Pith. sign in

REVIEW 3 major objections 6 minor 28 references

Detect, Classify, Act: Categorizing Industrial Anomalies with Multi-Modal Large Language Models

T0 review · 3 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read A two-stage pipeline—fast anomaly detector plus multimodal LLM—classifies industrial defects at 84 percent accuracy without task-specific training.

desk verdict The pipeline idea is genuinely useful, but the headline accuracy numbers don't match the paper's own metric equation, so the SOTA claim is uninterpretable until they clarify what was computed. read the letter →

arxiv 2505.02626 v1 pith:N44RVE35 submitted 2025-05-05 cs.CV

Sassan Mokhtar , Arian Mousakhan , Silvio Galesso , Jawad Tayyub , Thomas Brox This is my paper · ORCID
classification cs.CV
keywords anomalyclassificationindustrialinspectionmultimodallargelanguagemodelsdetectionMVTec-ADVisAvision-languagezero-shot
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

The paper sets out to turn industrial anomaly detection into anomaly classification: instead of only flagging an image as normal or abnormal, a system should name the type of defect and support decisions about whether it matters. The proposed pipeline, VELM, pairs a fast unsupervised anomaly detector with a multimodal LLM (a language model that also takes images as input), and the LLM has no task-specific training. The detector filters normal images and outlines suspicious regions; the LLM, guided by a text prompt that defines the normal object, the candidate defect classes, and a decision strategy, assigns the class. On the original MVTec-AD benchmark the method reports 81.4 percent accuracy, five points above the previous best, and on the authors' relabeled MVTec-AC dataset it reaches 84.0 percent with a real detector. The paper also contributes corrected class labels for two standard datasets, arguing that reliable evaluation of classification requires cleaner annotations than existing detection benchmarks provide.

What carries the argument

The load-bearing mechanism is the division of labor between a fast pixel-level anomaly detector and a multimodal LLM, joined by two kinds of prompts. The visual prompt is a red-line contour overlay on the localized anomaly, presented next to the query image and a normal reference image; the text prompt is structured into a normal-object description, anomaly class definitions, and an explicit classification strategy. The detector's job is to keep classification cheap and accurate by filtering normal samples and telling the LLM where to look, while the LLM's job is to apply semantic knowledge and follow user-defined class definitions. Ablation results show that removing the anomaly descriptions causes the largest accuracy drop, and that both the reference image and the contour overlay contribute; the complete prompt gives the best accuracy, supporting the claim that both visual and textual context matter.

What would settle it

Have independent annotators re-label a random subset of MVTec-AC and VisA-AC from the original images using the paper's class definitions and compare their labels with the published ones; low inter-annotator agreement or systematic disagreement with the original dataset labels would mean the reported accuracy figures are not a valid measure of classification performance.

Watch

Extended reading notes

Core claim

The central claim is that a specialized localization module and a semantically rich language model can be composed into a classifier that needs no training for the classification task. In VELM, a pixel-level anomaly detector acts as the Vision Expert: it decides whether an image is normal, and when it detects an anomaly it produces an anomaly map. That map is converted into a visual prompt by drawing a red contour around the detected region, and the LLM receives the query image, the contour image, a normal reference image, and a structured text prompt containing a normal-object description, definitions of every anomaly class, and a classification strategy. The paper reports that on MVTec-AD this composition reaches 81.4 percent accuracy, outperforming prior anomaly classification methods by 5 percentage points; on the refined MVTec-AC benchmark it reaches 84.0 percent with the DDAD detector and 87.8 percent when ground-truth masks are used instead. The authors further claim the same pipeline separates benign anomalies from critical defects with 89.8 percent accuracy in a simulated three-way normal/anomaly/defect task.

Load-bearing premise

The headline accuracy numbers rest on the paper's own manual relabeling of MVTec-AD and VisA—36 corrected samples, merged classes, and removal of small classes—being the correct ground truth, and no independent annotator or external audit checks those labels.

Editorial extensions

If this is right

  • Anomaly classification becomes a zero-shot task: with an off-the-shelf detector and a pretrained multimodal LLM, an inspection system can name defect types without collecting labeled defect examples for each new object.
  • Classification accuracy is bounded by localization quality; the gap between oracle masks (87.8 percent) and real detectors (84.0 percent for DDAD, 78.1 percent for PatchCore) implies that better anomaly segmentation directly raises classification performance.
  • Prompt structure is a real component of the method; removing anomaly descriptions lowers accuracy by about five points, so the way classes are defined in language is part of the approach, not a detail.
  • A single pipeline can triage anomalies into normal, negligible anomaly, and critical defect at 89.8 percent accuracy in the paper's simulated setting, which is the decision problem industrial inspection actually faces.
  • MVTec-AC and VisA-AC offer a benchmark for classification rather than detection, and their corrected and merged labels can support comparable evaluation of future methods.

Reading between the lines

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

  • One implication left implicit is that the random choice of a normal reference image introduces variance in the reported numbers; a natural extension is to evaluate across multiple reference images per query or to learn which reference best supports the prompt.
  • If the dependency on localization quality holds generally, then the practical bottleneck for new industrial objects is the detector, not the language model; swapping in a detector specialized per object family should be tested as a way to close the gap to the oracle result.
  • The closed-set limitation named in the paper suggests a next step: using the LLM's open vocabulary to output a description when the anomaly fits no user-defined class, which would measure how much the semantic model adds beyond the predefined labels.
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 proposes VELM, a two-stage pipeline for industrial anomaly classification that combines a fast unsupervised anomaly detector (DDAD or PatchCore) with a multimodal LLM (GPT-4o or GPT-4o-mini). When the detector flags an anomaly, the LLM receives a normal reference image, the query image, a red-contour visual prompt derived from the detector's localization, and a structured text prompt that defines the anomaly classes and a classification strategy. The authors also introduce two refined datasets, MVTec-AC and VisA-AC, with corrected, merged, and filtered anomaly class labels, and evaluate VELM on these and on the original MVTec-AD. They report state-of-the-art accuracy on MVTec-AD (81.4% in the body, 80.4% in the abstract) and strong performance on MVTec-AC (84.0% with DDAD+GPT-4o), together with ablations of the prompt components and a simulated anomaly-vs-defect experiment.

Significance. The two-stage design is practically motivated: using a fast detector to filter normal images and to localize anomalies before invoking the LLM is a sensible way to balance speed and semantic flexibility. The ablations in Table 5 clearly show the value of the reference image, visual prompt, and text descriptions. The paper provides a code link and proposes concrete benchmark refinements that could be useful to the community. However, the central quantitative claims depend on an apparently mis-defined evaluation metric and on benchmarks whose labels are authored by the same researchers without external validation or a public data release. Until these issues are resolved, the reported accuracy numbers and the claimed state of the art are not interpretable.

major comments (3)
  1. [§4.1, Eq. (1)] The 'macro accuracy' defined in Eq. (1) is not an accuracy measure; it is the mean per-class Jaccard index (intersection over union), because the denominator TP_c + FP_c + FN_c is the union of ground-truth and predicted positive images for class c and excludes true negatives. In a single-label multi-class setting, this quantity is far lower than conventional accuracy for the same predictions; for a balanced three-class problem with 84% per-class recall and precision, Eq. (1) would give about 0.72, and with realistic zero-shot precision it would be substantially lower. The values reported in Tables 2 and 3 (e.g., 84.0 and 87.6) are therefore not reproducible from the published formula. Please clarify the actual computed metric (e.g., macro recall or overall accuracy), correct the equation, or recompute every headline number. Since the abstract and Tables 1–3 all depend on this quantity, the central SOTA claim is currently uninterpretable.
  2. [§3.3, Tables 2–3, Figure 4] The MVTec-AC and VisA-AC benchmarks are central contributions, but their labels are the authors' manual corrections (36 corrected samples, merged classes, removed low-support classes) with no inter-annotator agreement, no external audit, and no public dataset link; the paper only provides a code repository. Moreover, the text prompts used for classification (Figure 4) encode the same class definitions and classification strategy that the authors used to assign the corrected labels. This creates a circularity risk: the LLM may be following the label-generation rules rather than independently recognizing visual anomalous patterns. Please release the datasets and a detailed annotation protocol, report inter-annotator statistics, and discuss how the evaluation avoids this circularity, for example by holding out a test set labeled by independent experts.
  3. [§4.3, Table 4] The anomaly-vs-defect experiment does not provide a valid measure of the claimed capability. The 'Normal' column is trivially 1.0 for all but one object category because normal images are filtered before the LLM; the random designation of 30% of anomaly classes as 'negligible' conflates class-level severity with instance-level acceptability; and the 89.8% mean accuracy aggregates this trivial component with the two-class anomaly/defect split. Please redesign the experiment with instance-level severity labels and a proper three-way evaluation, or clearly restrict the claim to the distinction between the two anomaly groups.
minor comments (6)
  1. [Abstract vs. §4.2.1, Table 1] The reported accuracy on MVTec-AD is 80.4% in the abstract and 81.4% in Section 4.2.1 and Table 1; please correct the inconsistency and verify the margin over Echo, since 81.4 − 72.9 = 8.5, not 9.5.
  2. [§3.3] The paper gives no public link for the MVTec-AC and VisA-AC datasets, only the code repository; please provide a dataset release to enable reproducibility and independent verification.
  3. [§4.2] The evaluation randomly selects a reference normal image from the training set, but the random seed is not reported; since the results may depend on this choice, please report seeds or average over multiple selections.
  4. [Table 1] F1-score and Cohen's kappa are reported for VELM but not for the baseline methods Echo and MCAD; please add these metrics or state why they are unavailable.
  5. [§3.3.1] Please specify which four 'combined' anomaly classes were excluded and list the exact per-object class merges, so that the benchmark construction is fully reproducible.
  6. [§5] The phrase 'first anomaly classification framework' is too strong, as prior work (MCAD, Echo) already addresses anomaly classification; please soften this claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: VELM's anomaly-class output comes from an off-the-shelf multimodal LLM with no task-specific training; the author-provided DDAD/PatchCore detectors are loaded as external tools, and the classification numbers are not forced by construction.

full rationale

The paper makes no parameter fitting that would make a reported number equivalent to its input. The classification step is performed by GPT-4o/GPT-4o-mini on images plus prompts; the vision expert only decides whether to invoke the LLM and provides a red-contour localization. None of the evaluation labels are used to fit weights or prompts, and ablations show the classification accuracy depends on the detector (Oracle 87.8, DDAD 84.0, PatchCore 78.1) rather than being fixed by any single input. The self-citations to DDAD and PatchCore are standard uses of prior detectors, not load-bearing uniqueness claims: the paper also evaluates with ground-truth masks and with GPT-4o-mini, so the claimed numbers do not reduce to a self-citation chain. The MVTec-AC/VisA-AC labels are author-created, but prompt definitions of anomaly classes are descriptions of class semantics, not per-image answer keys; without evidence that the prompts encode the specific corrected labels, no definitional equivalence can be exhibited. The main internal inconsistency is metric definition: Eq. (1) in Section 4.1 defines mean per-class Jaccard (TP/(TP+FP+FN)), not accuracy, yet the tables report the resulting numbers as 'Acc'. This makes the headline numbers hard to interpret and would be a correctness/reproducibility issue, but it is not a circular derivation of a prediction from its own input.

Assumptions & free parameters 3 free parameters · 4 assumptions · 2 invented entities

The central claim rests on a small number of hand choices: benchmark composition rules (which classes to merge or delete), the per-class text prompts that encode the decision rules, and the unexamined assumption that GPT-4o and the DDAD masks are reliable enough to ground those rules. There are no learned parameters or fitted equations, so the free-parameter count is low, but the load is carried by unverified human choices.

free parameters (3)
  • MVTec-AC class merge/exclusion policy = 4 merged pairs; toothbrush removed; 4 combined classes excluded
    The composition of the benchmark is chosen by the authors (Sec 3.3.1). Merging visually similar classes (poke/crack, cut/hole, thread side/thread top, broken teeth/rough) and removing categories changes the number of classes and the difficulty of the task; the reported 84% is defined on this hand-built label set.
  • VisA-AC minimum class size threshold = 10 samples
    Classes with fewer than 10 samples are removed (Sec 3.3.2). This threshold is chosen by hand and removes rare classes that would be hard to classify, biasing the benchmark toward easy classes.
  • Per-category LLM prompt contents = hand-written normal descriptions, anomaly definitions, classification strategies
    The text prompts in Figure 4 encode the classification rules (e.g., the 20% circumference rule for broken large vs broken small) and normal object knowledge. These prompts are not produced by an automatic procedure and are tuned to the benchmark classes; no validation set for prompt selection is described.
assumptions (4)
  • domain assumption GPT-4o can visually ground industrial close-up images well enough to map the supplied rules to pixels
    The whole classification stage depends on the closed API model's pretraining generalizing to MVTec/VisA test images (Sec 4.2). This is not verified with open weights or local inference.
  • domain assumption The vision expert (DDAD) localizes anomalies well enough that red-contour prompts help rather than mislead
    Section 3.1 assumes the detector's masks are useful; the VisA-AC results (69.6% vs 87.6% oracle) show this assumption is only partially met, yet the method still relies on it.
  • domain assumption The authors' manual corrections and merges of MVTec-AD and VisA labels are ground truth
    Sec 3.3 states 36 samples are corrected and classes are merged based on manual review, with no inter-annotator agreement or external audit. The evaluation inherits this assumption.
  • ad hoc to paper Equation (1) is a valid accuracy measure
    The paper defines 'macro accuracy' as TP/(TP+FP+FN) averaged, which is neither standard accuracy nor recall because it ignores true negatives. Treating this as accuracy is an unexamined modeling choice.
invented entities (2)
  • MVTec-AC
    purpose: Refined anomaly-class benchmark for evaluating anomaly classification on MVTec-AD images
    The labels are produced by the authors' manual review and the dataset is not linked in the paper; there is no external validation or release link to verify.
  • VisA-AC
    purpose: Refined anomaly-class benchmark for VisA images
    Same as MVTec-AC; classes with fewer than 10 samples are removed and similar classes merged, all by author judgment.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Detect, Classify, Act: Categorizing Industrial Anomalies with Multi-Modal Large Language Models." pith.science (2026). https://pith.science/paper/N44RVE35

@misc{pith2026250502626,
  author       = {Pith},
  title        = {Pith review of: Detect, Classify, Act: Categorizing Industrial Anomalies with Multi-Modal Large Language Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/N44RVE35}},
  note         = {Machine review of arXiv:2505.02626}
}
read the original abstract

Recent advances in visual industrial anomaly detection have demonstrated exceptional performance in identifying and segmenting anomalous regions while maintaining fast inference speeds. However, anomaly classification-distinguishing different types of anomalies-remains largely unexplored despite its critical importance in real-world inspection tasks. To address this gap, we propose VELM, a novel LLM-based pipeline for anomaly classification. Given the critical importance of inference speed, we first apply an unsupervised anomaly detection method as a vision expert to assess the normality of an observation. If an anomaly is detected, the LLM then classifies its type. A key challenge in developing and evaluating anomaly classification models is the lack of precise annotations of anomaly classes in existing datasets. To address this limitation, we introduce MVTec-AC and VisA-AC, refined versions of the widely used MVTec-AD and VisA datasets, which include accurate anomaly class labels for rigorous evaluation. Our approach achieves a state-of-the-art anomaly classification accuracy of 80.4% on MVTec-AD, exceeding the prior baselines by 5%, and 84% on MVTec-AC, demonstrating the effectiveness of VELM in understanding and categorizing anomalies. We hope our methodology and benchmark inspire further research in anomaly classification, helping bridge the gap between detection and comprehensive anomaly characterization.

Figures

Figures reproduced from arXiv: 2505.02626 by the authors.

Figure 1
Figure 1. Example of the proposed use case for our anomaly detec [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The figure illustrates four anomaly cases detected by a [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Overview of VELM. Given a query image, VELM first processes it using a Vision Expert, which performs both anomaly detection [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Example of a structured text prompt used for anomaly [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 6
Figure 6. Figure 6: Examples of broken teeth and rough anomaly classes in the MVTec-AD dataset. Despite their visual similarity, these anomalies are categorized into distinct classes, demonstrating the necessity for dataset refinement. gories (see [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

28 extracted references · 11 canonical work pages

  1. [1]

    Gpt-4 technical report

    Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ah- mad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774 ,

    arXiv

  2. [2]

    Qwen technical report

    Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei 8 Huang, et al. Qwen technical report. arXiv preprint arXiv:2309.16609, 2023. 3

    arXiv 2023

  3. [3]

    Efficien- tad: Accurate visual anomaly detection at millisecond-level latencies

    Kilian Batzner, Lars Heckler, and Rebecca K ¨onig. Efficien- tad: Accurate visual anomaly detection at millisecond-level latencies. In Proceedings of the IEEE/CVF Winter Confer- ence on Applications of Computer Vision , pages 128–138,

  4. [4]

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

    Paul Bergmann, Michael Fauser, David Sattlegger, and Carsten Steger. Mvtec ad–a comprehensive real-world dataset for unsupervised anomaly detection. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9592–9600, 2019. 3, 4

    work page 2019

  5. [5]

    Grounding everything: Emerging localiza- tion properties in vision-language transformers

    Walid Bousselham, Felix Petersen, Vittorio Ferrari, and Hilde Kuehne. Grounding everything: Emerging localiza- tion properties in vision-language transformers. In Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 3828–3837, 2024. 3

    2024

  6. [6]

    Clip-ad: A language-guided staged dual- path model for zero-shot anomaly detection

    Xuhai Chen, Jiangning Zhang, Guanzhong Tian, Haoyang He, Wuhao Zhang, Yabiao Wang, Chengjie Wang, Yunsheng Wu, and Yong Liu. Clip-ad: A language-guided staged dual- path model for zero-shot anomaly detection. arXiv preprint arXiv:2311.00453, 2023. 3

    arXiv 2023

  7. [7]

    Can multimodal large language models be guided to improve industrial anomaly detection? arXiv preprint arXiv:2501.15795, 2025

    Zhiling Chen, Hanning Chen, Mohsen Imani, and Farhad Imani. Can multimodal large language models be guided to improve industrial anomaly detection? arXiv preprint arXiv:2501.15795, 2025. 2, 3, 6

    arXiv 2025

  8. [8]

    Sub-image anomaly detec- tion with deep pyramid correspondences

    Niv Cohen and Yedid Hoshen. Sub-image anomaly detec- tion with deep pyramid correspondences. arXiv preprint arXiv:2005.02357, 2020. 2

    arXiv 2005

Show all 28 references
  1. [9]

    Padim: a patch distribution modeling framework for anomaly detection and localization

    Thomas Defard, Aleksandr Setkov, Angelique Loesch, and Romaric Audigier. Padim: a patch distribution modeling framework for anomaly detection and localization. In Inter- national conference on pattern recognition, pages 475–489. Springer, 2021

    2021

  2. [10]

    Anomaly detection via reverse distillation from one-class embedding

    Hanqiu Deng and Xingyu Li. Anomaly detection via reverse distillation from one-class embedding. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9737–9746, 2022. 2

    2022

  3. [11]

    Visual prompt engineering for medical vision language models in radiology

    Stefan Denner, Markus Bujotzek, Dimitrios Bounias, David Zimmerer, Raphael Stock, Paul F J ¨ager, and Klaus Maier- Hein. Visual prompt engineering for medical vision language models in radiology. arXiv preprint arXiv:2408.15802 ,

    arXiv

  4. [12]

    Diffusion for out-of-distribution detection on road scenes and beyond

    Silvio Galesso, Philipp Schr ¨oppel, Hssan Driss, and Thomas Brox. Diffusion for out-of-distribution detection on road scenes and beyond. In European Conference on Computer Vision, pages 110–126. Springer, 2024. 1

    2024

  5. [13]

    Anomalygpt: Detecting in- dustrial anomalies using large vision-language models

    Zhaopeng Gu, Bingke Zhu, Guibo Zhu, Yingying Chen, Ming Tang, and Jinqiao Wang. Anomalygpt: Detecting in- dustrial anomalies using large vision-language models. In Proceedings of the AAAI conference on artificial intelli- gence, pages 1932–1940, 2024. 3

    1932

  6. [14]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016. 2

    2016

  7. [15]

    Denoising dif- fusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising dif- fusion probabilistic models. Advances in neural information processing systems, 33:6840–6851, 2020. 2

    2020

  8. [16]

    Winclip: Zero- /few-shot anomaly classification and segmentation

    Jongheon Jeong, Yang Zou, Taewan Kim, Dongqing Zhang, Avinash Ravichandran, and Onkar Dabeer. Winclip: Zero- /few-shot anomaly classification and segmentation. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 19606–19616, 2023. 2, 3

    2023

  9. [17]

    Mmad: The first-ever comprehensive benchmark for mul- timodal large language models in industrial anomaly detec- tion

    Xi Jiang, Jian Li, Hanqiu Deng, Yong Liu, Bin-Bin Gao, Yifeng Zhou, Jialin Li, Chengjie Wang, and Feng Zheng. Mmad: The first-ever comprehensive benchmark for mul- timodal large language models in industrial anomaly detec- tion. arXiv preprint arXiv:2410.09453, 2024. 2, 3

    arXiv

  10. [18]

    Myriad: Large multimodal model by applying vi- sion experts for industrial anomaly detection

    Yuanze Li, Haolin Wang, Shihao Yuan, Ming Liu, Debin Zhao, Yiwen Guo, Chen Xu, Guangming Shi, and Wang- meng Zuo. Myriad: Large multimodal model by applying vi- sion experts for industrial anomaly detection. arXiv preprint arXiv:2310.19070, 2023. 3

    arXiv

  11. [19]

    Mcad: Multi- classification anomaly detection with relational knowledge distillation

    Zhuo Li, Yifei Ge, Xuebin Yue, and Lin Meng. Mcad: Multi- classification anomaly detection with relational knowledge distillation. Neural Computing and Applications , 36(23): 14543–14557, 2024. 3, 6

    2024

  12. [20]

    Anomaly detection with conditioned denoising diffusion models

    Arian Mousakhan, Thomas Brox, and Jawad Tayyub. Anomaly detection with conditioned denoising diffusion models. arXiv preprint arXiv:2305.15956 , 2023. 1, 2, 4, 6

    2023 arXiv

  13. [21]

    Henriques, and Fatma G¨uney

    Nazir Nayal, Mısra Yavuz, Jo ˜ao F. Henriques, and Fatma G¨uney. Rba: Segmenting unknown regions rejected by all. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023. 1

    2023

  14. [22]

    Learning transferable visual models from natural language supervi- sion

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervi- sion. In International conference on machine learning, ...

    2021

  15. [23]

    Towards to- tal recall in industrial anomaly detection

    Karsten Roth, Latha Pemula, Joaquin Zepeda, Bernhard Sch¨olkopf, Thomas Brox, and Peter Gehler. Towards to- tal recall in industrial anomaly detection. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 14318–14328, 2022. 1, 6

    2022

  16. [24]

    Score-based generative modeling through stochastic differential equa- tions

    Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Ab- hishek Kumar, Stefano Ermon, and Ben Poole. Score-based generative modeling through stochastic differential equa- tions. arXiv preprint arXiv:2011.13456, 2020. 2

    2011 arXiv

  17. [25]

    Diffusion models for medical anomaly detection

    Julia Wolleb, Florentin Bieder, Robin Sandk ¨uhler, and Philippe C Cattin. Diffusion models for medical anomaly detection. In International Conference on Medical image computing and computer-assisted intervention, pages 35–45. Springer, 2022. 1

    2022

  18. [26]

    Unsupervised surface anomaly detec- tion with diffusion probabilistic model

    Xinyi Zhang, Naiqi Li, Jiawei Li, Tao Dai, Yong Jiang, and Shu-Tao Xia. Unsupervised surface anomaly detec- tion with diffusion probabilistic model. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 6782–6791, 2023. 2

    2023

  19. [27]

    Anomalyclip: Object-agnostic prompt learn- 9 ing for zero-shot anomaly detection

    Qihang Zhou, Guansong Pang, Yu Tian, Shibo He, and Jiming Chen. Anomalyclip: Object-agnostic prompt learn- 9 ing for zero-shot anomaly detection. arXiv preprint arXiv:2310.18961, 2023. 3

    2023

  20. [28]

    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. In Eu- ropean Conference on Computer Vision , pages 392–408. Springer, 2022. 3, 5 10

    2022

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.