Pith. sign in

REVIEW 3 major objections 6 minor 1 cited by

Visual Anomaly Detection under Complex View-Illumination Interplay: A Large-Scale Benchmark

T0 review · 3 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A new benchmark shows anomaly detectors fail when viewpoint and lighting vary together.

desk verdict A useful new benchmark that fills a real gap in VAD evaluation, with a load-bearing detectability filter that needs sensitivity analysis before the invariant-protocol numbers become standard. read the letter →

arxiv 2505.10996 v1 pith:KJWLUEST submitted 2025-05-16 cs.CV

Yunkang Cao , Yuqi Cheng , Xiaohao Xu , Yiheng Zhang , Yihan Sun , Yuxiang Tan , Yuxin Zhang , Xiaonan Huang
show 1 more author
Weiming Shen
This is my paper · ORCID
classification cs.CV
keywords visualanomalydetectionbenchmarkdatasetmulti-viewimagingmulti-illuminationview-illuminationinterplayindustrialinspectionunsupervisedrobustnessevaluation
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

Visual anomaly detection — finding defects in images of manufactured objects — looks nearly solved on standard datasets, where the top method reports 99.6% AUROC. The paper argues this is misleading because current benchmarks keep viewpoint and illumination fixed or vary only one at a time, while real inspections face both changing together. To test that, it introduces M2AD, a benchmark of 119,880 high-resolution images of 999 specimens, each captured under 120 combinations of 12 viewpoints and 10 illumination conditions. On the two proposed protocols, M2AD-Synergy and M2AD-Invariant, state-of-the-art unsupervised methods degrade sharply; the same top method falls to 81.3% image-level AUROC on M2AD-Invariant. The paper concludes that benchmark saturation on standard industrial data overstates real-world readiness.

What carries the argument

The load-bearing object is the M2AD dataset itself, built by a capture rig that combines a motorized turntable providing 12 angular views in 30-degree steps with ten programmable illumination configurations, giving 120 calibrated images per specimen at 3,648 by 5,472 resolution. The paper adds a detectability filter for the M2AD-Invariant protocol: only abnormal images that three supervised detection models independently flag at IoU of at least 0.3 with confidence of at least 0.5 are retained, so the invariant benchmark measures robustness on anomalies that are arguably visible. The two evaluation protocols then define what counts as success, with M2AD-Synergy aggregating predictions across configurations to test fusion and M2AD-Invariant testing single-image robustness. The synchronized factorial design is what carries the argument, because without it the joint effect of view and illumination could not be separated from ordinary dataset difficulty.

What would settle it

Take the abnormal images that the detectability filter excluded and ask independent human annotators to mark the anomalies without seeing the supervised-model predictions; if humans can reliably label a large fraction of the excluded images, then the IoU and confidence thresholds are not measuring human-detectable anomalies, and the M2AD-Invariant numbers would need to be recomputed on a differently filtered set.

Watch

Extended reading notes

Core claim

The central discovery the paper tries to establish is that the view-illumination interplay, not any single imaging factor, is what breaks current visual anomaly detectors. On M2AD-Synergy, where a model can use all 120 configurations of a specimen, the best evaluated method, Dinomaly, reaches 90.0% object-level AUROC and 83.0% image-level AUROC; on M2AD-Invariant, which keeps single images but includes realistic view-illumination variation, the best method reaches only 81.3% image-level AUROC and 83.3% pixel-level AUPRO. These numbers contrast with the same method's 99.6% AUROC on MVTec AD. The paper also finds that simply averaging scores over more configurations does not help and can hurt, and that raising input resolution recovers several points, implying the remaining gap is partly about fine defect detail as well as about fusion.

Load-bearing premise

The M2AD-Invariant protocol assumes that anomalies agreed on by three supervised detection models at IoU of at least 0.3 and confidence of at least 0.5 are exactly the anomalies an unsupervised method should be expected to detect; if that filter is wrong, all M2AD-Invariant difficulty numbers inherit the error.

Editorial extensions

If this is right

  • State-of-the-art methods that saturate standard benchmarks cannot be assumed deployment-ready; M2AD-style evaluation is needed to expose the gap.
  • Current score-averaging fusion strategies over multiple views and illuminations are insufficient, making feature-level or physics-informed fusion a concrete target.
  • High-resolution inputs recover meaningful performance, around 3 to 6 percentage points in object-level AUROC, so resolution cannot be treated as a free parameter in industrial inspection.
  • The dual sub-categories in each of the ten classes make M2AD a usable testbed for cross-category generalization and zero-shot or few-shot anomaly detection.
  • Methods that explicitly model illumination and geometry, such as photometric-stereo or multi-view-stereo inspired approaches, become testable on real synchronized data.

Reading between the lines

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

  • The 12 by 10 factorial design also allows isolating how much of the failure is due to view changes versus illumination changes; the paper reports ablations on configuration count but not a full variance decomposition, so that decomposition is a natural next analysis.
  • If the detectability filter's IoU and confidence thresholds are miscalibrated, the M2AD-Invariant numbers would shift; a human re-annotation study of the excluded images would settle whether the filter is fair.
  • The synchronized multi-light capture means M2AD could double as a photometric-stereo benchmark, letting anomaly detectors consume estimated surface normals rather than raw pixel images; that is an extension the paper does not itself explore.
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 / 6 minor

Summary. The paper introduces M2AD, a large-scale visual anomaly detection benchmark with 119,880 high-resolution images of 999 specimens across 10 categories, captured under 12 synchronized viewpoints and 10 illumination conditions (120 configurations per specimen). The authors define two evaluation protocols: M2AD-Synergy, which aggregates predictions across all configurations to test multi-view/multi-illumination fusion, and M2AD-Invariant, which evaluates single-image robustness on a subset of anomalies deemed 'detectable' by a supervised Mask R-CNN consensus filter. They benchmark five recent unsupervised methods (CDO, RD++, MSFlow, Dinomaly, INP-Former) and report substantial performance drops compared with published scores on MVTec AD, e.g., Dinomaly drops from 99.6% to 81.3% I-AUROC on M2AD-Invariant. Additional experiments examine input resolution, the number of configurations, and the number of illumination conditions.

Significance. If the dataset construction and protocols are sound, M2AD is a valuable contribution: it is the first large-scale VAD benchmark to jointly control viewpoint and illumination, it provides high-resolution imagery with sub-millimeter defects, and its two complementary protocols target different failure modes of current methods. The consistent performance drops across five independent methods and two protocols, the ablations with means and standard deviations, and the fact that the authors' own methods (CDO, INP-Former) perform relatively poorly all support the claim that existing VAD methods are not robust to view-illumination interplay. The planned public release of data, test suite, and imaging prototype design further strengthens reproducibility. However, the headline quantitative claims rest on two under-specified components: the supervised detectability filter that defines the M2AD-Invariant subset, and the train/test split used for both protocols. These issues must be resolved before the benchmark's central 'profound challenge' claim is fully secure.

major comments (3)
  1. [Section 3.1, 'Detectability Assessment' and Table 4] The M2AD-Invariant protocol retains only abnormal images where three supervised Mask R-CNN models agree with manual annotations at IoU >= 0.3 and confidence >= 0.5, yet no sensitivity analysis or external validation of these thresholds is provided. Because Table 4 and the headline 81.3% I-AUROC for Dinomaly are computed exclusively on this retained subset, a miscalibrated filter directly changes the central claim. The direction of the bias is unknown: if the filter preferentially retains high-contrast anomalies that supervised models find easy, the reported difficulty is understated; if it retains texture-based cues that supervised models exploit but unsupervised methods do not, the difficulty is overstated. I request a sensitivity analysis over threshold values, and ideally a control experiment on the full unfiltered set or a comparison with human detectability ratings.
  2. [Section 4.1, 'Benchmark Setups' and Section 4.3, 'M2AD-Invariant Results'] The paper never specifies the exact train/test split for either protocol. The statement in Section 4.3 that 'all imaging configurations are utilized for training' is ambiguous: if all normal images from all 120 configurations are used in training and the same normal images are then used to compute I-AUROC, the evaluation violates the standard unsupervised VAD protocol of held-out normal test images and would artificially inflate scores. The benchmark setup in Section 4.1 also lacks a clear statement of how normal and abnormal specimens or images are partitioned. This is a load-bearing specification issue, as the reported performance gaps are only interpretable if the split is standard and reproducible.
  3. [Section 3.1, 'Detectability Assessment' and Fig. 2(c)] The manual annotation procedure for images in which a defect is visually absent or undetectable is not documented. The figure refers to 'Manual Labeling' and 'Human Checking', but the paper does not state how annotators decided that an anomaly is not visible under a given view-illumination configuration, nor how inter-annotator disagreements were resolved. Since these manual labels are the ground truth against which the supervised detectability filter is trained, the reliability of the entire M2AD-Invariant subset depends on this undocumented annotation step.
minor comments (6)
  1. [Throughout] The text contains inconsistent spacing in the abbreviation 'V AD' instead of 'VAD' in several places (e.g., the abstract and Section 1); this should be harmonized.
  2. [Fig. 3(a)] The caption and legend for Fig. 3(a) are difficult to parse; the relationship between 'All/Detectable Abnormal' counts and the text's statement that about 75% of abnormal images are detectable should be made explicit with per-category numbers.
  3. [Section 4.2, Fig. 4] The ablation in Fig. 4 randomly selects configurations and illumination conditions three times, but the sampling procedure (whether subsets are per specimen, whether they are stratified by view/illumination, and what random seeds are used) is not described, which limits reproducibility.
  4. [Section 4.3, 'Quantitative Results'] The sentence stating that 'all imaging configurations are utilized for training' belongs in the protocol definition in Section 4.1, not in the results section; currently it is easy to miss when reading the benchmark setup.
  5. [Table 1] The row for PAD lists '20' under the main category column and '30' under total number of categories, which is confusing; the column semantics for 'Main' and 'Sub.' should be clarified, and the counts should be double-checked against the cited source.
  6. [References] Reference [8] (Dinomaly) is an arXiv preprint without a year or venue; since the paper is from CVPR 2025, the citation should be updated to the published version if available.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: M2AD is an externally constructed benchmark, the evaluated methods are independent baselines, and the authors' own methods perform poorly, so the central 'profound challenge' claim is an empirical finding rather than a construction artifact.

full rationale

The paper's central claim—that state-of-the-art VAD methods degrade on M2AD relative to MVTec AD—is an empirical benchmark result, not a derivation from the benchmark's construction. The M2AD-Invariant protocol selects images using three supervised Mask R-CNN models trained on manual annotations (Section 3.1); the evaluated methods (CDO, RD++, MSFlow, Dinomaly, INP-Former) are unsupervised, and none is used to define the subset or to set any threshold, so no reported score is fitted or equivalent to an input by construction. The authors' own methods (CDO [24] and INP-Former [22]) perform poorly (72.2% and 67.3% I-AUROC on M2AD-Invariant), which argues against self-selection bias. Self-citations appear only as evaluated baselines or related-work pointers, not as load-bearing justification for the benchmark's conclusions; there is no uniqueness theorem or ansatz imported from prior work that forces the outcome. The supervised detectability filter could affect the interpretation of the absolute numbers, but that is a validity or calibration concern about threshold choice, not a circular reduction, and no sensitivity analysis is being passed off as a prediction. Hence no circular step is identifiable.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

No invented physical entities appear. The only tunable inputs are the detectability filter thresholds, which directly control the composition of the M2AD-Invariant test set. The axioms listed are the background assumptions about annotation quality, representativeness of defects, and the validity of the supervised detectability oracle. These assumptions are standard for benchmark papers but should be stated explicitly because the central numbers depend on them.

free parameters (2)
  • IoU threshold for detectability filter = 0.3
    Images where Mask R-CNN predictions have IoU below 0.3 with manual annotations are excluded from M2AD-Invariant. This threshold determines which anomalies are considered 'detectable' and shapes all I-AUROC/AUPRO results in Table 4.
  • Confidence threshold for detectability filter = 0.5
    Predictions with confidence below 0.5 are excluded, further shaping the M2AD-Invariant subset. The threshold is chosen ad hoc and no sensitivity analysis is provided.
assumptions (4)
  • domain assumption Manual annotations accurately label all anomalous regions.
    The dataset relies on human labeling as ground truth (Section 3.1). If labels are noisy or miss defects, benchmark numbers become unreliable.
  • ad hoc to paper Mask R-CNN prediction consensus is a valid proxy for anomaly detectability.
    M2AD-Invariant only retains images where three Mask R-CNN models agree with human annotation at IoU>=0.3 and p>=0.5. This assumes a supervised detector's salience matches what an unsupervised method should detect, which is not established.
  • domain assumption Engineered defects are representative of real-world manufacturing anomalies.
    The defects include perforations, abrasions, deformations, and bending (Section 3.1). If these do not match real defect distributions, the benchmark's ecological validity is limited.
  • domain assumption The 12-view by 10-illumination capture protocol models realistic imaging conditions.
    The authors assume synchronized rotation and programmable illumination covers all exposed surfaces and reflects real operational variability (Section 3.1). Occluded basal regions remain unobserved, so coverage is not complete.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Visual Anomaly Detection under Complex View-Illumination Interplay: A Large-Scale Benchmark." pith.science (2026). https://pith.science/paper/KJWLUEST

@misc{pith2026250510996,
  author       = {Pith},
  title        = {Pith review of: Visual Anomaly Detection under Complex View-Illumination Interplay: A Large-Scale Benchmark},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KJWLUEST}},
  note         = {Machine review of arXiv:2505.10996}
}
read the original abstract

The practical deployment of Visual Anomaly Detection (VAD) systems is hindered by their sensitivity to real-world imaging variations, particularly the complex interplay between viewpoint and illumination which drastically alters defect visibility. Current benchmarks largely overlook this critical challenge. We introduce Multi-View Multi-Illumination Anomaly Detection (M2AD), a new large-scale benchmark comprising 119,880 high-resolution images designed explicitly to probe VAD robustness under such interacting conditions. By systematically capturing 999 specimens across 10 categories using 12 synchronized views and 10 illumination settings (120 configurations total), M2AD enables rigorous evaluation. We establish two evaluation protocols: M2AD-Synergy tests the ability to fuse information across diverse configurations, and M2AD-Invariant measures single-image robustness against realistic view-illumination effects. Our extensive benchmarking shows that state-of-the-art VAD methods struggle significantly on M2AD, demonstrating the profound challenge posed by view-illumination interplay. This benchmark serves as an essential tool for developing and validating VAD methods capable of overcoming real-world complexities. Our full dataset and test suite will be released at https://hustcyq.github.io/M2AD to facilitate the field.

Figures

Figures reproduced from arXiv: 2505.10996 by the authors.

Figure 1
Figure 1. Motivation. (a) Anomaly detectability is governed by complex view-illumination interplay. Each image pair shows the original input (left) alongside its corresponding ground truth (right), with anomaly regions highlighted in white. (b) To address this challenge, our M2AD introduces multi￾view and multi-illumination acquisition protocols, enabling robust anomaly detection across diverse conditions. Zoom in for a clear… view at source ↗
Figure 2
Figure 2. Data collection pipeline of M2AD. A three-step process is employed. (a) Object preparation and defect engineering. (b) Design and construction of a configurable imaging prototype capable of capturing multi-view, multi-illumination images programmatically. (c) Assessing detectability by evaluating the consistency between predictions and annotations for M2AD-Invariant [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Statistics of M2AD. (a) Distribution of normal and abnormal object, view, and image counts across different categories. “Detectable” refers to the abnormal images retained in Sec. 3.1. (b) Percentage of image area occupied by anomaly regions. (c) Aspect ratio statistics of the minimum bounding rectangle of defects. Deliberately eschewing mechanical complexity inherent in industrial inspection systems requiring speci… view at source ↗
Figures from the paper (17 more)
Figure 4
Figure 4. Figure 4: Ablation study results. (a) O-AUROC under different imaging configuration numbers (24, 48, 72, 96, and 120). (b) I-AUROC and AUPRO under different combinations of illumination conditions (2, 4, 6, 8, and 10). We randomly select the configurations and illumination condi…
Figure 5
Figure 5. Figure 5: Visualization of anomaly detection results. (a) Input image, (b) ground truth (anomalies highlighted in white), (c) predicted anomaly maps by the best-performing model Dinomaly [8]. Despite its robustness, the visualization demonstrates Dinomaly’s limitations in captur…
Figure 6
Figure 6. Figure 6: Visualization of anomaly detection results. (a) Input image, (b) ground truth (anomalies highlighted in white), (c)∼(g) predicted anomaly maps by CDO [24], RD++ [37], MSFlow [38] Dinomaly [8], and INP-Former [22], respectively. Zoom in for a clearer view. A.3 Anomaly D…
Figure 7
Figure 7. Figure 7: Visualization of anomaly detection results for multi-illumination images. 01∼10 corresponds to the illumination conditions in [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]
Figure 8
Figure 8. Figure 8: Visualization of anomaly detection results for multi-illumination images. 01∼10 corresponds to the illumination conditions in [PITH_FULL_IMAGE:figures/full_fig_p016_8.png]
Figure 9
Figure 9. Figure 9: Visualization of all the categories in M2AD. Each group presents dual sub-categories. 01 02 03 04 05 06 07 08 09 10 [PITH_FULL_IMAGE:figures/full_fig_p017_9.png]
Figure 10
Figure 10. Figure 10: Schematic illustrations of distinct illumination configurations. Through programmable control of the photometric illumination module, we sequentially generate ten distinct illumination conditions and acquire corresponding multi-illumination image sequences for compreh…
Figure 11
Figure 11. Figure 11: Visualization of Bird. From top to bottom: images organized by views; from left to right: images organized by illumination condition. Normal images are highlighted with green borders, whereas abnormal images are marked with red borders for comparison. 18 [PITH_FULL_I…
Figure 12
Figure 12. Figure 12: Visualization of Car. From top to bottom: images organized by views; from left to right: images organized by illumination condition. Normal images are highlighted with green borders, whereas abnormal images are marked with red borders for comparison. 19 [PITH_FULL_IM…
Figure 13
Figure 13. Figure 13: Visualization of Cube. From top to bottom: images organized by views; from left to right: images organized by illumination condition. Normal images are highlighted with green borders, whereas abnormal images are marked with red borders for comparison. 20 [PITH_FULL_I…
Figure 14
Figure 14. Figure 14: Visualization of Dice. From top to bottom: images organized by views; from left to right: images organized by illumination condition. Normal images are highlighted with green borders, whereas abnormal images are marked with red borders for comparison. 21 [PITH_FULL_I…
Figure 15
Figure 15. Figure 15: Visualization of Doll. From top to bottom: images organized by views; from left to right: images organized by illumination condition. Normal images are highlighted with green borders, whereas abnormal images are marked with red borders for comparison. 22 [PITH_FULL_I…
Figure 16
Figure 16. Figure 16: Visualization of Holder. From top to bottom: images organized by views; from left to right: images organized by illumination condition. Normal images are highlighted with green borders, whereas abnormal images are marked with red borders for comparison. 23 [PITH_FULL…
Figure 17
Figure 17. Figure 17: Visualization of Motor. From top to bottom: images organized by views; from left to right: images organized by illumination condition. Normal images are highlighted with green borders, whereas abnormal images are marked with red borders for comparison. 24 [PITH_FULL_…
Figure 18
Figure 18. Figure 18: Visualization of Ring. From top to bottom: images organized by views; from left to right: images organized by illumination condition. Normal images are highlighted with green borders, whereas abnormal images are marked with red borders for comparison. 25 [PITH_FULL_I…
Figure 19
Figure 19. Figure 19: Visualization of Teapot. From top to bottom: images organized by views; from left to right: images organized by illumination condition. Normal images are highlighted with green borders, whereas abnormal images are marked with red borders for comparison. 26 [PITH_FULL…
Figure 20
Figure 20. Figure 20: Visualization of Tube. From top to bottom: images organized by views; from left to right: images organized by illumination condition. Normal images are highlighted with green borders, whereas abnormal images are marked with red borders for comparison. 27 [PITH_FULL_I…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. A Comprehensive Survey for Real-World Industrial Defect Detection: Challenges, Approaches, and Prospects

    cs.CV 2025-07 conditional novelty 3.0 of 10

    A broad survey of industrial defect detection that structures the field by closed-set vs open-set and 2D vs 3D methods, with an emphasis on the rise of open-set anomaly detection.

Reference graph

Works this paper leans on

39 extracted references · 30 canonical work pages · cited by 1 Pith paper

  1. [1]

    The MVTec anomaly detection dataset: A comprehensive real-world dataset for unsupervised anomaly detection

    Paul Bergmann, Kilian Batzner, Michael Fauser, David Sattlegger, and Carsten Steger. The MVTec anomaly detection dataset: A comprehensive real-world dataset for unsupervised anomaly detection. International Journal of Computer Vision, 129(4):1038–1059, 2021

    work page 2021

  2. [2]

    Spot-the-difference self-supervised pre-training for anomaly detection and segmentation

    Yang Zou, Jongheon Jeong, Latha Pemula, Dongqing Zhang, and Onkar Dabeer. Spot-the-difference self-supervised pre-training for anomaly detection and segmentation. InEuropean Conference on Computer Vision, pages 392–408. Springer, 2022

    work page 2022

  3. [3]

    Real-iad: A real-world multi-view dataset for benchmarking versatile industrial anomaly detection

    Chengjie Wang, Wenbing Zhu, Bin-Bin Gao, Zhenye Gan, Jiangning Zhang, Zhihao Gu, Shuguang Qian, Mingang Chen, and Lizhuang Ma. 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, pages 22883–22892, 2024

    work page 2024

  4. [4]

    Pad: A dataset and benchmark for pose-agnostic anomaly detection

    Qiang Zhou, Weize Li, Lihan Jiang, Guoliang Wang, Guyue Zhou, Shanghang Zhang, and Hao Zhao. Pad: A dataset and benchmark for pose-agnostic anomaly detection. In A. Oh, T. Naumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine, editors, Advances in Neural Information Processing Systems , volume 36, pages 44558–44571. Curran Associates, Inc., 2023

    work page 2023

  5. [5]

    The mvtec ad 2 dataset: Advanced scenarios for unsupervised anomaly detection

    Lars Heckler-Kram, Jan-Hendrik Neudeck, Ulla Scheler, Rebecca König, and Carsten Steger. The mvtec ad 2 dataset: Advanced scenarios for unsupervised anomaly detection. ArXiv, abs/2503.21622, 2025

    arXiv 2025

  6. [6]

    The eyecandies dataset for unsupervised multimodal anomaly detection and localization

    Luca Bonfiglioli, Marco Toschi, Davide Silvestri, Nicola Fioraio, and Daniele De Gregorio. The eyecandies dataset for unsupervised multimodal anomaly detection and localization. In Lei Wang, Juergen Gall, Tat- Jun Chin, Imari Sato, and Rama Chellappa, editors, Asian Conference on Computer Vision, volume 13845 of Lecture Notes in Computer Science , pages 4...

    work page 2022

  7. [7]

    Adaclip: Adapting clip with hybrid learnable prompts for zero-shot anomaly detection

    Yunkang Cao, Jiangning Zhang, Luca Frittoli, Yuqi Cheng, Weiming Shen, and Giacomo Boracchi. Adaclip: Adapting clip with hybrid learnable prompts for zero-shot anomaly detection. In European Conference on Computer Vision, 2024

    work page 2024

  8. [8]

    Dinomaly: The less is more philosophy in multi-class unsupervised anomaly detection

    Jia Guo, Shuai Lu, Weihang Zhang, Fang Chen, Huiqi Li, and Hongen Liao. Dinomaly: The less is more philosophy in multi-class unsupervised anomaly detection. ArXiv, 2025

    work page 2025

Show all 39 references
  1. [9]

    Mixed supervision for surface-defect detection: From weakly to fully supervised learning

    Jakob Božiˇc, Domen Tabernik, and Danijel Skoˇcaj. Mixed supervision for surface-defect detection: From weakly to fully supervised learning. Computers in Industry, 129:103459, 2021

    2021

  2. [10]

    VT-ADL: A vision transformer network for image anomaly detection and localization

    Pankaj Mishra, Riccardo Verk, Daniele Fornasier, Claudio Piciarelli, and Gian Luca Foresti. VT-ADL: A vision transformer network for image anomaly detection and localization. InIEEE International Symposium on Industrial Electronics, pages 01–06. IEEE, 2021

    2021

  3. [11]

    Surface defect saliency of magnetic tile

    Yibin Huang, Congying Qiu, Yue Guo, Xiaonan Wang, and Kui Yuan. Surface defect saliency of magnetic tile. In International Conference on Automation Science and Engineering (CASE) , pages 612–617, 2018

    2018

  4. [12]

    The mvtec 3d-ad dataset for unsupervised 3d anomaly detection and localization

    Paul Bergmann, Xin Jin, David Sattlegger, and Carsten Steger. The mvtec 3d-ad dataset for unsupervised 3d anomaly detection and localization. Proceedings of the 17th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications , 5:20...

    2022 doi

  5. [13]

    Real3d- ad: A dataset of point cloud anomaly detection

    Jiaqi Liu, Guoyang Xie, Xinpeng Li, Jinbao Wang, Yong Liu, Chengjie Wang, Feng Zheng, et al. Real3d- ad: A dataset of point cloud anomaly detection. In Thirty-seventh Conference on Neural Information Processing Systems Datasets and Benchmarks Track, volume 36, 2024

    2024

  6. [14]

    Real-iad d3: A real-world 2d/pseudo-3d/3d dataset for industrial anomaly detection

    Wenbing Zhu, Lidong Wang, Ziqing Zhou, Chengjie Wang, Yurui Pan, Ruoyi Zhang, Zhuhao Chen, Linjie Cheng, Bin-Bin Gao, Jiangning Zhang, Zhenye Gan, Yuxie Wang, Yulong Chen, Shuguang Qian, Mingmin Chi, Bo Peng, and Lizhuang Ma. Real-iad d3: A real-world 2d/pseudo-3d/3d dataset f...

    2025

  7. [15]

    Rad: A dataset and benchmark for real-life anomaly detection with robotic observations

    Kaichen Zhou, Yang Cao, Teawhan Kim, Hao Zhao, Hao Dong, Kai Ming Ting, and Ye Zhu. Rad: A dataset and benchmark for real-life anomaly detection with robotic observations. ArXiv, abs/2410.00713, 2024

    2024 arXiv

  8. [16]

    Manta: A large-scale multi-view and visual-text anomaly detection dataset for tiny objects

    Lei Fan, Dongdong Fan, Zhiguang Hu, Yiwen Ding, Donglin Di, Kai Yi, Maurice Pagnucco, and Yang Song. Manta: A large-scale multi-view and visual-text anomaly detection dataset for tiny objects. arXiv preprint arXiv:2412.04867, 2024. 10

    2024 arXiv

  9. [17]

    Industrial anomaly detection with domain shift: A real-world dataset and masked multi-scale reconstruction

    Zilong Zhang, Zhibin Zhao, Xingwu Zhang, Chuang Sun, and Xuefeng Chen. Industrial anomaly detection with domain shift: A real-world dataset and masked multi-scale reconstruction. Computers in Industry, 151:103990, 2023

    2023

  10. [18]

    IRON: inverse rendering by optimizing neural sdfs and materials from photometric images

    Kai Zhang, Fujun Luan, Zhengqi Li, and Noah Snavely. IRON: inverse rendering by optimizing neural sdfs and materials from photometric images. In IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR 2022, New Orleans, LA, USA, June 18-24, 2022 , pages 5555–5564....

    2022

  11. [19]

    Geo-neus: Geometry-consistent neural implicit surfaces learning for multi-view reconstruction

    Qiancheng Fu, Qingshan Xu, Yew Soon Ong, and Wenbing Tao. Geo-neus: Geometry-consistent neural implicit surfaces learning for multi-view reconstruction. In Sanmi Koyejo, S. Mohamed, A. Agarwal, Danielle Belgrave, K. Cho, and A. Oh, editors, Advances in Neural Information Proce...

    2022

  12. [20]

    Towards visual discrimination and reasoning of real-world physical dynamics: Physics-grounded anomaly detection

    Wenqiao Li, Yao Gu, Xintao Chen, Xiaohao Xu, Ming Hu, Xiaonan Huang, and Yingna Wu. Towards visual discrimination and reasoning of real-world physical dynamics: Physics-grounded anomaly detection. ArXiv, abs/2503.03562, 2025

    2025 arXiv

  13. [21]

    Holmes-vau: Towards long-term video anomaly understanding at any granularity

    Huaxin Zhang, Xiaohao Xu, Xiangdong Wang, Jia li Zuo, Xiaonan Huang, Changxin Gao, Shanjun Zhang, Li Yu, and Nong Sang. Holmes-vau: Towards long-term video anomaly understanding at any granularity. ArXiv, abs/2412.06171, 2024

    2024 arXiv

  14. [22]

    Exploring intrinsic normal prototypes within a single image for universal anomaly detection

    Wei Luo, Yunkang Cao, Haiming Yao, Xiaotian Zhang, Jianan Lou, Yuqi Cheng, Weiming Shen, and Wenyong Yu. Exploring intrinsic normal prototypes within a single image for universal anomaly detection. arXiv preprint arXiv:2503.02424, 2025

    2025 arXiv

  15. [23]

    MambaAD: Exploring State Space Models for Multi-class Unsupervised Anomaly Detection

    Haoyang He, Yuhu Bai, Jiangning Zhang, Qingdong He, Hongxu Chen, Zhenye Gan, Chengjie Wang, Xiangtai Li, Guanzhong Tian, and Lei Xie. MambaAD: Exploring State Space Models for Multi-class Unsupervised Anomaly Detection. In Amir Globersons, Lester Mackey, Danielle Belgrave, Ang...

    2024

  16. [24]

    Collaborative discrepancy optimization for reliable image anomaly localization

    Yunkang Cao, Xiaohao Xu, Zhaoge Liu, and Weiming Shen. Collaborative discrepancy optimization for reliable image anomaly localization. IEEE Transactions on Industrial Informatics , pages 1–10, 2023

    2023

  17. [25]

    Remembering Normality: Memory-guided Knowledge Distillation for Unsupervised Anomaly Detection

    Zhihao Gu, Liang Liu, Xu Chen, Ran Yi, Jiangning Zhang, Yabiao Wang, Chengjie Wang, Annan Shu, Guannan Jiang, and Lizhuang Ma. Remembering Normality: Memory-guided Knowledge Distillation for Unsupervised Anomaly Detection. pages 16355–16363, Paris, France, October 1-6, 2023, O...

    2023

  18. [26]

    A unified anomaly synthesis strategy with gradient ascent for industrial anomaly detection and localization

    Qiyu Chen, Huiyuan Luo, Chengkan Lv, and Zhengtao Zhang. A unified anomaly synthesis strategy with gradient ascent for industrial anomaly detection and localization. pages 37–54, Milan, Italy, September 29-October 4, 2024, 2025. Springer

    2024

  19. [27]

    Hierarchical gaussian mixture normalizing flow modeling for unified anomaly detection

    Xincheng Yao, Ruoqi Li, Zefeng Qian, Lu Wang, and Chongyang Zhang. Hierarchical gaussian mixture normalizing flow modeling for unified anomaly detection. In European Conference on Computer Vision, pages 92–108. Springer, 2024

    2024

  20. [28]

    Towards Total Recall in Industrial Anomaly Detection

    Karsten Roth, Latha Pemula, Joaquin Zepeda, Bernhard Scholkopf, Thomas Brox, and Peter Gehler. Towards Total Recall in Industrial Anomaly Detection. pages 14298–14308, New Orleans, LA, USA, June 18-24, 2022, June 2022. IEEE

    2022

  21. [29]

    Anomalyclip: Object-agnostic prompt learning for zero-shot anomaly detection

    Qihang Zhou, Guansong Pang, Yu Tian, Shibo He, and Jiming Chen. Anomalyclip: Object-agnostic prompt learning for zero-shot anomaly detection. In International Conference on Learning Representations, , 2024

    2024

  22. [30]

    Multimodal industrial anomaly detection via hybrid fusion

    Yue Wang, Jinlong Peng, Jiangning Zhang, Ran Yi, Yabiao Wang, and Chengjie Wang. Multimodal industrial anomaly detection via hybrid fusion. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 8032–8041, 2023

    2023

  23. [31]

    Easynet: An easy network for 3d industrial anomaly detection

    Ruitao Chen, Guoyang Xie, Jiaqi Liu, Jinbao Wang, Ziqi Luo, Jinfan Wang, and Feng Zheng. Easynet: An easy network for 3d industrial anomaly detection. In Proceedings of the 31st ACM International Conference on Multimedia, pages 7038–7046, 2023

    2023

  24. [32]

    Multi-sensor object anomaly detection: Unifying appearance, geometry, and internal properties

    Wenqiao Li, Bozhong Zheng, Xiaohao Xu, Jinye Gan, Fading Lu, Xiang Li, Na Ni, Zheng Tian, Xiaonan Huang, Shenghua Gao, et al. Multi-sensor object anomaly detection: Unifying appearance, geometry, and internal properties. arXiv preprint arXiv:2412.14592, 2024. 11

    2024 arXiv

  25. [33]

    Learning diffusion models for multi-view anomaly detection

    Chieh Liu, Yu-Min Chu, Ting-I Hsieh, Hwann-Tzong Chen, and Tyng-Luh Liu. Learning diffusion models for multi-view anomaly detection. In European Conference on Computer Vision, pages 328–345. Springer, 2024

    2024

  26. [34]

    Attention fusion reverse distillation for multi-lighting image anomaly detection

    Yiheng Zhang, Yunkang Cao, Tianhang Zhang, and Weiming Shen. Attention fusion reverse distillation for multi-lighting image anomaly detection. In 2024 IEEE 20th International Conference on Automation Science and Engineering (CASE) , pages 2134–2139. IEEE, 2024

    2024

  27. [35]

    Learning multi-view anomaly detection

    Haoyang He, Jiangning Zhang, Guanzhong Tian, Chengjie Wang, and Lei Xie. Learning multi-view anomaly detection. arXiv preprint arXiv:2407.11935, 2024

    2024 arXiv

  28. [36]

    Splatpose & detect: Pose- agnostic 3d anomaly detection

    Mathis Kruse, Marco Rudolph, Dominik Woiwode, and Bodo Rosenhahn. Splatpose & detect: Pose- agnostic 3d anomaly detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 3950–3960, 2024

    2024

  29. [37]

    Revisiting reverse distillation for anomaly detection

    Tran Dinh Tien, Anh Tuan Nguyen, Nguyen Hoang Tran, Ta Duc Huy, Soan Duong, Chanh D Tr Nguyen, and Steven QH Truong. Revisiting reverse distillation for anomaly detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 24511–24520,...

    2023

  30. [38]

    Msflow: Multiscale flow-based framework for unsupervised anomaly detection

    Yixuan Zhou, Xing Xu, Jingkuan Song, Fumin Shen, and Heng Tao Shen. Msflow: Multiscale flow-based framework for unsupervised anomaly detection. IEEE Transactions on Neural Networks and Learning Systems, 2024

    2024

  31. [39]

    Black Bird

    Yunkang Cao, Haiming Yao, Wei Luo, and Weiming Shen. Varad: Lightweight high-resolution image anomaly detection via visual autoregressive modeling. IEEE Transactions on Industrial Informatics , 21(4):3246–3255, 2025. 12 A Appendix The supplementary material includes the follow...

    2025

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.