pith. sign in

arxiv: 2606.24375 · v1 · pith:7RY3HWBKnew · submitted 2026-06-23 · 💻 cs.CV

MATCH: Flow Matching for Multi-View Anomaly Detection

Mathis Kruse , Melissa Schween , Bodo Rosenhahn This is my paper

Pith reviewed 2026-06-26 00:30 UTC · model grok-4.3

classification 💻 cs.CV
keywords multi-view anomaly detectionflow matchinglikelihood estimationindustrial inspectionreal-time anomaly detectionpixel-level segmentationobject anomaly detection
0
0 comments X

The pith

Flow matching uses ODE likelihood estimates to detect and segment anomalies in multi-view industrial images at object, image, and pixel levels.

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

The paper introduces MATCH as the first multi-view anomaly detection method built on flow matching. It leverages the ODE formulation of flow matching to compute likelihoods that serve as anomaly scores across object, image, and pixel levels. The approach reaches state-of-the-art results on the Real-IAD and MANTA-Tiny datasets while running on consumer hardware. By skipping the divergence term, it supports real-time use in production settings. A sympathetic reader would care because multi-view analysis is required for complex manufactured objects, and practical efficiency matters for deployment.

Core claim

MATCH adapts flow matching to multi-view anomaly detection by using its ODE formulation to estimate likelihoods and derive anomaly scores at object, image, and pixel levels. The architectural flexibility of the models allows efficient mapping of features with varying spatial sizes to a normal distribution. This yields state-of-the-art detection and segmentation performance on Real-IAD and MANTA-Tiny while omitting the divergence term to enable real-time operation on consumer hardware.

What carries the argument

The ODE formulation of flow matching, which transforms multi-view features into a normal distribution to enable direct likelihood estimation for anomaly scoring.

If this is right

  • Anomaly scores become available at object, image, and pixel levels from a single likelihood computation.
  • Real-time production use becomes feasible on standard consumer hardware.
  • The same model handles features of different spatial sizes without extra architectural changes.
  • Comprehensive benchmarks establish superiority over prior anomaly detection methods on both Real-IAD and MANTA-Tiny.

Where Pith is reading between the lines

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

  • Omitting the divergence term may trade some theoretical properties of continuous normalizing flows for practical speed in time-critical settings.
  • The multi-view likelihood approach could extend to other generative tasks that require consistent scoring across viewpoints.
  • If the likelihoods remain stable across datasets, the method reduces dependence on reconstruction error or prototype memory for anomaly detection.

Load-bearing premise

The ODE formulation of flow matching directly produces reliable likelihood-based anomaly scores for multi-view data at object, image, and pixel levels without post-hoc adjustments.

What would settle it

A head-to-head evaluation on Real-IAD showing that MATCH's AUROC or AUPRO scores fall below those of prior methods such as reconstruction or memory-bank baselines would falsify the performance claim.

Figures

Figures reproduced from arXiv: 2606.24375 by Bodo Rosenhahn, Mathis Kruse, Melissa Schween.

Figure 1
Figure 1. Figure 1: Qualitative comparison of Multi-Flow [43], Reverse Distillation (RD4AD) [17], and MATCH (ours) on the white tablet and short button category of MANTA-Tiny. While Multi-Flow struggles with its foreground extraction and RD4AD detects some false positives, MATCH most accurately segments all anomalies. Currently, a large variety of well-performing AD models use Normalizing Flows for estimating the likelihood o… view at source ↗
Figure 2
Figure 2. Figure 2: Sampling the "Two Moons" data set, with Flow Matching and RealNVP. Sim￾ilar to AD problems, where small deviances need to be detected in a high dimensional feature space, we embed the two-dimensional moons in d-dimensional space. While the expressiveness of Flow Matching handles high dimensions well, the less flexible Real￾NVP architecture starts to struggle early on. Experimental details are in the append… view at source ↗
Figure 3
Figure 3. Figure 3: Overview of MATCH. A frozen feature encoder Ei is used for latent embeddings. With these, intermediate latent vectors x i t are sampled, and, together with time t and view v information, fed through a bottleneck, and into the decoder layers Di. The Flow Matching loss L OT CFM is applied and lets our model learn the trajectory from data space q to Gaussian space p. Running the model with an ODE from t = 1 t… view at source ↗
Figure 4
Figure 4. Figure 4: A selection of qualitative anomaly maps for classes audiojack, PCB, and regu￾lator of Real-IAD (top row). The bottom row shows goji berries, soybeans, and wafer resistors of MANTA-Tiny. Pictured are the original RGB image, ground truth segmen￾tation maps and the resulting anomaly maps overlayed on RGB. category of MANTA-Tiny, as well as an analysis of some failure cases in the supplement in Sec. F, togethe… view at source ↗
Figure 5
Figure 5. Figure 5: Varying the feature dimension for bot￾tleneck and decoder from 32 to 768. All of the metrics linearly rise with higher dimensions. Including the divergence does not lead to any noticeable gains for the detection AUROC. The same behavior applies to all met￾rics, as completely shown in the supplementary materials. Most interestingly, with M = 1, the di￾vergence only makes up a total of .57% of the final scor… view at source ↗
Figure 1
Figure 1. Figure 1: Embedding Network, with posi￾tional encoding (PosEnc) and embedding layers (Emb). Bottleneck. The full bottleneck, in￾cluding feature alignment and bot￾tleneck blocks, is shown in [PITH_FULL_IMAGE:figures/full_fig_p023_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Alignment and Bottleneck Network [PITH_FULL_IMAGE:figures/full_fig_p024_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: ResNet-style Decoder Block, depending on choice of input (in), hidden (hid) and output (out) dimension. B Additional Multi-Class Anomaly Detection Additional evaluation is done with the popular multi-class AD setting, despite MATCH not being explicitly designed for this task. Here, all training splits of the classes of Real-IAD are combined. After training for 100 epochs, all classes are evaluated independ… view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative results for the MVTec AD data set [6] on the classes bottle, cable, and hazel￾nut. MATCH consistently segments the anomalous regions. We also test the new DINOv3 [60] model as a back￾bone for MATCH as well as baselines RD4AD [17] and PaDiM [16]. We chose these baselines, since they most naturally support swapping to a differ￾ent custom backbone. Since we still restrict our￾selves to consumer-le… view at source ↗
Figure 5
Figure 5. Figure 5: Anomaly maps of the most common failure cases in MANTA-Tiny [PITH_FULL_IMAGE:figures/full_fig_p029_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Qualitative anomaly maps for every category of MANTA-Tiny, with class and anomaly type designated. Despite the vast variety of products, MATCH regularly segments the anomalies, while avoiding excessive false positives [PITH_FULL_IMAGE:figures/full_fig_p030_6.png] view at source ↗
read the original abstract

Detecting anomalies in industrial objects is an important topic for increasing production efficiency. More complex objects often require the analysis of several view points, which has led to the field of multi-view anomaly detection. We present MATCH, the first multi-view anomaly detection method based on Flow Matching (FM). With the ODE formulation of Flow Matching, we can estimate likelihoods and thereby derive an anomaly score to detect anomalies in multi-view image data at object, image, and pixel-level. The architectural flexibility of FM models allows us to efficiently transform features of different spatial sizes to the normal distribution. We evaluate thoroughly on the already established Real-IAD data set and are also the first to provide a comprehensive evaluation of popular anomaly detection methods for the MANTA-Tiny data set. MATCH achieves state-of-the-art performance in both anomaly detection and segmentation, all while running on consumer-level hardware. By omitting the costly divergence term needed for likelihood estimation, we ensure that MATCH is usable in real-time production scenarios. Lastly, several ablation studies are conducted to validate the methodological choices.

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

1 major / 1 minor

Summary. The paper introduces MATCH, the first multi-view anomaly detection method based on Flow Matching. It uses the ODE formulation of FM to estimate likelihoods and derive anomaly scores at object, image, and pixel levels on multi-view industrial image data. The method is evaluated on Real-IAD and MANTA-Tiny, claims SOTA performance in detection and segmentation, runs on consumer hardware, and enables real-time use by omitting the divergence term required for likelihood estimation.

Significance. If the likelihood-based scoring mechanism without the divergence term can be shown to produce reliable anomaly rankings, the approach would offer an efficient, architecture-flexible alternative for multi-view industrial anomaly detection that supports real-time deployment.

major comments (1)
  1. [Abstract] Abstract: the central claim that 'likelihoods' are estimated via the ODE formulation of Flow Matching to produce anomaly scores at multiple levels is load-bearing, yet the text simultaneously states that the divergence term needed for likelihood estimation is omitted. No derivation of the exact anomaly score formula (or proxy) appears, nor any comparison to a standard CNF log-likelihood estimator or validation that ranking quality is preserved on Real-IAD or MANTA-Tiny.
minor comments (1)
  1. [Abstract] The abstract mentions 'several ablation studies' but provides no details on which methodological choices were tested or their outcomes.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive comment on the abstract and the load-bearing claim regarding likelihood estimation. We address this directly below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that 'likelihoods' are estimated via the ODE formulation of Flow Matching to produce anomaly scores at multiple levels is load-bearing, yet the text simultaneously states that the divergence term needed for likelihood estimation is omitted. No derivation of the exact anomaly score formula (or proxy) appears, nor any comparison to a standard CNF log-likelihood estimator or validation that ranking quality is preserved on Real-IAD or MANTA-Tiny.

    Authors: We agree the abstract is imprecise and that the manuscript lacks an explicit derivation or validation. MATCH computes anomaly scores from the flow-matching ODE trajectory (integrating the learned vector field) without the divergence term; the resulting quantity is a computationally efficient proxy rather than exact log-likelihood. We will revise the abstract to describe the scores as derived from the ODE formulation via this proxy. In the revised manuscript we will add (i) the precise mathematical definition of the proxy score in Section 3, (ii) a short comparison of proxy versus full CNF log-likelihood rankings on a held-out subset of Real-IAD, and (iii) a note confirming that the proxy preserves anomaly ordering on the evaluated datasets. These additions will be included in the next version. revision: yes

Circularity Check

0 steps flagged

No circularity: derivation chain self-contained with no reductions to inputs by construction

full rationale

The abstract and provided text contain no equations, no fitted parameters renamed as predictions, and no self-citations invoked as load-bearing uniqueness theorems. The ODE-based likelihood claim and omission of the divergence term are presented as a methodological choice for efficiency, without any self-definitional loop or renaming of known results. No step reduces a claimed prediction to its own input by construction; the central performance claims rest on external evaluation on Real-IAD and MANTA-Tiny rather than internal tautology.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only input supplies no equations, training details, or modeling choices; therefore no free parameters, axioms, or invented entities can be extracted.

pith-pipeline@v0.9.1-grok · 5713 in / 1096 out tokens · 33512 ms · 2026-06-26T00:30:07.389313+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

94 extracted references · 19 canonical work pages

  1. [1]

    In: 2022 IEEE International Conference on Image Processing (ICIP)

    Akcay, S., Ameln, D., Vaidya, A., Lakshmanan, B., Ahuja, N., Genc, U.: Anoma- lib: A deep learning library for anomaly detection. In: 2022 IEEE International Conference on Image Processing (ICIP). pp. 1706–1710. IEEE (2022)

    2022

  2. [2]

    arXiv preprint arXiv:1907.02392 (2019)

    Ardizzone, L., Lüth, C., Kruse, J., Rother, C., Köthe, U.: Guided image genera- tion with conditional invertible neural networks. arXiv preprint arXiv:1907.02392 (2019)

    Pith/arXiv arXiv 1907

  3. [3]

    arXiv preprint arXiv:2504.10612 (2025)

    Balcerak,M.,Amiranashvili,T.,Terpin,A.,Shit,S.,Bogensperger,L.,Kaltenbach, S., Koumoutsakos, P., Menze, B.: Energy matching: Unifying flow matching and energy-based models for generative modeling. arXiv preprint arXiv:2504.10612 (2025)

    arXiv 2025

  4. [4]

    In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)

    Batzner, K., Heckler, L., König, R.: Efficientad: Accurate visual anomaly detection at millisecond-level latencies. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV). pp. 128–138 (January 2024)

    2024

  5. [5]

    International Journal of Computer Vision130(4), 947–969 (2022)

    Bergmann,P.,Batzner,K.,Fauser,M.,Sattlegger,D.,Steger,C.:Beyonddentsand scratches: Logical constraints in unsupervised anomaly detection and localization. International Journal of Computer Vision130(4), 947–969 (2022)

    2022

  6. [6]

    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)

    2019

  7. [7]

    Bergmann,P.,Jin,X.,Sattlegger,D.,Steger,C.:Themvtec3d-addatasetforunsu- pervised 3d anomaly detection and localization. In: Proceedings of the 17th Inter- national Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications, VISIGRAPP 2022, Volume 5: VISAPP, Online Stream- ing, February 6-8, 2022. pp. 202–213. SCITEPRESS (2022)

    2022

  8. [8]

    Bergmann, P., Löwe, S., Fauser, M., Sattlegger, D., Steger, C.: Improving unsu- pervised defect segmentation by applying structural similarity to autoencoders. In: Proceedings of the 14th International Joint Conference on Computer Vision, Imag- ing and Computer Graphics Theory and Applications, VISIGRAPP 2019, Volume 5: VISAPP, Prague, Czech Republic, Feb...

    2019

  9. [9]

    Long, Y .-C

    Bhunia, A., Li, C., Bilen, H.: Looking 3d: Anomaly detection with 2d-3d align- ment. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2024, Seattle, WA, USA, June 16-22, 2024. pp. 17263–17272. IEEE (2024). https://doi.org/10.1109/CVPR52733.2024.01634,https://doi.org/10.1109/ CVPR52733.2024.01634

    work page doi:10.1109/cvpr52733.2024.01634 2024

  10. [10]

    Live: Learning video llm with stream- ing speech transcription at scale

    Bhunia, A., Li, C., Bilen, H.: Odd-one-out: Anomaly detection by compar- ing with neighbors. In: IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, CVPR 2025, Nashville, TN, USA, June 11-15, 2025. pp. 16 M. Kruse et al. 20395–20404. Computer Vision Foundation / IEEE (2025).https://doi.org/ 10.1109/CVPR52734.2025.01899,https://openaccess.the...

    work page doi:10.1109/cvpr52734.2025.01899 2025

  11. [11]

    In: Proceedings of the 16th Asian Conference on Computer Vision (ACCV2022 (2022), aCCV

    Bonfiglioli, L., Toschi, M., Silvestri, D., Fioraio, N., De Gregorio, D.: The eyecan- dies dataset for unsupervised multimodal anomaly detection and localization. In: Proceedings of the 16th Asian Conference on Computer Vision (ACCV2022 (2022), aCCV

    2022

  12. [12]

    arXiv preprint arXiv:2505.10996 (2025)

    Cao, Y., Cheng, Y., Xu, X., Zhang, Y., Sun, Y., Tan, Y., Zhang, Y., Huang, X., Shen, W.: Visual anomaly detection under complex view-illumination interplay: A large-scale benchmark. arXiv preprint arXiv:2505.10996 (2025)

    arXiv 2025

  13. [13]

    Advances in neural information processing systems31(2018)

    Chen, R.T., Rubanova, Y., Bettencourt, J., Duvenaud, D.K.: Neural ordinary dif- ferential equations. Advances in neural information processing systems31(2018)

    2018

  14. [14]

    arXiv preprint arXiv:2005.02357 (2020)

    Cohen, N., Hoshen, Y.: Sub-image anomaly detection with deep pyramid corre- spondences. arXiv preprint arXiv:2005.02357 (2020)

    arXiv 2005

  15. [15]

    arXiv preprint arXiv:2307.08698 (2023)

    Dao, Q., Phung, H., Nguyen, B., Tran, A.: Flow matching in latent space. arXiv preprint arXiv:2307.08698 (2023)

    arXiv 2023

  16. [16]

    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)

    2021

  17. [17]

    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)

    2022

  18. [18]

    In: 2009 IEEE conference on computer vision and pattern recognition

    Deng, J., Dong, W., Socher, R., Li, L.J., Li, K., Fei-Fei, L.: Imagenet: A large- scale hierarchical image database. In: 2009 IEEE conference on computer vision and pattern recognition. pp. 248–255. Ieee (2009)

    2009

  19. [19]

    Assembly101: A large-scale multi-view video dataset for understanding procedural activities

    Ding, C., Pang, G., Shen, C.: Catching both gray and black swans: Open-set supervised anomaly detection. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022, New Orleans, LA, USA, June 18-24, 2022. pp. 7378–7388. IEEE (2022).https://doi.org/10.1109/CVPR52688.2022.00724, https://doi.org/10.1109/CVPR52688.2022.00724

    work page doi:10.1109/cvpr52688.2022.00724 2022

  20. [20]

    In: Bengio, Y., LeCun, Y

    Dinh, L., Krueger, D., Bengio, Y.: NICE: non-linear independent components esti- mation. In: Bengio, Y., LeCun, Y. (eds.) 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Workshop Track Proceedings (2015),http://arxiv.org/abs/1410.8516

    Pith/arXiv arXiv 2015

  21. [21]

    In: 5th International Conference on Learning Representations, ICLR 2017, Toulon, France, April 24-26, 2017, Conference Track Proceedings

    Dinh, L., Sohl-Dickstein, J., Bengio, S.: Density estimation using real NVP. In: 5th International Conference on Learning Representations, ICLR 2017, Toulon, France, April 24-26, 2017, Conference Track Proceedings. OpenReview.net (2017), https://openreview.net/forum?id=HkpbnH9lx

    2017

  22. [22]

    Live: Learning video llm with stream- ing speech transcription at scale

    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: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2025, Nashville, TN, USA, June 11-15, 2025. pp. 25518–25527. Computer Vision Foundation / IEEE (2025).https://doi.or...

    work page doi:10.1109/cvpr52734.2025.02376 2025

  23. [23]

    Live: Learning video llm with stream- ing speech transcription at scale

    Galappaththige, C.J., Lai, J., Windrim, L., Dansereau, D.G., Sünderhauf, N., Miller, D.: Multi-view pose-agnostic change localization with zero labels. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2025, Nashville, TN, USA, June 11-15, 2025. pp. 11600–11610. Computer Vision Foundation / IEEE (2025).https://doi.org/10.1109/CVPR52...

    work page doi:10.1109/cvpr52734.2025.01083 2025

  24. [24]

    Savva, A

    Gong, D., Liu, L., Le, V., Saha, B., Mansour, M.R., Venkatesh, S., van den Hen- gel, A.: Memorizing normality to detect anomaly: Memory-augmented deep au- toencoder for unsupervised anomaly detection. In: 2019 IEEE/CVF International Conference on Computer Vision, ICCV 2019, Seoul, Korea (South), October 27 - November 2, 2019. pp. 1705–1714. IEEE (2019).ht...

    work page doi:10.1109/iccv 2019

  25. [25]

    In: 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019

    Grathwohl, W., Chen, R.T.Q., Bettencourt, J., Sutskever, I., Duvenaud, D.: FFJORD: free-form continuous dynamics for scalable reversible generative mod- els. In: 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019. OpenReview.net (2019),https://openreview. net/forum?id=rJxgknCcK7

    2019

  26. [26]

    In: Wooldridge, M.J., Dy, J.G., Natarajan, S

    Gu, Z., Zhu, B., Zhu, G., Chen, Y., Tang, M., Wang, J.: Anomalygpt: Detect- ing industrial anomalies using large vision-language models. In: Wooldridge, M.J., Dy, J.G., Natarajan, S. (eds.) Thirty-Eighth AAAI Conference on Artificial Intel- ligence, AAAI 2024. pp. 1932–1940. AAAI Press (2024).https://doi.org/10. 1609/AAAI.V38I3.27963,https://doi.org/10.16...

    work page doi:10.1609/aaai.v38i3.27963 2024

  27. [27]

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

    Gudovskiy,D.,Ishizaka,S.,Kozuka,K.:Cflow-ad:Real-timeunsupervisedanomaly detection with localization via conditional normalizing flows. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. pp. 98–107 (2022)

    2022

  28. [28]

    In: Oh, A., Naumann, T., Glober- son, A., Saenko, K., Hardt, M., Levine, S

    Guo, J., Lu, S., Jia, L., Zhang, W., Li, H.: Recontrast: Domain-specific anomaly detection via contrastive reconstruction. In: Oh, A., Naumann, T., Glober- son, A., Saenko, K., Hardt, M., Levine, S. (eds.) Advances in Neural Infor- mation Processing Systems 36: Annual Conference on Neural Information Pro- cessing Systems 2023, NeurIPS 2023, New Orleans, L...

    2023

  29. [29]

    Live: Learning video llm with stream- ing speech transcription at scale

    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: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2025, Nashville, TN, USA, June 11-15, 2025. pp. 20405–20415. Computer Vision Foundation / IEEE (2025).https://doi.org/10.1109/CVPR52734.2025.01...

    work page doi:10.1109/cvpr52734.2025.01900 2025

  30. [30]

    Springer (1993)

    Hairer, E., Wanner, G., Nørsett, S.P.: Solving ordinary differential equations I: Nonstiff problems. Springer (1993)

    1993

  31. [31]

    In: Globersons, A., Mackey, L., Belgrave, D., Fan, A., Paquet, U., Tomczak, J.M., Zhang, C

    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 unsuper- vised anomaly detection. In: Globersons, A., Mackey, L., Belgrave, D., Fan, A., Paquet, U., Tomczak, J.M., Zhang, C. (eds.) Advances in Neural Infor- mation Processing Systems 38: Annual Conference on N...

    2024

  32. [32]

    In: Wooldridge, M.J., Dy, J.G., Natarajan, S

    He, H., Zhang, J., Chen, H., Chen, X., Li, Z., Chen, X., Wang, Y., Wang, C., Xie, L.: A diffusion-based framework for multi-class anomaly detection. In: Wooldridge, M.J., Dy, J.G., Natarajan, S. (eds.) Thirty-Eighth AAAI Conference on Artificial 18 M. Kruse et al. Intelligence, AAAI 2024, Thirty-Sixth Conference on Innovative Applications of Artificial In...

    work page doi:10.1609/aaai.v38i8.28690 2024

  33. [33]

    IEEE Transactions on Multimedia (2026)

    He, H., Zhang, J., Tian, G., Wang, C., Xie, L.: Learning multi-view anomaly de- tection with efficient adaptive selection. IEEE Transactions on Multimedia (2026)

    2026

  34. [34]

    & Sun, J

    He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: 2016 IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2016, Las Vegas, NV, USA, June 27-30, 2016. pp. 770–778. IEEE Computer Society (2016).https://doi.org/10.1109/CVPR.2016.90,https://doi.org/10.1109/ CVPR.2016.90

    work page doi:10.1109/cvpr.2016.90 2016

  35. [35]

    Uncrtaints: Uncertainty quantification for cloud removal in optical satellite time series,

    Heckler, L., König, R., Bergmann, P.: Exploring the importance of pretrained feature extractors for unsupervised anomaly detection and localization. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2023 - Workshops, Vancouver, BC, Canada, June 17-24, 2023. pp. 2917–2926. IEEE (2023).https://doi.org/10.1109/CVPRW59228.2023.00293,htt...

    work page doi:10.1109/cvprw59228.2023.00293 2023

  36. [36]

    arXiv preprint arXiv:2503.21622 (2025)

    Heckler-Kram, L., Neudeck, J.H., Scheler, U., König, R., Steger, C.: The mvtec ad 2 dataset: Advanced scenarios for unsupervised anomaly detection. arXiv preprint arXiv:2503.21622 (2025)

    arXiv 2025

  37. [37]

    arXiv preprint arXiv:1606.08415 (2016)

    Hendrycks, D., Gimpel, K.: Gaussian error linear units (gelus). arXiv preprint arXiv:1606.08415 (2016)

    Pith/arXiv arXiv 2016

  38. [38]

    Communications in Statistics-Simulation and Com- putation18(3), 1059–1076 (1989)

    Hutchinson, M.F.: A stochastic estimator of the trace of the influence matrix for laplacian smoothing splines. Communications in Statistics-Simulation and Com- putation18(3), 1059–1076 (1989)

    1989

  39. [39]

    Query - dependent video representation for moment retrieval and highlight detection

    Jeong, J., Zou, Y., Kim, T., Zhang, D., Ravichandran, A., Dabeer, O.: Winclip: Zero-/few-shot anomaly classification and segmentation. In: IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition, CVPR 2023, Vancouver, BC, Canada, June 17-24, 2023. pp. 19606–19616. IEEE (2023).https://doi.org/ 10.1109/CVPR52729.2023.01878,https://doi.org/10.1109/C...

    work page doi:10.1109/cvpr52729.2023.01878 2023

  40. [40]

    In: Bengio, Y., LeCun, Y

    Kingma, D.P., Ba, J.: Adam: A method for stochastic optimization. In: Bengio, Y., LeCun, Y. (eds.) 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings (2015),http://arxiv.org/abs/1412.6980

    Pith/arXiv arXiv 2015

  41. [41]

    (eds.) Advances in Neural Information Processing Systems 31: Annual ConferenceonNeuralInformationProcessingSystems2018,NeurIPS2018,Decem- ber 3-8, 2018, Montréal, Canada

    Kingma, D.P., Dhariwal, P.: Glow: Generative flow with invertible 1x1 convolu- tions.In:Bengio,S.,Wallach,H.M.,Larochelle,H.,Grauman,K.,Cesa-Bianchi,N., Garnett, R. (eds.) Advances in Neural Information Processing Systems 31: Annual ConferenceonNeuralInformationProcessingSystems2018,NeurIPS2018,Decem- ber 3-8, 2018, Montréal, Canada. pp. 10236–10245 (2018...

    2018

  42. [42]

    In: Larochelle, H., Ranzato, M., Hadsell, R., Balcan, M., Lin, H

    Kirichenko, P., Izmailov, P., Wilson, A.G.: Why normalizing flows fail to detect out-of-distribution data. In: Larochelle, H., Ranzato, M., Hadsell, R., Balcan, M., Lin, H. (eds.) Advances in Neural Information Processing Systems 33: Annual Con- ference on Neural Information Processing Systems 2020, NeurIPS 2020, December 6-12, 2020, virtual (2020),https:...

    2020

  43. [43]

    In: IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, CVPR Workshops 2025, Nashville, TN, USA, June 11-15, 2025

    Kruse, M., Rosenhahn, B.: Multi-flow: Multi-view-enriched normalizing flows for industrial anomaly detection. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, CVPR Workshops 2025, Nashville, TN, USA, June 11-15, 2025. pp. 3933–3944. Computer Vision Foundation / IEEE (2025)

    2025

  44. [44]

    In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2024 - Workshops, Seattle, WA, USA, June 17- 18, 2024

    Kruse, M., Rudolph, M., Woiwode, D., Rosenhahn, B.: Splatpose & detect: Pose- agnostic 3d anomaly detection. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2024 - Workshops, Seattle, WA, USA, June 17- 18, 2024. pp. 3950–3960. IEEE (2024).https://doi.org/10.1109/CVPRW63382. 2024.00399,https://doi.org/10.1109/CVPRW63382.2024.00399

    work page doi:10.1109/cvprw63382 2024

  45. [45]

    IEEE Access10, 78446–78454 (2022)

    Lee, S., Lee, S., Song, B.C.: CFA: coupled-hypersphere-based feature adaptation for target-oriented anomaly localization. IEEE Access10, 78446–78454 (2022). https://doi.org/10.1109/ACCESS.2022.3193699,https://doi.org/10.1109/ ACCESS.2022.3193699

    work page doi:10.1109/access.2022.3193699 2022

  46. [46]

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

    Lei, J., Hu, X., Wang, Y., Liu, D.: Pyramidflow: High-resolution defect contrastive localization using pyramid normalizing flow. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 14143–14152 (2023)

    2023

  47. [47]

    In: The Eleventh International Conference on Learning Rep- resentations, ICLR 2023, Kigali, Rwanda, May 1-5, 2023

    Lipman, Y., Chen, R.T.Q., Ben-Hamu, H., Nickel, M., Le, M.: Flow matching for generative modeling. In: The Eleventh International Conference on Learning Rep- resentations, ICLR 2023, Kigali, Rwanda, May 1-5, 2023. OpenReview.net (2023), https://openreview.net/forum?id=PqvMRDCJT9t

    2023

  48. [48]

    arXiv preprint arXiv:2412.06264 (2024)

    Lipman, Y., Havasi, M., Holderrieth, P., Shaul, N., Le, M., Karrer, B., Chen, R.T., Lopez-Paz, D., Ben-Hamu, H., Gat, I.: Flow matching guide and code. arXiv preprint arXiv:2412.06264 (2024)

    Pith/arXiv arXiv 2024

  49. [49]

    arXiv preprint arXiv:2503.11088 (2025)

    Liu, Y., Xu, X., Li, S., Liao, J., Yang, X.: Multi-view industrial anomaly detec- tion with epipolar constrained cross-view fusion. arXiv preprint arXiv:2503.11088 (2025)

    arXiv 2025

  50. [50]

    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)

    2023

  51. [51]

    In: 7th Inter- national Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019

    Loshchilov, I., Hutter, F.: Decoupled weight decay regularization. In: 7th Inter- national Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019. OpenReview.net (2019),https://openreview.net/forum? id=Bkg6RiCqY7

    2019

  52. [52]

    Technometrics42(1), 55–61 (2000)

    McKay, M.D., Beckman, R.J., Conover, W.J.: A comparison of three methods for selecting values of input variables in the analysis of output from a computer code. Technometrics42(1), 55–61 (2000)

    2000

  53. [53]

    Journal of compu- tational physics122(2), 218–230 (1995)

    Morokoff, W.J., Caflisch, R.E.: Quasi-monte carlo integration. Journal of compu- tational physics122(2), 218–230 (1995)

    1995

  54. [54]

    The Journal of Machine Learning Research22(1), 2617–2680 (2021)

    Papamakarios, G., Nalisnick, E., Rezende, D.J., Mohamed, S., Lakshminarayanan, B.: Normalizing flows for probabilistic modeling and inference. The Journal of Machine Learning Research22(1), 2617–2680 (2021)

    2021

  55. [55]

    In: Inter- national conference on machine learning

    Rezende, D., Mohamed, S.: Variational inference with normalizing flows. In: Inter- national conference on machine learning. pp. 1530–1538. PMLR (2015)

    2015

  56. [56]

    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)

    2022

  57. [57]

    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) 20 M. Kruse et al

    1907

  58. [58]

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

    Rudolph, M., Wehrbein, T., Rosenhahn, B., Wandt, B.: Fully convolutional cross- scale-flows for image-based defect detection. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. pp. 1088–1097 (2022)

    2022

  59. [59]

    In: WACV

    Rudolph, M., Wehrbein, T., Rosenhahn, B., Wandt, B.: Asymmetric student- teacher networks for industrial anomaly detection. In: WACV. pp. 2591–2601. IEEE (2023)

    2023

  60. [60]

    arXiv preprint arXiv:2508.10104 (2025)

    Siméoni, O., Vo, H.V., Seitzer, M., Baldassarre, F., Oquab, M., Jose, C., Khali- dov, V., Szafraniec, M., Yi, S., Ramamonjisoa, M., et al.: Dinov3. arXiv preprint arXiv:2508.10104 (2025)

    Pith/arXiv arXiv 2025

  61. [61]

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

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

    2023

  62. [62]

    Tong, A., Fatras, K., Malkin, N., Huguet, G., Zhang, Y., Rector-Brooks, J., Wolf, G., Bengio, Y.: Improving and generalizing flow-based generative mod- els with minibatch optimal transport. Trans. Mach. Learn. Res.2024(2024), https://openreview.net/forum?id=CD9Snc73AW

    2024

  63. [63]

    Advances in neural information pro- cessing systems30(2017)

    Vaswani,A.,Shazeer,N.,Parmar,N.,Uszkoreit,J.,Jones,L.,Gomez,A.N.,Kaiser, Ł., Polosukhin, I.: Attention is all you need. Advances in neural information pro- cessing systems30(2017)

    2017

  64. [64]

    Long, Y .-C

    Wang, C., Zhu, W., Gao, 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: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2024, Seattle, WA, USA, June 16-22, 2024. pp. 22883–22892. IEEE (2024).https://doi.org/10.1109/CVPR52...

    work page doi:10.1109/cvpr52733.2024.02159 2024

  65. [65]

    In: British Machine Vision Conference (2021)

    Wang, G., Han, S., Ding, E., Huang, D.: Student-teacher feature pyramid matching for anomaly detection. In: British Machine Vision Conference (2021)

    2021

  66. [66]

    Yang, K., Cao, J., Bai, Z., Su, Z., Tagliasacchi, A.: Piad: Pose and illumination agnosticanomalydetection.In:ProceedingsoftheIEEE/CVFConferenceonCom- puter Vision and Pattern Recognition (CVPR). pp. 4734–4743 (June 2025)

    2025

  67. [67]

    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)

    2022

  68. [68]

    arXiv preprint arXiv:2111.07677 (2021)

    Yu, J., Zheng, Y., Wang, X., Li, W., Wu, Y., Zhao, R., Wu, L.: Fastflow: Unsuper- vised anomaly detection and localization via 2d normalizing flows. arXiv preprint arXiv:2111.07677 (2021)

    arXiv 2021

  69. [69]

    In: Wilson, R.C., Hancock, E.R., Smith, W.A.P

    Zagoruyko, S., Komodakis, N.: Wide residual networks. In: Wilson, R.C., Hancock, E.R., Smith, W.A.P. (eds.) Proceedings of the British Machine Vision Conference 2016, BMVC 2016, York, UK, September 19-22, 2016. BMVA Press (2016),https: //bmva-archive.org.uk/bmvc/2016/papers/paper087/index.html

    2016

  70. [70]

    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)

    2021

  71. [71]

    In: European conference on computer vision

    Zavrtanik, V., Kristan, M., Skočaj, D.: Dsr–a dual subspace re-projection network for surface anomaly detection. In: European conference on computer vision. pp. 539–554. Springer (2022)

    2022

  72. [72]

    Zeiler, Dilip Krishnan, Graham W

    Zeiler, M.D., Krishnan, D., Taylor, G.W., Fergus, R.: Deconvolutional networks. In: The Twenty-Third IEEE Conference on Computer Vision and Pattern Recog- nition, CVPR 2010, San Francisco, CA, USA, 13-18 June 2010. pp. 2528–2535. MATCH: Flow Matching for Multi-View Anomaly Detection 21 IEEE Computer Society (2010).https://doi.org/10.1109/CVPR.2010.5539957...

    work page doi:10.1109/cvpr.2010.5539957 2010

  73. [73]

    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)

    arXiv 2023

  74. [74]

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

    Zhang, J., He, H., Gan, Z., He, Q., Cai, Y., Xue, Z., Wang, Y., Wang, C., Xie, L., Liu, Y.: A comprehensive library for benchmarking multi-class visual anomaly de- tection. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 3524–3533 (2025)

    2025

  75. [75]

    Live: Learning video llm with stream- ing speech transcription at scale

    Zhang, J., Wang, G., Jin, Y., Huang, D.: Towards training-free anomaly detec- tion with vision and language foundation models. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2025, Nashville, TN, USA, June 11-15, 2025. pp. 15204–15213. Computer Vision Foundation / IEEE (2025).https: //doi.org/10.1109/CVPR52734.2025.01416,https://o...

    work page doi:10.1109/cvpr52734.2025.01416 2025

  76. [76]

    Scene Adaptive Sparse Transformer for Event-based Object Detection

    Zhang, X., Xu, M., Zhou, X.: Realnet: A feature selection network with realistic synthetic anomaly for anomaly detection. In: IEEE/CVF Conference on Com- puter Vision and Pattern Recognition, CVPR 2024, Seattle, WA, USA, June 16- 22, 2024. pp. 16699–16708. IEEE (2024).https://doi.org/10.1109/CVPR52733. 2024.01580,https://doi.org/10.1109/CVPR52733.2024.01580

    work page doi:10.1109/cvpr52733 2024

  77. [77]

    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)

    2023

  78. [78]

    In: Proceedings of the 23rd ACM SIGKDD international conference on knowledge discovery and data mining

    Zhou, C., Paffenroth, R.C.: Anomaly detection with robust deep autoencoders. In: Proceedings of the 23rd ACM SIGKDD international conference on knowledge discovery and data mining. pp. 665–674 (2017)

    2017

  79. [79]

    arXiv preprint arXiv:2410.00713 (2024)

    Zhou, K., Cao, Y., Kim, T., Zhao, H., Dong, H., Ting, K.M., Zhu, Y.: Rad: A dataset and benchmark for real-life anomaly detection with robotic observations. arXiv preprint arXiv:2410.00713 (2024)

    Pith/arXiv arXiv 2024

  80. [80]

    In: Oh, A., Naumann, T., Globerson, A., Saenko, K., Hardt, M., Levine, S

    Zhou, Q., Li, W., Jiang, L., Wang, G., Zhou, G., Zhang, S., Zhao, H.: PAD: A dataset and benchmark for pose-agnostic anomaly detection. In: Oh, A., Naumann, T., Globerson, A., Saenko, K., Hardt, M., Levine, S. (eds.) Ad- vances in Neural Information Processing Systems 36: Annual Conference on Neural Information Processing Systems 2023, NeurIPS 2023, New O...

    2023

Showing first 80 references.