pith. sign in

arxiv: 2607.02252 · v1 · pith:NUMGXKOWnew · submitted 2026-07-02 · 💻 cs.CV

ArcAD: Anomaly-Rectified Calibration for Cold-Start Supervised Anomaly Detection

Ningning Han , Lei Fan , Jia Guo , Yunkang Cao , Xiu Su , Feng Cao , Donglin Di , Tonghua Su This is my paper

Pith reviewed 2026-07-03 15:47 UTC · model grok-4.3

classification 💻 cs.CV
keywords anomaly detectioncold-start learningindustrial inspectionhypersphere projectionpush-pull trainingsupervised anomaly detectionreconstruction-based methodsdata scarcity
0
0 comments X

The pith

ArcAD calibrates reconstruction-based detectors to build compact normal boundaries from scarce samples and rare anomalies.

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

The paper presents ArcAD as a plug-and-play calibration method for industrial anomaly detection in cold-start regimes where normal examples are few and anomalies are rare. It projects the available normal samples onto a hypersphere and pulls them into multiple compact clusters to cover the normal manifold more fully. At the same time the method synthesizes pseudo-anomalies on the same hypersphere and uses the real anomalies to push the decision boundary inward, sharpening the separation between normal and defective regions. This push-pull process lets existing reconstruction models exploit the limited supervised signal without requiring large datasets. If the approach holds, it would allow reliable anomaly detection on manufacturing lines that cannot collect extensive normal or anomaly examples.

Core claim

ArcAD follows a push-pull learning paradigm to construct a compact and discriminative normal boundary under data scarcity. On the one hand it projects limited normal samples onto a hypersphere and pulls them into multiple compact clusters to maximize coverage of the normal manifold. On the other hand it synthesizes pseudo-anomalies on the hypersphere and leverages real anomalies to push the boundary inward and sharpen anomaly discrimination.

What carries the argument

The push-pull learning paradigm on a hypersphere that pulls limited normal samples into compact clusters while pushing with synthesized and real anomalies.

If this is right

  • ArcAD outperforms both state-of-the-art supervised and unsupervised methods on MVTec-AD, VisA, Real-IAD, and MANTA under single-class and multi-class cold-start conditions.
  • The framework works as a plug-and-play addition to any reconstruction-based industrial anomaly detection baseline.
  • Pulling normals into multiple clusters on the hypersphere increases coverage of the normal manifold when samples are scarce.
  • Using both synthesized pseudo-anomalies and real anomalies sharpens the boundary between normal and defective regions.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The hypersphere projection step could be replaced by other compact manifolds if the geometry of the normal data favors a different shape.
  • Performance gains may depend on how well the pseudo-anomaly synthesis matches the distribution of real defects in a given factory.
  • The same calibration idea might transfer to other reconstruction tasks that suffer from limited normal data, such as medical image screening.

Load-bearing premise

That projecting limited normal samples onto a hypersphere and applying push-pull forces with synthesized anomalies will reliably produce a compact and discriminative normal boundary despite data scarcity.

What would settle it

Cold-start experiments on MVTec-AD or VisA in which ArcAD produces no measurable improvement in detection metrics over the unmodified reconstruction baseline.

Figures

Figures reproduced from arXiv: 2607.02252 by Donglin Di, Feng Cao, Jia Guo, Lei Fan, Ningning Han, Tonghua Su, Xiu Su, Yunkang Cao.

Figure 1
Figure 1. Figure 1: Left: Decision boundaries in cold-start scenarios. Unsupervised methods (a) yield loose manifolds and supervised methods (b) overfit to known anomalies, whereas ArcAD (c) forms compact and unified boundaries. Right: Paradigm comparison on data efficiency, anomaly utilization, and generalization. frequently encounters a severe cold-start bottleneck. Unlike standard unsuper￾vised settings [12, 41] that assum… view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the proposed ArcAD. (a) Integration of the ArcAD framework into a standard reconstruction-based architecture. (b) ArcAD explicitly calibrates the latent feature distribution through Sinkhorn-based Prototype Modeling (SPM) and Defect-Guided Calibration (DGC). 3.2 Sinkhorn-based Prototype Modeling (SPM) vMF Modeling. Compared to the unbounded Euclidean space, the hypersphere provides a compact ma… view at source ↗
Figure 3
Figure 3. Figure 3: Statistics of standard and cold-start setting datasets (see details in Supp). In our cold-start setting, 30% of the standard training set is utilized as nor￾mal training data, which is then combined with a subset of anomalies sampled from the original test set. Typically, anomalies account for 10% of this newly constructed training set. Due to the severe scarcity of anomalous samples, we set this ratio to … view at source ↗
Figure 4
Figure 4. Figure 4: Performance of baseline methods (Dinomaly, ReContrast, and RD4AD) equipped with the ArcAD across four datasets under the multi-class setting. Single-class Performance. We evaluated ArcAD under the single-class setting across four datasets, as summarized in [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: Left: Qualitative results of anomaly localization for multi-class anomaly de￾tection. Test images with anomalous regions are highlighted in red contours. Right: Feature visualization of ArcAD before and after optimization on the MANTA dataset. egy yields hard anomalies, en￾abling accurate boundary delineation. 98.50 99.00 99.50 100 300 500 800 1000 Number of Prototypes (K) 91.00 91.50 92.00 92.50 93.00 93.… view at source ↗
Figure 5
Figure 5. Figure 5: Ablation of the number of proto￾types K in the SPM. Number of Prototypes K. To evalu￾ate the impact of the prototype quan￾tity in the SPM module, we con￾ducted an ablation study by varying the number of prototypes K across {100, 300, 500, 800, 1000}. The model reaches its highest I-AUROC (92.5%) at K = 500, with P-AUROC stay￾ing relatively stable above 99.0%. Al￾though the highest P-F1-max (50.2%) occurs a… view at source ↗
read the original abstract

The deployment of Industrial Anomaly Detection (IAD) in real-world manufacturing frequently encounters a challenging cold-start bottleneck, in which limited normal samples fail to represent the full normal distribution and only a few anomalies are available. Under such a regime, existing methods struggle to form compact normal boundaries and fail to effectively exploit supervised signals from rare defects. To address this challenge, we propose Anomaly-Rectified Cold-start AD (ArcAD), a plug-and-play calibration framework for reconstruction-based IAD baselines. ArcAD follows a push-pull learning paradigm to construct a compact and discriminative normal boundary under data scarcity. On the one hand, ArcAD projects limited normal samples onto a hypersphere and pulls them into multiple compact clusters to maximize coverage of the normal manifold. On the other hand, it synthesizes pseudo-anomalies on the hypersphere and leverages real anomalies to push the boundary inward and sharpen anomaly discrimination. Extensive experiments on MVTec-AD, VisA, Real-IAD, and MANTA demonstrate that ArcAD significantly outperforms state-of-the-art supervised and unsupervised methods in both single-class and multi-class settings under cold-start conditions. Code is available at: https://github.com/LGC-AD/ArcAD.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript proposes ArcAD, a plug-and-play calibration framework for reconstruction-based industrial anomaly detection (IAD) baselines under cold-start conditions with scarce normal samples and few anomalies. It adopts a push-pull paradigm: normal samples are projected onto a hypersphere and pulled into multiple compact clusters to maximize coverage of the normal manifold, while pseudo-anomalies are synthesized on the hypersphere and combined with real anomalies to push the decision boundary inward and improve discrimination. Experiments on MVTec-AD, VisA, Real-IAD, and MANTA report that ArcAD outperforms state-of-the-art supervised and unsupervised methods in both single-class and multi-class cold-start settings. Code is released at https://github.com/LGC-AD/ArcAD.

Significance. If the empirical gains hold under rigorous validation, ArcAD would address a practically important bottleneck in deploying IAD systems in manufacturing, where data scarcity is common. The plug-and-play design on existing reconstruction baselines and the public code release are strengths that support reproducibility and adoption.

major comments (3)
  1. [§3] §3 (push-pull paradigm): The central claim that projecting limited normal samples onto a hypersphere and pulling them into multiple compact clusters maximizes coverage of the normal manifold lacks any explicit verification mechanism or metric (e.g., no coverage ratio, manifold reconstruction error, or ablation on sample representativeness). When the scarce samples lie in a small region of feature space, the resulting tight clusters may still leave large portions of the true normal distribution uncovered, undermining generalization; this is load-bearing for the cold-start superiority claim.
  2. [§4] §4 (experiments): The reported outperformance is asserted for specific cold-start regimes, but the manuscript provides no details on how the limited normal samples are selected (random vs. fixed seeds), whether results are averaged over multiple draws, or variance across runs. Without this, it is impossible to assess whether the gains are robust or sensitive to the particular scarce-sample realizations.
  3. [§3.2] §3.2 (pseudo-anomaly synthesis): The procedure for synthesizing pseudo-anomalies on the hypersphere is described at a high level but lacks the precise generation rule, distance constraints, or hyperparameter values used; because this directly implements the 'push' component that sharpens discrimination, the absence of these details prevents assessment of whether the boundary refinement is general or tuned to the evaluated datasets.
minor comments (2)
  1. The abstract and introduction use 'significantly outperforms' without accompanying statistical significance tests or effect-size reporting in the experimental section; adding these would strengthen the presentation.
  2. Notation for the hypersphere radius and cluster centers is introduced without an explicit equation reference in the method overview; a single consolidated equation block would improve clarity.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive comments on our manuscript. We address each major point below and will revise the paper accordingly where the concerns are valid.

read point-by-point responses

  1. Referee: [§3] §3 (push-pull paradigm): The central claim that projecting limited normal samples onto a hypersphere and pulling them into multiple compact clusters maximizes coverage of the normal manifold lacks any explicit verification mechanism or metric (e.g., no coverage ratio, manifold reconstruction error, or ablation on sample representativeness). When the scarce samples lie in a small region of feature space, the resulting tight clusters may still leave large portions of the true normal distribution uncovered, undermining generalization; this is load-bearing for the cold-start superiority claim.

    Authors: We agree that the manuscript does not provide an explicit metric (such as coverage ratio or manifold reconstruction error) to directly verify maximization of normal manifold coverage. The multi-cluster design on the hypersphere is intended to improve representation of limited samples, but downstream task performance is the primary evidence presented. To address this, we will add a targeted analysis or ablation in the revision to quantify coverage under the proposed clustering. revision: yes

  2. Referee: [§4] §4 (experiments): The reported outperformance is asserted for specific cold-start regimes, but the manuscript provides no details on how the limited normal samples are selected (random vs. fixed seeds), whether results are averaged over multiple draws, or variance across runs. Without this, it is impossible to assess whether the gains are robust or sensitive to the particular scarce-sample realizations.

    Authors: The referee is correct that the experimental section lacks these implementation details. We will revise §4 to specify the sample selection procedure (random sampling with fixed seeds), confirm that results are averaged over multiple independent draws, and report mean performance with standard deviations to demonstrate robustness. revision: yes

  3. Referee: [§3.2] §3.2 (pseudo-anomaly synthesis): The procedure for synthesizing pseudo-anomalies on the hypersphere is described at a high level but lacks the precise generation rule, distance constraints, or hyperparameter values used; because this directly implements the 'push' component that sharpens discrimination, the absence of these details prevents assessment of whether the boundary refinement is general or tuned to the evaluated datasets.

    Authors: We acknowledge that §3.2 provides only a high-level description. In the revised manuscript we will expand this section with the exact generation rule, including distance constraints on the hypersphere and the specific hyperparameter values used across experiments. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper describes ArcAD as a plug-and-play calibration framework using hypersphere projection, cluster pulling for normals, and pseudo-anomaly synthesis plus real anomalies for pushing. No equations, uniqueness theorems, or predictions are presented that reduce by construction to fitted inputs or self-citation chains. Claims rest on empirical outperformance on MVTec-AD, VisA, Real-IAD, and MANTA rather than tautological mappings. The method is self-contained against external benchmarks with no load-bearing self-referential steps.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Only the abstract is available; no specific free parameters, axioms, or invented entities can be identified from the provided information.

pith-pipeline@v0.9.1-grok · 5765 in / 1114 out tokens · 34699 ms · 2026-07-03T15:47:19.646743+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

56 extracted references · 56 canonical work pages

  1. [1]

    In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition

    Bergmann, P., Fauser, M., Sattlegger, D., Steger, C.: Mvtec ad–a comprehen- sive real-world dataset for unsupervised anomaly detection. In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition. pp. 9592– 9600 (2019)

    work page 2019

  2. [2]

    Ad- vances in neural information processing systems26(2013)

    Cuturi, M.: Sinkhorn distances: Lightspeed computation of optimal transport. Ad- vances in neural information processing systems26(2013)

    work page 2013

  3. [3]

    In: 34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018

    Davidson, T.R., Falorsi, L., et al.: Hyperspherical variational auto-encoders. In: 34th Conference on Uncertainty in Artificial Intelligence 2018, UAI 2018. pp. 856–

    work page 2018

  4. [4]

    Association For Uncertainty in Artificial Intelligence (AUAI) (2018)

    work page 2018

  5. [5]

    In: International conference on pattern recognition

    Defard,T.,Setkov,A.,Loesch,A.,Audigier,R.:Padim:apatchdistributionmodel- ing framework for anomaly detection and localization. In: International conference on pattern recognition. pp. 475–489. Springer (2021)

    work page 2021

  6. [6]

    In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition

    Deng, H., Li, X.: Anomaly detection via reverse distillation from one-class em- bedding. In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition. pp. 9737–9746 (2022)

    work page 2022

  7. [7]

    In: Proceedings of the IEEE/CVF conference on Com- puter Vision and Pattern Recognition

    Ding, C., Pang, G., Shen, C.: Catching both gray and black swans: Open-set super- vised anomaly detection. In: Proceedings of the IEEE/CVF conference on Com- puter Vision and Pattern Recognition. pp. 7388–7398 (2022) 16 Han et al

    work page 2022

  8. [8]

    IEEE Trans- actions on Industrial Informatics21(5), 3627–3635 (2025)

    Fan, L., Fan, D., Ding, Y., Wu, Y., Di, D., Pagnucco, M., Song, Y.: Grainbrain: Multiview identification and stratification of defective grain kernels. IEEE Trans- actions on Industrial Informatics21(5), 3627–3635 (2025)

    work page 2025

  9. [9]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Fan, L., Fan, D., Hu, Z., Ding, Y., Di, D., Yi, K., Pagnucco, M., Song, Y.: Manta: A large-scale multi-view and visual-text anomaly detection dataset for tiny objects. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 25518–25527 (2025)

    work page 2025

  10. [10]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Fan, L., Huang, J., Di, D., Su, A., Song, T., Pagnucco, M., Song, Y.: Salvaging the overlooked: Leveraging class-aware contrastive learning for multi-class anomaly de- tection. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 21419–21428 (2025)

    work page 2025

  11. [11]

    In: Proceedings of the AAAI Conference on Artificial Intelligence

    Fu, L., Huang, S., Lai, Y., Liao, T., Zhang, C., Chen, C.: Beyond federated proto- typelearning:Learnablesemanticanchorswithhypersphericalcontrastfordomain- skewed data. In: Proceedings of the AAAI Conference on Artificial Intelligence. vol. 39, pp. 16648–16656 (2025)

    work page 2025

  12. [12]

    In: International Confer- ence on Machine Learning

    Gopal, S., Yang, Y.: Von mises-fisher clustering models. In: International Confer- ence on Machine Learning. pp. 154–162. PMLR (2014)

    work page 2014

  13. [13]

    arXiv preprint arXiv:2510.17611 (2025)

    Guo,J.,Lu,S.,Fan,L.,Li,Z.,Di,D.,Song,Y.,Zhang,W.,Zhu,W.,Yan,H.,Chen, F., et al.: One dinomaly2 detect them all: A unified framework for full-spectrum unsupervised anomaly detection. arXiv preprint arXiv:2510.17611 (2025)

    work page arXiv 2025

  14. [14]

    Advances in Neural Information Processing Systems36, 10721–10740 (2023)

    Guo, J., Lu, S., Jia, L., Zhang, W., Li, H.: Recontrast: Domain-specific anomaly de- tection via contrastive reconstruction. Advances in Neural Information Processing Systems36, 10721–10740 (2023)

    work page 2023

  15. [15]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Guo, J., Lu, S., Zhang, W., Chen, F., Li, H., Liao, H.: Dinomaly: The less is more philosophy in multi-class unsupervised anomaly detection. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 20405–20415 (2025)

    work page 2025

  16. [16]

    Advances in Neural Information Processing Systems37, 71162– 71187 (2024)

    He, H., Bai, Y., Zhang, J., He, Q., Chen, H., Gan, Z., Wang, C., Li, X., Tian, G., Xie, L.: Mambaad: Exploring state space models for multi-class unsupervised anomaly detection. Advances in Neural Information Processing Systems37, 71162– 71187 (2024)

    work page 2024

  17. [17]

    In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition

    He, K., Fan, H., Wu, Y., Xie, S., Girshick, R.: Momentum contrast for unsupervised visual representation learning. In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition. pp. 9729–9738 (2020)

    work page 2020

  18. [18]

    In: Proceedings of the IEEE/CVF International Conference on Com- puter Vision

    Hwang, J.J., Yu, S.X., et al.: Segsort: Segmentation by discriminative sorting of segments. In: Proceedings of the IEEE/CVF International Conference on Com- puter Vision. pp. 7334–7344 (2019)

    work page 2019

  19. [19]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Jeong, J., Zou, Y., Kim, T., Zhang, D., Ravichandran, A., Dabeer, O.: Winclip: Zero-/few-shot anomaly classification and segmentation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 19606– 19616 (2023)

    work page 2023

  20. [20]

    arXiv preprint arXiv:2509.24335 (2025)

    Ke, G., Xue, H.: Hyperspherical latents improve continuous-token autoregressive generation. arXiv preprint arXiv:2509.24335 (2025)

    work page arXiv 2025

  21. [21]

    In: Computer Vision and Pattern Recognition

    Kobayashi, T.: T-vmf similarity for regularizing intra-class feature distribution. In: Computer Vision and Pattern Recognition. pp. 6616–6625 (2021)

    work page 2021

  22. [22]

    In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition

    Li, C.L., Sohn, K., Yoon, J., Pfister, T.: Cutpaste: Self-supervised learning for anomaly detection and localization. In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition. pp. 9664–9674 (2021)

    work page 2021

  23. [23]

    In: The Thirteenth International Conference on Learning Representations ArcAD 17

    Li, C., Shi, Y., Hu, J., Zhu, X.X., Mou, L.: Scale-aware contrastive reverse distilla- tion for unsupervised medical anomaly detection. In: The Thirteenth International Conference on Learning Representations ArcAD 17

    work page

  24. [24]

    In: European Conference on Computer Vision

    Li, J.L., Chang, M.C., Chen, W.C.: Learning multi-manifold embedding for out-of- distribution detection. In: European Conference on Computer Vision. pp. 417–433. Springer (2024)

    work page 2024

  25. [25]

    In: Proceedings of the IEEE International Conference on Computer Vision

    Lin, T.Y., Goyal, P., Girshick, R., He, K., Dollár, P.: Focal loss for dense object de- tection. In: Proceedings of the IEEE International Conference on Computer Vision. pp. 2980–2988 (2017)

    work page 2017

  26. [26]

    arXiv preprint arXiv:2511.20169 (2025)

    Ling, H., Guo, J., Tao, Z., Cao, Y., Di, D., Xu, H., Su, X., Song, Y., Fan, L.: Adnet: A large-scale and extensible multi-domain benchmark for anomaly detection across 380 real-world categories. arXiv preprint arXiv:2511.20169 (2025)

    work page arXiv 2025

  27. [27]

    In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition

    Liu, Z., Zhou, Y., Xu, Y., Wang, Z.: Simplenet: A simple network for image anomaly detection and localization. In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition. pp. 20402–20411 (2023)

    work page 2023

  28. [28]

    In: The Thirteenth International Conference on Learning Representations

    Loshchilov, I., Hsieh, C.P., Sun, S., Ginsburg, B.: ngpt: Normalized transformer with representation learning on the hypersphere. In: The Thirteenth International Conference on Learning Representations

    work page

  29. [29]

    Learning with mixture of prototypes for out-of-distribution detection.arXiv preprint arXiv:2402.02653, 2024

    Lu, H., Gong, D., et al.: Learning with mixture of prototypes for out-of-distribution detection. arXiv preprint arXiv:2402.02653 (2024)

    work page arXiv 2024

  30. [30]

    John Wiley & Sons (2009)

    Mardia, K.V., Jupp, P.E.: Directional statistics. John Wiley & Sons (2009)

    work page 2009

  31. [31]

    In: Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision

    McIntosh, D., Albu, A.B.: Inter-realization channels: Unsupervised anomaly de- tection beyond one-class classification. In: Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision. pp. 6285–6295 (2023)

    work page 2023

  32. [32]

    Ming, Y., Sun, Y., et al.: How to exploit hyperspherical embeddings for out-of- distribution detection? arXiv preprint arXiv:2203.04450 (2022)

    work page arXiv 2022

  33. [33]

    IEEE Signal processing magazine13(6), 47–60 (1996)

    Moon, T.K.: The expectation-maximization algorithm. IEEE Signal processing magazine13(6), 47–60 (1996)

    work page 1996

  34. [34]

    Transactions on Machine Learning Research Journal (2024)

    Oquab, M., Darcet, T., Moutakanni, T., Vo, H., Szafraniec, M., Khalidov, V., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., et al.: Dinov2: Learning robust visual features without supervision. Transactions on Machine Learning Research Journal (2024)

    work page 2024

  35. [35]

    arXiv preprint arXiv:2108.00462 (2021)

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

    work page arXiv 2021

  36. [36]

    In: Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining

    Pang, G., Shen, C., Van Den Hengel, A.: Deep anomaly detection with deviation networks. In: Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining. pp. 353–362 (2019)

    work page 2019

  37. [37]

    In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition

    Park, H., Noh, J., Ham, B.: Learning memory-guided normality for anomaly de- tection. In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition. pp. 14372–14381 (2020)

    work page 2020

  38. [38]

    Journal of Machine Learning Research17(200), 1–51 (2016)

    Qin, X., Cunningham, P., et al.: Online trans-dimensional von mises-fisher mix- ture models for user profiles. Journal of Machine Learning Research17(200), 1–51 (2016)

    work page 2016

  39. [39]

    IEEE Trans- actions on Instrumentation and Measurement70, 1–13 (2021)

    Rippel, O., Mertens, P., König, E., Merhof, D.: Gaussian anomaly detection by modeling the distribution of normal data in pretrained deep features. IEEE Trans- actions on Instrumentation and Measurement70, 1–13 (2021)

    work page 2021

  40. [40]

    In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition

    Roth, K., Pemula, L., Zepeda, J., Schölkopf, B., Brox, T., Gehler, P.: Towards total recall in industrial anomaly detection. In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition. pp. 14318–14328 (2022)

    work page 2022

  41. [41]

    In: Proceedings of the IEEE/CVF winter conference on applications of computer vision

    Rudolph, M., Wandt, B., Rosenhahn, B.: Same same but differnet: Semi-supervised defect detection with normalizing flows. In: Proceedings of the IEEE/CVF winter conference on applications of computer vision. pp. 1907–1916 (2021) 18 Han et al

    work page 1907

  42. [42]

    Proceedings of the IEEE109(5), 756–795 (2021)

    Ruff, L., Kauffmann, J.R., Vandermeulen, R.A., Montavon, G., Samek, W., Kloft, M., Dietterich, T.G., Müller, K.R.: A unifying review of deep and shallow anomaly detection. Proceedings of the IEEE109(5), 756–795 (2021)

    work page 2021

  43. [43]

    Journal of Intelligent Manufacturing31(3), 759–776 (2020)

    Tabernik, D., Šela, S., Skvarč, J., Skočaj, D.: Segmentation-based deep-learning approach for surface-defect detection. Journal of Intelligent Manufacturing31(3), 759–776 (2020)

    work page 2020

  44. [44]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Wang, C., Zhu, W., Gao, B.B., Gan, Z., Zhang, J., Gu, Z., Qian, S., Chen, M., Ma, L.: Real-iad: A real-world multi-view dataset for benchmarking versatile industrial anomaly detection. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 22883–22892 (2024)

    work page 2024

  45. [45]

    In: Proceed- ings of the IEEE/CVF conference on Computer Vision and Pattern Recognition

    Wang, F., Zhang, T., Wang, Y., Qiu, Y., Liu, X., Guo, X., Cui, Z.: Distribution pro- totype diffusion learning for open-set supervised anomaly detection. In: Proceed- ings of the IEEE/CVF conference on Computer Vision and Pattern Recognition. pp. 20416–20426 (2025)

    work page 2025

  46. [46]

    In: International conference on ma- chine learning

    Wang, T., Isola, P.: Understanding contrastive representation learning through alignment and uniformity on the hypersphere. In: International conference on ma- chine learning. pp. 9929–9939. PMLR (2020)

    work page 2020

  47. [47]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Wei, S., Jiang, J., Xu, X.: Uninet: A contrastive learning-guided unified framework with feature selection for anomaly detection. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 9994–10003 (2025)

    work page 2025

  48. [48]

    In: Computer Vision and Pattern Recognition

    Xu,J.,Li,S.,Deng,A.,etal.:Probabilisticknowledgedistillationoffaceensembles. In: Computer Vision and Pattern Recognition. pp. 3489–3498 (2023)

    work page 2023

  49. [49]

    In: European Conference on Computer Vision

    Yao, X., Li, R., Qian, Z., Wang, L., Zhang, C.: Hierarchical gaussian mixture normalizing flow modeling for unified anomaly detection. In: European Conference on Computer Vision. pp. 92–108. Springer (2024)

    work page 2024

  50. [50]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Yao, X., Li, R., Zhang, J., Sun, J., Zhang, C.: Explicit boundary guided semi- push-pull contrastive learning for supervised anomaly detection. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 24490–24499 (2023)

    work page 2023

  51. [51]

    Advances in Neural Information Processing Systems 35, 4571–4584 (2022)

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

    work page 2022

  52. [52]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Zavrtanik, V., Kristan, M., Skočaj, D.: Draem-a discriminatively trained re- construction embedding for surface anomaly detection. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 8330–8339 (2021)

    work page 2021

  53. [53]

    arXiv preprint arXiv:2312.07495 (2023)

    Zhang, J., Chen, X., Wang, Y., Wang, C., Liu, Y., Li, X., Yang, M.H., Tao, D.: Exploring plain vit reconstruction for multi-class unsupervised anomaly detection. arXiv preprint arXiv:2312.07495 (2023)

    work page arXiv 2023

  54. [54]

    In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition

    Zhang, X., Li, S., Li, X., Huang, P., Shan, J., Chen, T.: Destseg: Segmentation guided denoising student-teacher for anomaly detection. In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition. pp. 3914– 3923 (2023)

    work page 2023

  55. [55]

    In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition

    Zhu, J., Ding, C., Tian, Y., Pang, G.: Anomaly heterogeneity learning for open-set supervised anomaly detection. In: Proceedings of the IEEE/CVF conference on Computer Vision and Pattern Recognition. pp. 17616–17626 (2024)

    work page 2024

  56. [56]

    In: European Conference on Computer Vision

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

    work page 2022