefit grows with the amount and variety of labeled anomalies: increasing the labeled anomaly ratio from 1% to 5% and the number of seen anomaly types from 1 to 3 consistently improves performance in the medical experiments.
  • The alignment perspective gives reconstruction-based detectors a principled way to use incomplete anomaly knowledge: at equilibrium, anomalous regions should be reconstructed as normal features, producing high error exactly where the defects are.
  • In the industrial setting, where no real anomaly images are available, the method is evaluated using elastically distorted normal images as pseudo-anomalies, and it remains competitive with strong unsupervised detectors.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The theorem's disjoint-support assumption suggests that the practical gain should track how well the pretrained features separate normal from anomalous patches; measuring that overlap with a classifier trained on held-out patches could predict when the method will help.
  • If the alignment story is the operative mechanism, the same energy-based regularizer should transfer to other reconstruction backbones and feature levels; a direct test is to swap ResNet features for another pretrained encoder and check whether the performance gain tracks alignment quality rather than architecture.
  • The patch-level result is essentially a noisy-label robustness theorem: treating all patches of an anomalous image as anomalous works even when most patches are normal, so the same principle could apply to other weakly supervised localization problems where negative bags contain mostly positives.
  • The industrial experiments rely on elastic distortions as stand-ins for defects; a natural extension is to generate pseudo-anomalies from normal images with local cuts, noise, or color shifts and use the same coarse-knowledge adversarial loss, making the method applicable where no real anomaly collection exists.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 4 minor

Summary. The paper proposes CKAAD, a feature-reconstruction anomaly detection method that trains an auto-encoder with the help of a small, image-level-labeled anomalous dataset. An energy-based discriminator regularizes the generator so that reconstructed features and, at patch level, reconstructed patch features are pushed toward the distribution of normal features. The main theoretical contribution is a pair of theorems claiming that, under disjoint-support assumptions, the adversarial training admits a Nash equilibrium in which the generated feature distribution exactly equals the normal feature distribution, even though patch-level anomaly labels are absent. The method is evaluated on four medical datasets (ISIC2018, Chest X-ray, Br35H, OCT) and two industrial datasets (MVTec AD, VisA), with ablations comparing reconstruction-only, reconstruction-minus-anomaly, pure GAN, image-level knowledge-aware, and patch-level knowledge-aware variants. Code is publicly released.

Significance. The empirical study is broad and generally well executed: six datasets, multiple anomaly-ratio and anomaly-type settings, standard deviations over three runs in the supplement, and ablations that isolate the components of the loss. The idea of exploiting coarse image-level anomaly knowledge to improve fine-grained localization is practically appealing, and the reported gains on medical datasets are consistent. The paper also ships a theoretical analysis and public code, which are strengths. However, the theoretical guarantee is substantially weaker than the text claims: the proofs establish the existence of one equilibrium with exact distribution matching, not that the training objective enforces this alignment, and the key disjoint-support assumption is unverified. The industrial experiments rely on elastically distorted normal images as synthetic anomalies, so they do not directly validate the use of real coarse labels in the motivating industrial scenario. If the claims are revised to match the actual guarantees and the assumptions are empirically checked, the contribution would still be valuable.

major comments (4)
  1. [Supplementary, proof of Theorem 1, Eq. (27); proof of Theorem 2, Eq. (38)] The proofs show that LG(D*,G) is minimized whenever the generated distribution Pg places all its mass on Supp(P+), not only when Pg=P+. Under disjoint supports, any such Pg yields the same value of the total-variation term in Eq. (23)/(34), so the lower bound in Eq. (27)/(38) is attained by a continuum of generators. Thus the losses (8)-(9) and (11)-(12) do not uniquely enforce the claimed distribution alignment; alternating optimization could converge to an equilibrium in which reconstructed features are normal-supported but do not match P+. Since the reconstruction loss L+rec only constrains outputs for normal inputs and is not part of the theorem, the claimed mechanism for localizing anomalies is not a consequence of the adversarial objective. The proof also assumes, rather than demonstrates, the existence of a generator achieving Pg=P+ ("Bring G*_theta such that Pg=P+"), and the theorems establish only existence of an equilibrium, not convergence of the alternating Adam procedure used in Eq. (13). The theorem statements and the surrounding text should be weakened accordingly.
  2. [Theorems 1 and 2 (main text and Supplementary)] Both theorems assume that P+ and P- are disjoint distributions. This is a strong assumption for fine-grained anomalies in pretrained ResNet feature space: a subtle defect may produce patch features that overlap substantially with normal patches. The paper never measures this separability. If the supports overlap, the total-variation lower-bound argument no longer forces the generator to place mass on the normal support, and the theoretical rationale for localization collapses. A concrete check would be to train a linear classifier on normal versus ground-truth anomalous patch features (e.g., on MVTec or on a medical dataset with pixel masks) and report the resulting separation. Until such evidence is provided, the theorems should be stated as conditional on an empirically unverified premise.
  3. [Supplementary, 'Datasets' and Figure 4; Table 2] For MVTec AD and VisA, the coarse anomaly knowledge is obtained by elastically distorting normal images, and no real coarse-labeled defects are used. The results in Table 2 therefore demonstrate the method's behavior under synthetic anomalies, not its ability to exploit real coarse image-level labels in the industrial setting. This matters because elastic distortion creates a known, localized deviation from the normal manifold that may be far easier for the adversarial discriminator to exploit than real industrial defects. The paper should add experiments with a small number of real labeled defect images or explicitly state this limitation in the main text and rephrase the industrial claims accordingly.
  4. [Training and Testing, Eq. (11)-(13); Supplementary, proof of Theorem 2] Theorem 2 is proved for a per-position discriminator D^{(h,w)} and generator G^{(h,w)}, but the actual algorithm uses one shared patch discriminator D_s for all positions and for all layers s. The paper notes this only in passing ("In experiments, a common patch feature discriminator D(p) is used for all patch feature generators"), without an argument that the equilibrium result transfers to the shared case. Since the shared discriminator is what the experiments use, the theoretical claim does not cover the evaluated system as stated.
minor comments (4)
  1. [Paragraph after Theorem 2] "path-level anomaly annotations" should be "patch-level anomaly annotations."
  2. [Paragraph after Theorem 2] The statement that the E_{P+}D term in (11) "has greater weight" is not made precise; after expansion, the coefficients on E_{P+}D, E_{Pg}D, and E_{P-}D are 1, gamma, and (1-gamma), and the comparison with the E_{Pm} term depends on the unknown beta. A short formal justification of the claimed override would be helpful.
  3. [Table 1] The paper reports "best F1" but does not describe the threshold selection procedure; please clarify whether the threshold is chosen on the test set or with a validation protocol, as this affects comparability across methods.
  4. [Section 4.1, Eq. (5) and implementation details] The mixture ratio alpha in Eq. (5) is not used in any of the training losses; the text says alpha and gamma are both set to 0.5, but only gamma appears in Eq. (8)-(12). Clarify the role of alpha or remove it from the formulation.

Circularity Check

1 steps flagged · score 2.0 of 10

Central empirical claims are held-out and non-circular, but the Abstract's 'rigorously prove alignment' overclaims: the Theorem 2 proof shows the generator loss is minimized by any Pg supported on the normal patch-feature support, and the equality Pg = P+ is inserted by hand rather than derived.

  1. self definitional [Abstract (main text); Supplementary Material, 'Theoretical Analysis', Proof of Theorem 2, Eqs. (35)-(39)]
    "Although no patch-level anomalous information is available, we rigorously prove that by simply viewing any patch features from anomalous images as anomalies, the proposed knowledge-aware method can also align the distribution of reconstructed patch features with the normal ones. ... Bring G∗ θ such that Pg = P+ into LG(D∗, Gθ), we can get LG(D∗, G∗ θ ) = γa(1 − η) / 2η. So we conclude there exists a Nash equilibrium system (D∗, G∗ ) such that Pg(p) = P+(p)."

    The proof's lower bound LG(D*,G) ≥ γa(1−η)/(2η) is attained by any Pg with all mass in Supp(P+) and ηPg ≤ P+ pointwise: the inequalities used, ∫(P+−ηPg) ≥ 0 and ∫_{P−}(ηPg+(1−η)P−) = 1−η, do not constrain the shape of Pg inside the normal support. Thus losses (11)-(12) force only support containment of the reconstructed patch distribution, not the full equality Pg = P+. The proof then inserts the strongest claim by hand ('Bring G∗θ such that Pg = P+'), and the Abstract converts this equilibrium-selection assumption into 'we rigorously prove ... align the distribution of reconstructed patch features with the normal ones'.

full rationale

The paper's empirical core is not circular. The test-time anomaly score (Eq. 14) is the raw cosine reconstruction error between input and reconstructed pretrained features, which is independent of the training objectives (Eqs. 8-9 and 11-12) and of the discriminator energy; hyperparameters (α = γ = 0.5, λ = 0.02, learning rates) are hand-set constants rather than fits to the test set; and every reported AUC/F1/PRO is computed on held-out test splits (ISIC2018, OCT, Chest X-ray, Br35H test sets; official MVTec/Visa test sets with pixel-level masks) against externally published baselines. No 'prediction' reduces to a fitted value. The authors' prior work (AA-BiGAN (Tian, Su, Yin 2022), purified GAN (Tian, Su, Yu 2023), Lv et al. 2024) is cited only as related work and baselines; both theorems are proven inside the Supplementary rather than imported by self-citation, so there is no load-bearing self-citation or imported uniqueness claim. The one flagged step is the theoretical derivation chain: the Abstract claims a rigorous proof of patch-level distributional alignment, but the proof's own equations establish only that the generator's output is concentrated on the normal patch-feature support, with Pg = P+ inserted as one particular minimizer. That is a real overclaim in the narrative 'aligns the distribution', though the weaker support-containment statement would already suffice for the localization mechanism. In addition, the load-bearing disjoint-support assumption for fine-grained anomaly features from pre-trained ResNet is never measured, and the industrial experiments substitute elastically distorted normal images for real anomalies; these are unverified premises and correctness risks, not input-output equivalences. Net: the central empirical claim has independent, held-out content and is honestly evaluated, so the circularity score stays low despite the theoretical overclaim.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on distributional assumptions about feature space and on the standard but unproven convergence of GAN training. The empirical setting adds a synthetic-anomaly assumption for industrial data. No new physical entities are introduced.

free parameters (4)
  • gamma = 0.5
    Balance coefficient between generated and anomalous feature penalties in the energy discriminator losses (8) and (11). No derivation or sensitivity analysis is reported.
  • lambda = 0.02
    Weight of the adversarial generator loss in the total update (13). Chosen by hand, with no validation procedure reported.
  • a = not reported
    Energy threshold in the [0, a-D]+ penalty in (8) and (11). It bounds the discriminator output and is not justified or tuned in the text.
  • alpha = 0.5
    Mixture ratio in the input feature distribution P(F)=alpha P+(F)+(1-alpha)P-(F) in Eq. (5). Set to 0.5 in implementation, though it does not directly appear in the implemented loss.
assumptions (5)
  • domain assumption Normal and anomalous feature distributions P+(F) and P-(F) are disjoint.
    Required by Theorems 1 and 2. If the supports overlap in pretrained ResNet feature space, the proof partition into P+ and its complement does not hold as stated.
  • domain assumption Alternating optimization of the discriminator and generator reaches the Nash equilibrium described by the theorems.
    The theorems prove existence of an equilibrium, not convergence. The practical training uses the same loss forms, but the proofs do not establish that gradient-based training attains the equilibrium.
  • domain assumption Pretrained ResNet features retain enough normal-versus-anomaly information for the energy discriminator to separate them.
    The whole method operates on frozen ResNet features. If these features do not distinguish anomalies, the discriminator cannot assign high energy to anomalous features as required.
  • domain assumption An anomalous image is a mixture of normal and anomalous patches with unknown ratio beta, and the normal patch component follows the same distribution as patches from normal images.
    Used in Theorem 2 and in the patch-level training loss. If the mixture assumption fails, the proof's P'_m construction does not apply.
  • ad hoc to paper Elastic distortion of normal images produces anomalies that are representative enough to train the method for industrial datasets.
    Since MVTec AD and Visa provide no training anomaly images, the authors synthesize anomalies via elastic transformation. This is an extra assumption not present in the medical setting.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Boosting Fine-Grained Visual Anomaly Detection with Coarse-Knowledge-Aware Adversarial Learning." pith.science (2026). https://pith.science/paper/VTYXMZAW

@misc{pith2026241212850,
  author       = {Pith},
  title        = {Pith review of: Boosting Fine-Grained Visual Anomaly Detection with Coarse-Knowledge-Aware Adversarial Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VTYXMZAW}},
  note         = {Machine review of arXiv:2412.12850}
}
read the original abstract

Many unsupervised visual anomaly detection methods train an auto-encoder to reconstruct normal samples and then leverage the reconstruction error map to detect and localize the anomalies. However, due to the powerful modeling and generalization ability of neural networks, some anomalies can also be well reconstructed, resulting in unsatisfactory detection and localization accuracy. In this paper, a small coarsely-labeled anomaly dataset is first collected. Then, a coarse-knowledge-aware adversarial learning method is developed to align the distribution of reconstructed features with that of normal features. The alignment can effectively suppress the auto-encoder's reconstruction ability on anomalies and thus improve the detection accuracy. Considering that anomalies often only occupy very small areas in anomalous images, a patch-level adversarial learning strategy is further developed. Although no patch-level anomalous information is available, we rigorously prove that by simply viewing any patch features from anomalous images as anomalies, the proposed knowledge-aware method can also align the distribution of reconstructed patch features with the normal ones. Experimental results on four medical datasets and two industrial datasets demonstrate the effectiveness of our method in improving the detection and localization performance.

Figures

Figures reproduced from arXiv: 2412.12850 by the authors.

Figure 1
Figure 1. Framework of our proposed CKAAD. With the normal and anomalous feature maps extracted from pre-trained [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Histograms of performance (Average AUC and F1 [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Visualization of detection results. Images, anomaly [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Visualization of elastic transformation of images. [PITH_FULL_IMAGE:figures/full_fig_p013_4.png]
Figure 5
Figure 5. Figure 5: The two rows separately exhibit anomaly score distribution on the test set trained without/with [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]
Figure 6
Figure 6. Figure 6: Detection visualization on medical datasets. In [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. CXR-AD: Component X-ray Image Dataset for Industrial Anomaly Detection

    cs.CV 2025-05 conditional novelty 4.0 of 10

    CXR-AD is a new X-ray benchmark for internal component defects, with 653 normal and 561 defect images across five component categories; three standard anomaly detection methods all show substantial performance drops c...

Reference graph

Works this paper leans on

53 extracted references · 40 canonical work pages · cited by 1 Pith paper

  1. [1]

    Akcay, S.; Atapour-Abarghouei, A.; and Breckon, T. P. 2019. Ganomaly: Semi-supervised anomaly detection via adversarial training. In 14th Asian Conference on Computer Vision, 622--637

  2. [2]

    Bergmann, P.; Fauser, M.; Sattlegger, D.; and Steger, C. 2019. MVTec AD--A comprehensive real-world dataset for unsupervised anomaly detection. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 9592--9600

  3. [3]

    Cho, H.; Seol, J.; and Lee, S.-g. 2021. Masked Contrastive Learning for Anomaly Detection. In Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, 1434--1441

  4. [4]

    E.; Dusza, S.; Gutman, D.; Helba, B.; Kalloo, A.; Liopyris, K.; Marchetti, M.; et al

    Codella, N.; Rotemberg, V.; Tschandl, P.; Celebi, M. E.; Dusza, S.; Gutman, D.; Helba, B.; Kalloo, A.; Liopyris, K.; Marchetti, M.; et al. 2019. Skin lesion analysis toward melanoma detection 2018: A challenge hosted by the international skin imaging collaboration (isic). arXiv:1902.03368

  5. [5]

    Defard, T.; Setkov, A.; Loesch, A.; and Audigier, R. 2021. Padim: a patch distribution modeling framework for anomaly detection and localization. In International Conference on Pattern Recognition, 475--489

  6. [6]

    Deng, H.; and Li, X. 2022. Anomaly detection via reverse distillation from one-class embedding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 9737--9746

  7. [7]

    Ding, C.; Pang, G.; and Shen, C. 2022. Catching both gray and black swans: Open-set supervised anomaly detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 7388--7398

  8. [8]

    Fernando, T.; Gammulle, H.; Denman, S.; Sridharan, S.; and Fookes, C. 2021. Deep learning for medical anomaly detection--a survey. ACM Computing Surveys (CSUR), 54(7): 1--37

Show all 53 references
  1. [9]

    Goodfellow, I.; Pouget-Abadie, J.; Mirza, M.; Xu, B.; Warde-Farley, D.; Ozair, S.; Courville, A.; and Bengio, Y. 2014. Generative adversarial nets. Advances in neural information processing systems, 27

  2. [10]

    Gui, X.; Wu, D.; Chang, Y.; and Fan, S. 2022. Constrained Adaptive Projection with Pretrained Features for Anomaly Detection. In Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence, 2059--2065

  3. [11]

    Guo, J.; Lu, S.; Jia, L.; Zhang, W.; and Li, H. 2023. ReContrast: Domain-Specific Anomaly Detection via Contrastive Reconstruction. In Advances in Neural Information Processing Systems, 10721--10740

  4. [12]

    Hamada . 2020. Br35h: Brain tumor detection 2020. https://www.kaggle.com/datasets/ahmedhamada0/brain-tumor-detection, 2020

  5. [13]

    He, K.; Zhang, X.; Ren, S.; and Sun, J. 2016. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, 770--778

  6. [14]

    S.; Goldbaum, M.; Cai, W.; Valentim, C

    Kermany, D. S.; Goldbaum, M.; Cai, W.; Valentim, C. C.; Liang, H.; Baxter, S. L.; McKeown, A.; Yang, G.; Wu, X.; Yan, F.; et al. 2018. Identifying medical diagnoses and treatable diseases by image-based deep learning. cell, 172(5): 1122--1131

  7. [15]

    Kimura, D.; Chaudhury, S.; Narita, M.; Munawar, A.; and Tachibana, R. 2020. Adversarial discriminative attention for robust anomaly detection. In Proceedings of the IEEE/CVF winter conference on applications of computer vision, 2172--2181

  8. [16]

    Li, C.-L.; Sohn, K.; Yoon, J.; and Pfister, T. 2021. Cutpaste: Self-supervised learning for anomaly detection and localization. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 9664--9674

  9. [17]

    Liu, J.; Xie, G.; Wang, J.; Li, S.; Wang, C.; Zheng, F.; and Jin, Y. 2024. Deep industrial image anomaly detection: A survey. Machine Intelligence Research, 21(1): 104--135

  10. [18]

    Liu, W.; Chang, H.; Ma, B.; Shan, S.; and Chen, X. 2023 a . Diversity-measurable anomaly detection. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 12147--12156

  11. [19]

    Liu, Z.; Zhou, Y.; Xu, Y.; and Wang, Z. 2023 b . Simplenet: A simple network for image anomaly detection and localization. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 20402--20411

  12. [20]

    Lv, W.; Su, Q.; Wan, H.; Xu, H.; and Xu, W. 2024. Contamination-Resilient Anomaly Detection via Adversarial Learning on Partially-Observed Normal and Anomalous Data. In Forty-first International Conference on Machine Learning

  13. [21]

    Pang, G.; Ding, C.; Shen, C.; and Hengel, A. v. d. 2021 a . Explainable deep few-shot anomaly detection with deviation networks. arXiv:2108.00462

  14. [22]

    Pang, G.; Shen, C.; Jin, H.; and van den Hengel, A. 2023. Deep weakly-supervised anomaly detection. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 1795--1807

  15. [23]

    Pang, G.; van den Hengel, A.; Shen, C.; and Cao, L. 2021 b . Toward deep supervised anomaly detection: Reinforcement learning from partially labeled anomaly data. In Proceedings of the 27th ACM SIGKDD conference on knowledge discovery & data mining, 1298--1308

  16. [24]

    Reiss, T.; Cohen, N.; Bergman, L.; and Hoshen, Y. 2021. Panda: Adapting pretrained features for anomaly detection and segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2806--2814

  17. [25]

    Reiss, T.; and Hoshen, Y. 2023. Mean-shifted contrastive loss for anomaly detection. In Proceedings of the AAAI Conference on Artificial Intelligence, 2155--2162

  18. [26]

    Roth, K.; Pemula, L.; Zepeda, J.; Sch \"o lkopf, B.; Brox, T.; and Gehler, P. 2022 a . Towards total recall in industrial anomaly detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 14318--14328

  19. [27]

    Roth, K.; Pemula, L.; Zepeda, J.; Sch \"o lkopf, B.; Brox, T.; and Gehler, P. 2022 b . Towards total recall in industrial anomaly detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 14318--14328

  20. [28]

    Rudolph, M.; Wandt, B.; and Rosenhahn, B. 2021. Same same but differnet: Semi-supervised defect detection with normalizing flows. In Proceedings of the IEEE/CVF winter conference on applications of computer vision, 1907--1916

  21. [29]

    A.; Binder, A.; M \"u ller, E.; and Kloft, M

    Ruff, L.; Vandermeulen, R.; Goernitz, N.; Deecke, L.; Siddiqui, S. A.; Binder, A.; M \"u ller, E.; and Kloft, M. 2018. Deep one-class classification. In International conference on machine learning, 4393--4402

  22. [30]

    o rnitz, N.; Binder, A.; M \

    Ruff, L.; Vandermeulen, R. A.; G \"o rnitz, N.; Binder, A.; M \"u ller, E.; M \"u ller, K.-R.; and Kloft, M. 2019. Deep Semi-Supervised Anomaly Detection. In International Conference on Learning Representations

  23. [31]

    H.; and Rabiee, H

    Salehi, M.; Sadjadi, N.; Baselizadeh, S.; Rohban, M. H.; and Rabiee, H. R. 2021. Multiresolution knowledge distillation for anomaly detection. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 14902--14912

  24. [32]

    M.; Schmidt-Erfurth, U.; and Langs, G

    Schlegl, T.; Seeb \"o ck, P.; Waldstein, S. M.; Schmidt-Erfurth, U.; and Langs, G. 2017. Unsupervised anomaly detection with generative adversarial networks to guide marker discovery. In International conference on information processing in medical imaging, 146--157

  25. [33]

    Tack, J.; Mo, S.; Jeong, J.; and Shin, J. 2020. CSI : novelty detection via contrastive learning on distributionally shifted instances. In Proceedings of the 34th International Conference on Neural Information Processing Systems, 11839--11852

  26. [34]

    Tebbe, J.; and Tayyub, J. 2024. Dynamic Addition of Noise in a Diffusion Model for Anomaly Detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 3940--3949

  27. [35]

    Tian, B.; Su, Q.; and Yin, J. 2022. Anomaly Detection by Leveraging Incomplete Anomalous Knowledge with Anomaly-Aware Bidirectional GANs. In Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence, 2255--2261

  28. [36]

    Tian, B.; Su, Q.; and Yu, J. 2023. Leveraging contaminated datasets to learn clean-data distribution with purified generative adversarial networks. In Proceedings of the AAAI Conference on Artificial Intelligence, 9989--9996

  29. [37]

    D.; Nguyen, A

    Tien, T. D.; Nguyen, A. T.; Tran, N. H.; Huy, T. D.; Duong, S.; Nguyen, C. D. T.; and Truong, S. Q. 2023. Revisiting reverse distillation for anomaly detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 24511--24520

  30. [38]

    Tschandl, P.; Rosendahl, C.; and Kittler, H. 2018. The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions. Scientific data, 5(1): 1--9

  31. [39]

    K.; and Cremers, D

    Vasilev, A.; Golkov, V.; Meissner, M.; Lipp, I.; Sgarlata, E.; Tomassini, V.; Jones, D. K.; and Cremers, D. 2020. q-Space novelty detection with variational autoencoders. In Computational Diffusion MRI: MICCAI Workshop, Shenzhen, China, October 2019, 113--124

  32. [40]

    Xia, X.; Pan, X.; Li, N.; He, X.; Ma, L.; Zhang, X.; and Ding, N. 2022. GAN-based anomaly detection: A review. Neurocomputing, 493: 497--535

  33. [41]

    Yang, J.; Xu, R.; Qi, Z.; and Shi, Y. 2022. Visual anomaly detection for images: A systematic survey. Procedia computer science, 199: 471--478

  34. [42]

    Yi, J.; and Yoon, S. 2020. Patch svdd: Patch-level svdd for anomaly detection and segmentation. In Proceedings of the Asian conference on computer vision, 375--390

  35. [43]

    You, Z.; Cui, L.; Shen, Y.; Yang, K.; Lu, X.; Zheng, Y.; and Le, X. 2022. A unified model for multi-class anomaly detection. Advances in Neural Information Processing Systems, 35: 4571--4584

  36. [44]

    Yu, J.; Zheng, Y.; Wang, X.; Li, W.; Wu, Y.; Zhao, R.; and Wu, L. 2021. Fastflow: Unsupervised anomaly detection and localization via 2d normalizing flows. arXiv:2111.07677

  37. [45]

    Zavrtanik, V.; Kristan, M.; and Sko c aj, D. 2021 a . Draem-a discriminatively trained reconstruction embedding for surface anomaly detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 8330--8339

  38. [46]

    Zavrtanik, V.; Kristan, M.; and Sko c aj, D. 2021 b . Reconstruction by inpainting for visual anomaly detection. Pattern Recognition, 112: 107706

  39. [47]

    Zhang, J.; Wang, S.; and Chen, S. 2022. Reconstruction Enhanced Multi-View Contrastive Learning for Anomaly Detection on Attributed Networks. In Proceedings of the Thirty-First International Joint Conference on Artificial Intelligence, IJCAI 2022, Vienna, Austria, 23-29 July 2...

  40. [48]

    Zhang, X.; Li, S.; Li, X.; Huang, P.; Shan, J.; and Chen, T. 2023. DeSTSeg: Segmentation Guided Denoising Student-Teacher for Anomaly Detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 3914--3923

  41. [49]

    Zhao, Y.; Ding, Q.; and Zhang, X. 2023. AE - FLOW : Autoencoders with Normalizing Flows for Medical Images Anomaly Detection. In The Eleventh International Conference on Learning Representations

  42. [50]

    Zhou, Q.; Pang, G.; Tian, Y.; He, S.; and Chen, J. 2024. Anomaly CLIP : Object-agnostic Prompt Learning for Zero-shot Anomaly Detection. In The Twelfth International Conference on Learning Representations

  43. [51]

    Zou, Y.; Jeong, J.; Pemula, L.; Zhang, D.; and Dabeer, O. 2022. Spot-the-difference self-supervised pre-training for anomaly detection and segmentation. In European Conference on Computer Vision, 392--408

  44. [52]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all...

  45. [53]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.