Pith. sign in

REVIEW 3 major objections 4 minor 55 references

AMI-Net: Adaptive Mask Inpainting Network for Industrial Anomaly Detection and Localization

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read AMI-Net's core claim is that reconstruction-based anomaly detection should mask predicted anomalies, not random patches, and reconstruct multi-scale features, not pixels; on MVTec AD this reaches 99.0% image and 98.2% pixel AUROC.

desk verdict Solid engineering paper with a genuinely new adaptive test-time mask idea, but the central ablation is missing and the cluster-boundary rule is underspecified. read the letter →

arxiv 2412.11802 v1 pith:ICC6ASVC submitted 2024-12-16 cs.CV cs.AI

Wei Luo , Haiming Yao , Wenyong Yu , Zhengyong Li This is my paper · ORCID
classification cs.CVcs.AI
keywords anomalydetectionlocalizationadaptivemaskinpaintingfeaturereconstructionclustertokensvisiontransformerindustrialdefectMVTecAD
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

AMI-Net is an unsupervised industrial defect detection method built on a simple bet: if a reconstruction network can be made to inpaint the defective region rather than copy it, the defect will stand out as reconstruction error. To do this, the paper replaces pixel targets with multi-scale feature maps from a pre-trained WideResNet50, trains an inpainting transformer under random masks of variable size and position, and adds an adaptive mask generator that identifies which feature tokens are anomalous by their distance to learned cluster tokens. At test time the generator masks those anomalous tokens, the inpainter restores them from visible normal context, and the difference between input and reconstructed features localizes the defect. On MVTec AD the method reports 99.0% image-level and 98.2% pixel-level AUROC in the one-for-one setting, 97.2%/97.5% in the one-for-all setting, and 11.48 ms per image inference. A sympathetic reader would care because the method addresses the core weakness of reconstruction approaches, defect reconstruction, without requiring defect samples or multiple test-time passes.

What carries the argument

The load-bearing mechanism is the adaptive mask generator: a set of P trainable cluster tokens aggregated by a one-block transformer (the semantic aggregation network) over feature tokens that carry fixed positional embeddings. Distance between a cluster token and a feature token is defined as the Euclidean distance times the cosine dissimilarity; each feature token is assigned to its nearest cluster, and a cluster's boundary is $r_i = \mathrm{mean}(d_{ij}) + 0.5\,\mathrm{std}(d_{ij})$. Tokens beyond this boundary are masked, and the inpainting network, a stack of eight transformer blocks, reconstructs the masked tokens from the visible ones. A second load-bearing component is the reconstruction target itself: multi-scale features from layers 2 through 4 of a pre-trained WideResNet50, resized and concatenated, which the paper argues are more semantically separable between normal and abnormal than pixels. Finally, the training-time random positional and quantitative masking, with the mask ratio sampled uniformly from 0 to 1, teaches the inpainter to handle defects of arbitrary size, and the feature jittering variant adds robustness to noise.

What would settle it

Run the released code on MVTec AD, record the adaptive mask for every test image, and compare it with the ground-truth defect mask: if the mean intersection-over-union is near zero on any category, or if defects like cable swaps are left unmasked, the central claim that the generator masks anomalous regions while preserving normal ones is falsified. Sweeping $\lambda$ in Eq. 16 and observing a narrow accuracy peak would further show that the boundary is miscalibrated.

Watch

Extended reading notes

Core claim

The central claim is that the right reconstruction target is not the image but a multi-scale semantic feature representation, and the right mask is not random but adaptive. Given a test image, AMI-Net extracts features with a frozen pre-trained CNN, projects them into tokens, and assigns each token to one of P learned cluster tokens using a distance that multiplies Euclidean distance by cosine dissimilarity. The cluster boundary is set as mean distance plus 0.5 standard deviation; tokens beyond the boundary are treated as anomalous and masked, while normal tokens stay visible. Positional embedding is added before clustering so global defects, such as swapped cables or misplaced transistors, which look locally normal but appear in wrong positions, are caught. The inpainting transformer then fills the masked positions from the visible normal context, so defective regions are suppressed in the reconstruction. The anomaly score is the product of Euclidean distance and cosine dissimilarity between input and reconstructed features, upsampled to the original resolution. The paper argues this design prevents the generalization failure of plain autoencoders, removes the test-time randomness of prior mask-based methods, and needs only a single forward pass.

Load-bearing premise

The method assumes that, on every test image, normal feature tokens form clusters tight enough that a per-cluster boundary of mean distance plus 0.5 standard deviation separates anomalous tokens from normal ones; the paper does not state whether these statistics come from the test image itself or from training data, and small miscalibration would either leave defects unmasked or mask large normal regions.

Editorial extensions

If this is right

  • On MVTec AD, AMI-Net with feature jittering reports 99.0% image-level and 98.2% pixel-level AUROC in the one-for-one setting, trailing PatchCore by only 0.1% in detection while running more than four times faster.
  • On BTAD, it reports 95.1% image and 97.5% pixel AUROC, the best among compared methods, supporting better generalization to texture-dominated datasets.
  • In the one-for-all setting, a single unified model reports 97.2% image and 97.5% pixel AUROC on MVTec AD, surpassing UniAD by 0.7 percentage points in both metrics.
  • In the few-shot setting with 2, 4, or 8 training images per category, AMI-Net achieves the best localization and second-best detection among the compared few-shot methods.
  • The single forward pass at 11.48 ms per image makes the method suitable for real-time industrial inspection without multiple complementary masks.

Reading between the lines

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

  • The adaptive mask generator effectively turns a reconstruction network into a two-stage detector: first a fast clustering-based proposal of anomalous tokens, then an inpainting-based verification; this suggests the mask quality, not the inpainter, is the accuracy bottleneck, so improving cluster boundary calibration should pay off more than enlarging the inpainter.
  • Because the mask is generated per test image from cluster statistics, the method should transfer to new object categories without retraining only if the pre-trained features remain tightly clustered; a natural test is to evaluate the generator's mask precision on categories unseen during training.
  • The paper's sketched extension to abnormal training samples, pasting extracted defect regions onto normal samples, could turn AMI-Net into a semi-supervised method; one could test whether the adaptive mask generator then learns to ignore pasted defects during training and still detects real ones.
  • The distance-as-product-of-Euclidean-and-cosine design couples scale and direction, so normal features with large magnitude but slightly wrong direction may be masked; this could explain the reported false positives on low-contrast noise.
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

3 major / 4 minor

Summary. The paper proposes AMI-Net, an unsupervised industrial anomaly detection and localization method that reconstructs multi-scale features extracted by a pre-trained WideResNet50. During training the model randomly masks variable numbers of feature tokens and learns to inpaint them, while also training cluster tokens with a clustering loss. During testing, an adaptive mask generator clusters feature tokens and masks tokens whose distance to the nearest cluster center exceeds a per-cluster boundary, after which an inpainting transformer reconstructs the masked features; the final anomaly score is the distance between the input and reconstructed features. Experiments on MVTec AD and BTAD are reported under one-for-one, one-for-all, and few-shot settings, with inference time reported as 11.48 ms.

Significance. If the adaptive mask generator works as claimed, AMI-Net would be a practically valuable contribution: it offers competitive image- and pixel-level AUROC on MVTec AD (99.0/98.2), best results on BTAD (95.1/97.5), strong one-for-all performance (97.2/97.5), reasonable few-shot localization, and a fast single-pass inference. The paper also ships code, which is a concrete reproducibility asset, and the experiments span multiple training regimes and include ablations over key hyperparameters. However, the central claim is that the adaptive mask hides anomalous tokens while preserving normal tokens; the manuscript does not provide direct evidence of mask quality, does not isolate the adaptive mask from random masking in a controlled within-model ablation, and contains an inconsistency in the cluster-distance equations that makes the mask generation procedure not precisely reproducible as written.

major comments (3)
  1. [§III-D, Eqs. (13)–(16)] The distance definitions are internally inconsistent and the cluster boundary is underspecified. Eq. (13) defines d_ij as the minimum over i of R(T^i_clu, T^{2j}_F(I)), so d_ij is independent of i; Eq. (14) then defines d_i as the sum over j of d_ij, which makes d_i identical for every cluster i. The intended per-cluster distance set is never written down, and this also undermines the clustering loss in Eq. (15), whose intra-class term becomes vacuous under the written definitions. Eq. (16) additionally uses the notation d_ij without specifying whether the mean and standard deviation are computed per test image, from a training-set buffer, or from a fitted distribution. This is load-bearing because the adaptive mask is the core novelty: if, for example, the mean and std are computed per test image, a heavily defective image will inflate them and push the boundary above anomalous tokens, leaving defects unmasked; if they come from training statistics, distribution shift can mask normal regions. The equations must be corrected and the statistics precisely specified.
  2. [§III-D2 and §IV-E] The claimed benefit of adaptive masking over random masking is not demonstrated by a controlled within-model experiment. The training procedure uses random positional and quantitative masking, while testing uses the adaptive mask, so the reported accuracy could in principle be attributed to the training strategy, the reconstruction objective, or the cluster token branch rather than to the adaptive mask itself. The paper should add an ablation in which the same trained inpainting network is evaluated with (a) the adaptive mask, (b) a random mask of the same average mask ratio, and (c) no mask, reporting image/pixel AUROC and AP. The paper also never reports mask-level precision/recall of the adaptive mask against ground-truth defect maps, which is the most direct way to confirm that the mask actually covers defects and preserves normal regions. The failure cases in §IV-G (missed subtle defects, false positives on background noise) are consistent with a miscalibrated boundary, which makes this missing evidence particularly important.
  3. [§III-D, §IV-E6] The sensitivity of the final anomaly score to the mask generator is not characterized. Because the final score in Eq. (18) is computed after masking exactly the tokens that the adaptive mask generator flagged, the reported detection and localization results partly inherit the mask generator's decisions; this is a design coupling, not necessarily a fatal flaw, but the manuscript presents the adaptive mask as if it were validated independently. The ablation in Fig. 9(f) reports that varying lambda does not affect the pixel AUROC metric at all, while the text says lambda controls whether normal features are erroneously masked. This should be analyzed: if pixel AUROC is insensitive to the cluster boundary, the paper should explain why, and it should report mask-level metrics that respond to lambda. Without such analysis, the role of the adaptive mask generator in the overall performance is not established.
minor comments (4)
  1. [§I, §II-A] There are several typos and inconsistent names: 'establishs' in the Introduction, 'TrsutMAE' in §II-A1 and Table I, and 'Cluser Tokens' in the heading of §IV-E3.
  2. [References] Reference [13] is cited multiple times in section headings and the bibliography entry appears incomplete or misformatted; please verify the MVTec AD reference and all other entries against the publisher's format.
  3. [Table IV] The DiffNet row for k=8 reports '82.3-' with a missing pixel-level AUROC; please fill in the value or mark it clearly as not reported.
  4. [§IV-E6] The text says 'The variation in parameter lambda does not affect the pixel AUROC metric' but Fig. 9(f) appears to show small fluctuations; please reconcile the text with the figure or explain why these fluctuations are within noise.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: AMI-Net's adaptive mask is a heuristic pre-filter, and the reported anomaly score is an independently computed reconstruction residual benchmarked on external datasets.

full rationale

The derivation chain is self-contained. AMI-Net trains on normal samples with L_rec (Eqs. 7-9) and L_clu (Eq. 15) and reports AUROC on held-out MVTec AD and BTAD test sets, so there is no parameter fitted to test labels being renamed as a prediction. The adaptive mask generator (Eqs. 10-16) is a distance-to-cluster-center heuristic that flags candidate anomalous tokens; the final anomaly map As (Eq. 18) is a separate reconstruction residual between input features and inpainted features. Nothing in Eqs. 12-18 makes As equal to the mask by construction: a masked token can still be reconstructed with low residual (false negative), and visible normal tokens can produce high residual (false positive), as the paper's own failure cases in Sec. IV-G show. The self-citations (Refs. [8]-[10], [25]) appear only as related-work examples and are not load-bearing; no uniqueness theorem or ansatz is imported from the authors' prior work. The underspecification of whether mean/std in Eq. 16 are per-test-image or training-buffer statistics, and the index inconsistency between Eqs. 13-14, are correctness/reproducibility risks rather than circularity, because they do not make the reported result equivalent to an input by construction.

Assumptions & free parameters 8 free parameters · 5 assumptions · 0 invented entities

The central claim rests on five tuned hyperparameters, a modeling choice on mask-ratio distribution, an unspecified feature jittering strength, and standard deep-learning assumptions about pretrained feature discriminability and cluster compactness. No new physical entities are introduced; cluster tokens are trainable model parameters rather than independent evidence.

free parameters (8)
  • lambda (cluster boundary scaling) = 0.5
    Tuned via ablation on the MVTec AD test set (Fig. 9f); controls mask aggressiveness and directly affects false positives and false negatives.
  • P (number of cluster tokens) = 8
    Chosen by ablation (Fig. 9c) for a speed-accuracy trade-off.
  • K (patch size) = 4
    Chosen by ablation (Fig. 9b); affects token granularity and inference speed.
  • Ni (inpainting transformer blocks) = 8
    Chosen by ablation (Fig. 9d) balancing restoration capability and speed.
  • Ns (semantic aggregation transformer blocks) = 1
    Chosen by ablation (Fig. 9e); larger Ns degrades detection accuracy.
  • loss weights w1, w2, w3, w4 = 1, 5, 1, 0.1
    Set by design in Sections III-C and III-D; not systematically ablated, and w2=5 favors cosine similarity over MSE.
  • mask ratio distribution = uniform(0,1)
    Introduced in Section III-C as the multi-scale masking strategy; no evidence that it matches the real defect scale distribution.
  • feature jittering strength = not specified
    Borrowed from UniAD [40] for the Ours* variants; the strength is not stated, affecting reproducibility of the best reported numbers.
assumptions (5)
  • domain assumption Pretrained WideResNet50 features separate normal from anomalous industrial patterns.
    Invoked in Section III-B via reference [35]; if the feature space is not discriminative, both cluster distances and reconstruction targets lose meaning.
  • domain assumption Normal training feature tokens form compact clusters in the combined Euclidean-times-cosine distance space.
    Section III-D2; the cluster boundary rule only works if within-cluster distances have a usable mean and standard deviation separation.
  • ad hoc to paper Random masking with ratio sampled uniformly from 0 to 1 during training covers the scale diversity of industrial defects.
    Section III-C; this is a modeling choice not derived from defect size statistics or a stated defect scale prior.
  • domain assumption A transformer with global self-attention can use visible normal context to inpaint masked anomalous regions.
    Section III-D5; the entire reconstruction advantage depends on the inpainting network succeeding with visible context.
  • domain assumption Positional embeddings make global defects detectable by clustering.
    Section III-D4 argues position encoding helps for global defects like transistor misplacement; this assumes cluster centers encode positional normalcy.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AMI-Net: Adaptive Mask Inpainting Network for Industrial Anomaly Detection and Localization." pith.science (2026). https://pith.science/paper/ICC6ASVC

@misc{pith2026241211802,
  author       = {Pith},
  title        = {Pith review of: AMI-Net: Adaptive Mask Inpainting Network for Industrial Anomaly Detection and Localization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ICC6ASVC}},
  note         = {Machine review of arXiv:2412.11802}
}
read the original abstract

Unsupervised visual anomaly detection is crucial for enhancing industrial production quality and efficiency. Among unsupervised methods, reconstruction approaches are popular due to their simplicity and effectiveness. The key aspect of reconstruction methods lies in the restoration of anomalous regions, which current methods have not satisfactorily achieved. To tackle this issue, we introduce a novel \uline{A}daptive \uline{M}ask \uline{I}npainting \uline{Net}work (AMI-Net) from the perspective of adaptive mask-inpainting. In contrast to traditional reconstruction methods that treat non-semantic image pixels as targets, our method uses a pre-trained network to extract multi-scale semantic features as reconstruction targets. Given the multiscale nature of industrial defects, we incorporate a training strategy involving random positional and quantitative masking. Moreover, we propose an innovative adaptive mask generator capable of generating adaptive masks that effectively mask anomalous regions while preserving normal regions. In this manner, the model can leverage the visible normal global contextual information to restore the masked anomalous regions, thereby effectively suppressing the reconstruction of defects. Extensive experimental results on the MVTec AD and BTAD industrial datasets validate the effectiveness of the proposed method. Additionally, AMI-Net exhibits exceptional real-time performance, striking a favorable balance between detection accuracy and speed, rendering it highly suitable for industrial applications. Code is available at: https://github.com/luow23/AMI-Net

Figures

Figures reproduced from arXiv: 2412.11802 by the authors.

Figure 1
Figure 1. Comparison of different unsupervised anomaly detection methods. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The schematic diagrams of different methods. (a) Normal-data-based [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Overall architecture of proposed AMI-Net. Firstly, multi-scale features are extracted using a pretrained CNN. (a) During the training phase, AMI-Net [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Examples of the effectiveness of the adaptive mask generator. First Row: the defective image. Second Row: the distance map formed by the distance [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: Issues arising from clustering methods that do not take positional [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Examples of the role of positional information. First Column: [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Localization Results of AMI-Net on MVTec AD [13] and BTAD [14]. For each set, from top to bottom, there are the defect image, detection heat [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 8
Figure 8. Figure 8: Inference time versus Image level AUROC on MVTec AD dataset [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: The ablation experiment results. (a) Influence of pre-trained network. (b) Impact of patch size [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 11
Figure 11. Figure 11: Some instances of detection failures by AMI-Net. Top Row: [PITH_FULL_IMAGE:figures/full_fig_p013_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 51 canonical work pages

  1. [1]

    Reducing the dimensionality of data with neural networks,

    G. E. Hinton and R. Salakhutdinov, “Reducing the dimensionality of data with neural networks,” Science, 2006. 1, I, II-A1, 10

    work page 2006

  2. [2]

    Reconstruction by inpainting for visual anomaly detection,

    V . Zavrtanik, M. Kristan, and D. Sko ˇcaj, “Reconstruction by inpainting for visual anomaly detection,” Pattern Recognition, vol. 112, p. 107706,

    work page

  3. [3]

    Multi-category decom- position editing network for the accurate visual inspection of texture defects,

    H. Yang, H. Zhu, J. Li, J. Chen, and Z. Yin, “Multi-category decom- position editing network for the accurate visual inspection of texture defects,” IEEE Transactions on Automation Science and Engineering , pp. 1–11, 2023. I

    work page 2023

  4. [4]

    Defect classification and detection using a multitask deep one-class cnn,

    X. Dong, C. J. Taylor, and T. F. Cootes, “Defect classification and detection using a multitask deep one-class cnn,” IEEE Transactions on Automation Science and Engineering , vol. 19, no. 3, pp. 1719–1730,

    work page

  5. [5]

    Visual anomaly detection via partition memory bank module and error estimation,

    P. Xing and Z. Li, “Visual anomaly detection via partition memory bank module and error estimation,” IEEE Transactions on Circuits and Systems for Video Technology, 2023. I

    work page 2023

  6. [6]

    Pga-net: Pyramid feature fusion and global context attention network for automated surface defect detection,

    H. Dong, K. Song, Y . He, J. Xu, Y . Yan, and Q. Meng, “Pga-net: Pyramid feature fusion and global context attention network for automated surface defect detection,” IEEE Transactions on Industrial Informatics , 2020. I

    work page 2020

  7. [7]

    A-net: An a-shape lightweight neural network for real-time surface defect segmentation,

    B. Chen, T. Niu, W. Yu, R. Zhang, Z. Wang, and B. Li, “A-net: An a-shape lightweight neural network for real-time surface defect segmentation,” IEEE Transactions on Instrumentation and Measurement, pp. 1–1, 2023. I

    work page 2023

  8. [8]

    Normal reference attention and defective feature perception network for surface defect detection,

    W. Luo, H. Yao, and W. Yu, “Normal reference attention and defective feature perception network for surface defect detection,” IEEE Transac- tions on Instrumentation and Measurement , vol. 72, pp. 1–14, 2023. I, II-A2

    work page 2023

Show all 55 references
  1. [9]

    A feature memory rearrangement network for visual inspection of textured surface defects toward edge intelligent manufacturing,

    H. Yao, W. Yu, and X. Wang, “A feature memory rearrangement network for visual inspection of textured surface defects toward edge intelligent manufacturing,” IEEE Transactions on Automation Science and Engineering, 2022. I, II-A2

    2022

  2. [10]

    Unsupervised defect segmentation via forgetting-inputting-based feature fusion and multiple hierarchical feature difference,

    W. Luo, T. Niu, H. Yao, L. Tang, W. Yu, and B. Li, “Unsupervised defect segmentation via forgetting-inputting-based feature fusion and multiple hierarchical feature difference,” IEEE Sensors Journal , 2023. I

    2023

  3. [11]

    Self-supervised masking for unsupervised anomaly detection and localization,

    C. Huang, Q. Xu, Y . Wang, Y . Wang, and Y . Zhang, “Self-supervised masking for unsupervised anomaly detection and localization,” IEEE Transactions on Multimedia , pp. 1–1, 2022. I, II-A3

    2022

  4. [12]

    Masked swin transformer unet for industrial anomaly detection,

    J. Jiang, J. Zhu, M. Bilal, Y . Cui, N. Kumar, R. Dou, F. Su, and X. Xu, “Masked swin transformer unet for industrial anomaly detection,” IEEE Transactions on Industrial Informatics , vol. 19, no. 2, pp. 2200–2209,

  5. [13]

    Mvtec ad — a comprehensive real-world dataset for unsupervised anomaly detection,

    P. Bergmann, M. Fauser, D. Sattlegger, and C. Steger, “Mvtec ad — a comprehensive real-world dataset for unsupervised anomaly detection,” computer vision and pattern recognition , 2019. I, III-D4, I, IV, IV-B1, 7, 8, IV-C, III, IV, IV-D

    2019

  6. [14]

    Vt- adl: A vision transformer network for image anomaly detection and localization,

    P. Mishra, R. Verk, D. Fornasier, C. Piciarelli, and G. L. Foresti, “Vt- adl: A vision transformer network for image anomaly detection and localization,” in 2021 IEEE 30th International Symposium on Industrial Electronics (ISIE). IEEE, 2021, pp. 01–06. I, IV, IV-A4, II, IV-B2,...

    2021

  7. [15]

    Memorizing normality to detect anomaly: Memory- augmented deep autoencoder for unsupervised anomaly detection,

    D. Gong, L. Liu, V . Le, B. Saha, M. R. Mansour, S. Venkatesh, and A. van den Hengel, “Memorizing normality to detect anomaly: Memory- augmented deep autoencoder for unsupervised anomaly detection,” in- ternational conference on computer vision , 2019. II-A1, II-A1

    2019

  8. [16]

    Trustmae: A noise-resilient defect classification framework using memory-augmented auto-encoders with trust regions,

    D. S. Tan, Y .-C. Chen, T. P.-C. Chen, and W.-C. Chen, “Trustmae: A noise-resilient defect classification framework using memory-augmented auto-encoders with trust regions,” workshop on applications of computer vision, 2021. II-A1, II-A1, IV-A4, IV-B1

    2021

  9. [17]

    Divide- and-assemble: Learning block-wise memory for unsupervised anomaly detection,

    J. Hou, Y . Zhang, Q. Zhong, D. Xie, S. Pu, and H. Zhou, “Divide- and-assemble: Learning block-wise memory for unsupervised anomaly detection,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2021, pp. 8791–8800. II-A1, II-A1

    2021

  10. [18]

    An unsupervised-learning-based approach for automated defect inspection on textured surfaces,

    S. Mei, H. Yang, and Z. Yin, “An unsupervised-learning-based approach for automated defect inspection on textured surfaces,”IEEE Transactions on Instrumentation and Measurement , 2018. II-A1

    2018

  11. [19]

    Multiscale feature-clustering- based fully convolutional autoencoder for fast accurate visual inspection of texture surface defects,

    H. Yang, Y . Chen, K. Song, and Z. Yin, “Multiscale feature-clustering- based fully convolutional autoencoder for fast accurate visual inspection of texture surface defects,” IEEE Transactions on Automation Science and Engineering, 2019. II-A1, III-D4

    2019

  12. [20]

    The unreasonable effectiveness of deep features as a perceptual metric,

    R. Zhang, P. Isola, A. A. Efros, E. Shechtman, and O. Wang, “The unreasonable effectiveness of deep features as a perceptual metric,” computer vision and pattern recognition , 2018. II-A1

    2018

  13. [21]

    Cutpaste: Self-supervised learning for anomaly detection and localization,

    C.-L. Li, K. Sohn, J. Yoon, and T. Pfister, “Cutpaste: Self-supervised learning for anomaly detection and localization,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 9664–9674. II-A2

    2021

  14. [22]

    An anomaly feature-editing- based adversarial network for texture defect visual inspection,

    H. Yang, Q. Zhou, K. Song, and Z. Yin, “An anomaly feature-editing- based adversarial network for texture defect visual inspection,” IEEE Transactions on Industrial Informatics , 2021. II-A2, III-D4

    2021

  15. [23]

    Draem-a discriminatively trained reconstruction embedding for surface anomaly detection,

    V . Zavrtanik, M. Kristan, and D. Sko ˇcaj, “Draem-a discriminatively trained reconstruction embedding for surface anomaly detection,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 8330–8339. II-A2, IV-A4, IV-B1, IV-C

    2021

  16. [24]

    Mldfr: A multilevel features restoration method based on damaged images for anomaly detection and localization,

    Y . Guo, M. Jiang, Q. Huang, Y . Cheng, and J. Gong, “Mldfr: A multilevel features restoration method based on damaged images for anomaly detection and localization,” IEEE Transactions on Industrial Informatics, pp. 1–10, 2023. II-A2

    2023

  17. [25]

    Siamese transition masked autoen- coders as uniform unsupervised visual anomaly detector,

    H. Yao, X. Wang, and W. Yu, “Siamese transition masked autoen- coders as uniform unsupervised visual anomaly detector,” arXiv preprint arXiv:2211.00349, 2022. II-A3

    2022 arXiv

  18. [26]

    Swin transformer: Hierarchical vision transformer using shifted windows,

    Z. Liu, Y . Lin, Y . Cao, H. Hu, Y . Wei, Z. Zhang, S. Lin, and B. Guo, “Swin transformer: Hierarchical vision transformer using shifted windows,” in Proceedings of the IEEE/CVF international conference on computer vision, 2021, pp. 10 012–10 022. II-A3

    2021

  19. [27]

    Inpainting transformer for anomaly detection,

    J. Pirnay and K. Chai, “Inpainting transformer for anomaly detection,” in Image Analysis and Processing–ICIAP 2022: 21st International Con- ference, Lecce, Italy, May 23–27, 2022, Proceedings, Part II . Springer, 2022, pp. 394–406. II-A3

    2022

  20. [28]

    Deep one-class classification,

    L. Ruff, R. Vandermeulen, N. Goernitz, L. Deecke, S. A. Siddiqui, A. Binder, E. M ¨uller, and M. Kloft, “Deep one-class classification,” in International conference on machine learning . PMLR, 2018, pp. 4393–4402. II-B

    2018

  21. [29]

    Patch svdd: Patch-level svdd for anomaly detection and segmentation,

    J. Yi and S. Yoon, “Patch svdd: Patch-level svdd for anomaly detection and segmentation,” asian conference on computer vision , 2020. II-B, IV-A4, IV-B2, IV-C

    2020

  22. [30]

    Panda: Adapting pretrained features for anomaly detection and segmentation,

    T. Reiss, N. Cohen, L. Bergman, and Y . Hoshen, “Panda: Adapting pretrained features for anomaly detection and segmentation,” computer vision and pattern recognition , 2020. II-B, IV-A4, IV-B2

    2020

  23. [31]

    Towards total recall in industrial anomaly detection,

    K. Roth, L. Pemula, J. Zepeda, B. Sch ¨olkopf, T. Brox, and P. Gehler, “Towards total recall in industrial anomaly detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 14 318–14 328. II-B, IV-A4, IV-B1, IV-B2, IV-C

    2022

  24. [32]

    Industrial image anomaly localization based on gaussian clustering of pre-trained feature,

    Q. Wan, L. Gao, X. Li, and L. Wen, “Industrial image anomaly localization based on gaussian clustering of pre-trained feature,” IEEE Transactions on Industrial Electronics , 2021. II-B

    2021

  25. [33]

    Anomaly detection via reverse distillation from one-class embedding,

    H. Deng and X. Li, “Anomaly detection via reverse distillation from one-class embedding,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 9737–9746. II-B

    2022

  26. [34]

    Unsupervised image anomaly detection and segmentation based on pre-trained feature mapping,

    Q. Wan, L. Gao, X. Li, and L. Wen, “Unsupervised image anomaly detection and segmentation based on pre-trained feature mapping,” IEEE Transactions on Industrial Informatics , 2022. II-B, IV-A4, IV-B1

    2022

  27. [35]

    Multiresolution knowledge distillation for anomaly detection,

    M. Salehi, N. Sadjadi, S. Baselizadeh, M. H. Rohban, and H. R. Rabiee, “Multiresolution knowledge distillation for anomaly detection,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 14 902–14 912. III-B, IV-A4, IV-B1, IV-C

    2021

  28. [36]

    Wide residual networks,

    S. Zagoruyko and N. Komodakis, “Wide residual networks,” arXiv preprint arXiv:1605.07146, 2016. III-B

    2016 arXiv

  29. [37]

    Imagenet large scale visual recognition challenge,

    O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. S. Bernstein, A. C. Berg, and L. Fei-Fei, “Imagenet large scale visual recognition challenge,” International Journal of Computer Vision , 2014. III-B

    2014

  30. [38]

    Masked au- toencoders are scalable vision learners,

    K. He, X. Chen, S. Xie, Y . Li, P. Doll ´ar, and R. Girshick, “Masked au- toencoders are scalable vision learners,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 16 000–16 009. III-C IEEE TRANSACTIONS ON AUTOMATION SCIENCE AND ...

    2022

  31. [39]

    Anomaly composition and decompo- sition network for accurate visual inspection of texture defects,

    K. Song, H. Yang, and Z. Yin, “Anomaly composition and decompo- sition network for accurate visual inspection of texture defects,” IEEE Transactions on Instrumentation and Measurement , vol. 71, pp. 1–14,

  32. [40]

    A unified model for multi-class anomaly detection,

    Z. You, L. Cui, Y . Shen, K. Yang, X. Lu, Y . Zheng, and X. Le, “A unified model for multi-class anomaly detection,” Advances in Neural Information Processing Systems, vol. 35, pp. 4571–4584, 2022. I, IV-A2, IV-A4, II, IV-C, IV-C, III, IV

    2022

  33. [41]

    Pyramidflow: High-resolution defect contrastive localization using pyramid normalizing flow,

    J. Lei, X. Hu, Y . Wang, and D. Liu, “Pyramidflow: High-resolution defect contrastive localization using pyramid normalizing flow,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 14 143–14 152. II

    2023

  34. [42]

    Revisiting reverse distillation for anomaly detection,

    T. D. Tien, A. T. Nguyen, N. H. Tran, T. D. Huy, S. Duong, C. D. T. Nguyen, and S. Q. Truong, “Revisiting reverse distillation for anomaly detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 24 511–24 520. II

    2023

  35. [43]

    Improv- ing unsupervised defect segmentation by applying structural similarity to autoencoders,

    P. Bergmann, S. L ¨owe, M. Fauser, D. Sattlegger, and C. Steger, “Improv- ing unsupervised defect segmentation by applying structural similarity to autoencoders,” arXiv preprint arXiv:1807.02011, 2018. IV-A4, IV-B1

    2018 arXiv

  36. [44]

    Unsupervised anomaly segmentation via deep feature reconstruction,

    Y . Shi, J. Yang, and Z. Qi, “Unsupervised anomaly segmentation via deep feature reconstruction,” Neurocomputing, 2021. IV-A4, IV-B1

    2021

  37. [45]

    Fastflow: Unsupervised anomaly detection and localization via 2d normalizing flows,

    J. Yu, Y . Zheng, X. Wang, W. Li, Y . Wu, R. Zhao, and L. Wu, “Fastflow: Unsupervised anomaly detection and localization via 2d normalizing flows,” arXiv preprint arXiv:2111.07677 , 2021. IV-A4, IV-B2

    2021 arXiv

  38. [46]

    Cfa: Coupled-hypersphere-based fea- ture adaptation for target-oriented anomaly localization,

    S. Lee, S. Lee, and B. C. Song, “Cfa: Coupled-hypersphere-based fea- ture adaptation for target-oriented anomaly localization,” IEEE Access, vol. 10, pp. 78 446–78 454, 2022. IV-A4, IV-B2

    2022

  39. [47]

    Uninformed students: Student-teacher anomaly detection with discriminative latent embeddings

    P. Bergmann, M. Fauser, D. Sattlegger, and C. Steger, “Uninformed students: Student-teacher anomaly detection with discriminative latent embeddings.” computer vision and pattern recognition , 2019. IV-A4, IV-C

    2019

  40. [48]

    Padim: A patch distribution modeling framework for anomaly detection and localization,

    D. Thomas, S. Aleksandr, L. Angelique, and A. Romaric, “Padim: A patch distribution modeling framework for anomaly detection and localization,” Lecture Notes in Computer Science , 2021. IV-A4, IV-C

    2021

  41. [49]

    Simplenet: A simple network for image anomaly detection and localization,

    Z. Liu, Y . Zhou, Y . Xu, and Z. Wang, “Simplenet: A simple network for image anomaly detection and localization,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 20 402–20 411. IV-A4, IV-C

    2023

  42. [50]

    A hierarchical transformation- discriminating generative model for few shot anomaly detection,

    S. Sheynin, S. Benaim, and L. Wolf, “A hierarchical transformation- discriminating generative model for few shot anomaly detection,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 8495–8504. IV-A4, IV-D

    2021

  43. [51]

    Same same but differnet: Semi-supervised defect detection with normalizing flows,

    M. Rudolph, B. Wandt, and B. Rosenhahn, “Same same but differnet: Semi-supervised defect detection with normalizing flows,” in Proceed- ings of the IEEE/CVF winter conference on applications of computer vision, 2021, pp. 1907–1916. IV-A4, IV-D

    2021

  44. [52]

    Registration based few-shot anomaly detection,

    C. Huang, H. Guan, A. Jiang, Y . Zhang, M. Spratling, and Y .-F. Wang, “Registration based few-shot anomaly detection,” in European Conference on Computer Vision. Springer, 2022, pp. 303–319. IV-A4, IV-D

    2022

  45. [53]

    Hierarchical vector quantized transformer for multi-class unsupervised anomaly detection,

    R. Lu, Y . Wu, L. Tian, D. Wang, B. Chen, X. Liu, and R. Hu, “Hierarchical vector quantized transformer for multi-class unsupervised anomaly detection,” arXiv preprint arXiv:2310.14228 , 2023. III

    2023 arXiv

  46. [54]

    Explicit boundary guided semi-push-pull contrastive learning for supervised anomaly detection,

    X. Yao, R. Li, J. Zhang, J. Sun, and C. Zhang, “Explicit boundary guided semi-push-pull contrastive learning for supervised anomaly detection,” in 2023 IEEE/CVF Conference on Computer Vision and Pattern Recog- nition (CVPR), 2023, pp. 24 490–24 499. IV-H Wei Luo (Student Membe...

    2023

  47. [2021]

    1, I, II-A3, IV-A4, IV-B1, 10

Pith tools

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