Pith. sign in

REVIEW 3 major objections 2 minor 1 cited by

DictAS: A Framework for Class-Generalizable Few-Shot Anomaly Segmentation via Dictionary Lookup

T0 review · 3 major / 2 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read DictAS claims a unified model can segment anomalies in unseen classes using only a few normal reference images as prompts.

desk verdict The paper is not the paper: the body is an unrelated stat.ME manuscript, so DictAS rests on an abstract alone — bounce it back, don't review it. read the letter →

arxiv 2508.13560 v2 pith:Y5FO4ENP submitted 2025-08-19 cs.CV

Zhen Qu , Xian Tao , Xinyi Gong , ShiChen Qu , Xiaopei Zhang , Xingang Wang , Fei Shen , Zhengtao Zhang
show 2 more authors
Mukesh Prasad Guiguang Ding
This is my paper
classification cs.CV
keywords few-shotanomalysegmentationdictionarylookupclass-generalizableself-supervisedlearningvisualpromptCLIPdetection
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

DictAS proposes that anomaly segmentation can be cast as a dictionary lookup task: build a dictionary from a few normal reference images, and label any query region that cannot be looked up as anomalous. The paper claims this yields a unified model that generalizes to completely unseen object classes without retraining, unlike prior few-shot anomaly segmentation methods that rely on supervised prompt learning with seen abnormal examples. The proposed framework adds two constraints, contrastive query and text alignment, during self-supervised training so that the lookup skill, rather than fixed normal/abnormal patterns, is transferred. If the claim holds, practitioners could deploy a single anomaly detector across new product lines or new medical image types by providing only a handful of normal examples as prompts. This reading pass only includes the abstract; the supplied full text is a different manuscript on Bayesian quantile regression, so the experimental support described in the abstract could not be checked.

What carries the argument

Dictionary Lookup with sparse retrieval: region features of the query image are matched against a dictionary built from a few normal reference images; a query that cannot be retrieved is flagged anomalous. The dictionary separates index and content, and the lookup is made sparse. Two auxiliary objectives, Contrastive Query Constraint and Text Alignment Constraint, regularize the learned feature space so transferable lookup, not memorization, is acquired.

What would settle it

Run DictAS on a held-out class it never saw during training, using only one or a few normal reference images, and compare pixel-wise anomaly detection against a simple nearest-normal-feature baseline that uses the same backbone. If DictAS does not beat that baseline by a clear margin, the claim of transferable dictionary lookup is not supported.

Watch

Extended reading notes

Core claim

The central discovery claimed is that dictionary lookup capabilities can be transferred to unseen-class anomaly segmentation via self-supervised learning. The paper's approach, DictAS, constructs a dictionary whose index and content are simulated from the normal reference images, performs sparse lookup for each query region feature, and treats retrieval failure as an anomaly. To sharpen discrimination, it introduces Contrastive Query Constraint and Text Alignment Constraint, which make abnormal features harder to retrieve and align normal features with text semantics. The abstract states that extensive experiments on seven public industrial and medical datasets show consistent state-of-the-a

Load-bearing premise

The claim that dictionary-lookup skills learned from normal references on training classes generalize to completely unseen classes without retraining is load-bearing; the abstract asserts it, and in this reading pass the supplied full text is an unrelated manuscript, so the supporting experiments are not available.

Editorial extensions

If this is right

  • A single model can be applied to a new industrial or medical category by supplying a few normal images as prompts, with no retraining on the target class.
  • Anomaly segmentation becomes a promptable task, analogous to prompting in other vision tasks: the reference images define 'normal' without requiring anomalous examples.
  • The sparse lookup design gives a natural interpretability handle: the dictionary content can show which normal prototype a region matched or failed to match.
  • The method promises practical deployment in settings where abnormal samples are rare, expensive, or undefined.
  • The abstract's claim, if verified, would let one anomaly detector serve many categories, reducing the per-category engineering typical of current FSAS systems.

Reading between the lines

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

  • The dictionary-lookup framing suggests a route to unify anomaly segmentation across industrial and medical domains with shared visual features; a natural test is whether the learned lookup skill transfers across modality boundaries (e.g., from industrial parts to medical scans).
  • A testable extension is to vary the number of reference images and measure the performance curve; if the method truly learns a lookup capability, performance should improve smoothly with more references and should never require anomalous examples.
  • The reliance on CLIP-style features implies that text prompts could further steer the definition of 'normal'; a direct experiment could use category-specific text descriptions as an auxiliary anchor to see whether segmentation accuracy improves.
  • Because the actual full text was not available in this reading pass, every experimental comparison and ablation mentioned in the abstract remains unverified from this manuscript snapshot.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 2 minor

Summary. The submission declares (Abstract) a framework, DictAS, for class-generalizable few-shot anomaly segmentation via self-supervised dictionary lookup, requiring no retraining on target data and only a few normal reference images as visual prompts. It claims state-of-the-art results on seven industrial and medical datasets. However, the full text of the submission is not the DictAS paper; it is arXiv:2508.13556, a Bayesian multinomial probit joint quantile regression manuscript. There is no description of Dictionary Construction, Dictionary Lookup, Query Discrimination Regularization, the datasets, baselines, or experimental protocol, and no derivations or numerical results pertaining to anomaly segmentation. Consequently the paper's central claims are entirely unsupported by the submitted evidence.

Significance. If the approach described in the abstract were fully instantiated and validated, it would be a meaningful advance: few-shot anomaly segmentation over unseen classes without target retraining, using only a handful of normal reference images, is an important capability, and a dictionary-lookup mechanism with self-supervised transfer is a plausible route. The claimed consistency across seven industrial and medical benchmarks would be an interesting empirical result. However, because the submitted manuscript contains none of the technical content, the significance cannot be assessed beyond the abstract, and I can credit no derivations, reproducibility artifacts, or falsifiable predictions from the body.

major comments (3)
  1. [Full Text vs. Abstract] The full text (pp. 1–15) is entirely a multinomial probit/quantile regression paper (arXiv:2508.13556) by different authors. It contains no anomaly segmentation content. The central claim of the abstract—that DictAS outperforms state-of-the-art FSAS methods on seven datasets—has no supporting experiments, tables, or figures in this submission. This is a load-bearing omission: no component (Dictionary Construction, Dictionary Lookup, Query Discrimination Regularization, Contrastive Query Constraint, Text Alignment Constraint) can be checked.
  2. [Abstract] The abstract does not provide any quantitative result, dataset names, or protocol; 'consistently outperforms state-of-the-art FSAS methods' is an unverifiable assertion. In a submitted manuscript, the body is expected to substantiate this. Since the body is unrelated, the claim is unsupported.
  3. [Full Text] The core premise—that dictionary lookup capabilities learned via self-supervised learning from normal reference images generalize to unseen classes—is load-bearing and requires empirical validation. The submitted text provides no description of the training procedure, network architecture, lookup mechanism, regularization losses, or evaluation protocol, so this premise cannot be tested. This is a missing-evidence problem rather than an identified internal inconsistency of the proposed method, but the absence of evidence is complete.
minor comments (2)
  1. [Title/Authorship] The title and author list implied by the abstract do not match the title and author list of the full text. If this is a packaging error, the correct manuscript must be provided; otherwise the submission is not a coherent paper.
  2. [References] The bibliography in the full text concerns econometrics and Bayesian statistics and is entirely irrelevant to the abstract. No reference is made to prior few-shot anomaly segmentation methods, CLIP-style vision-language models, or industrial/medical anomaly detection benchmarks.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity detected; supplied full text is an unrelated paper, so no derivation chain can be audited.

full rationale

The submitted full text is arXiv:2508.13556, a Bayesian multinomial probit quantile regression paper by Okabe et al., which has no connection to DictAS (arXiv:2508.13560). The only DictAS content available is the abstract, which describes a three-component framework (Dictionary Construction, Dictionary Lookup, Query Discrimination Regularization) for few-shot anomaly segmentation. The abstract contains no equations, no fitted parameters, no self-citations, and no derivation chain that could reduce to its own inputs. Without the method section, experiments, or theoretical derivations, there is no textual basis to claim that any 'prediction' is equivalent to its inputs by construction, nor that any load-bearing premise is justified solely by self-citation. The abstract's claim of cross-class generalization via self-supervised dictionary lookup is an empirical assertion that cannot be verified from the provided text, but absence of evidence is not circularity. Per the hard rules, circularity may only be flagged when the paper itself provides a specific reduction (Eq. X = Eq. Y by construction) or when a cited result is self-referential and load-bearing; neither condition is met here. Therefore the honest finding is no significant circularity, score 0.

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

Only the abstract is available; no free parameters or invented entities are mentioned. The axioms above are inferred from the described approach.

assumptions (2)
  • domain assumption Normal reference images contain sufficient feature patterns to form a dictionary for anomaly detection.
    Implicit in the abstract's description of dictionary construction from normal reference images; without this, the dictionary lookup cannot work.
  • domain assumption Self-supervised learning can transfer dictionary lookup capabilities to unseen classes.
    Stated as the core insight of DictAS; this is an empirical hypothesis that the abstract claims but the missing body does not support.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DictAS: A Framework for Class-Generalizable Few-Shot Anomaly Segmentation via Dictionary Lookup." pith.science (2026). https://pith.science/paper/Y5FO4ENP

@misc{pith2026250813560,
  author       = {Pith},
  title        = {Pith review of: DictAS: A Framework for Class-Generalizable Few-Shot Anomaly Segmentation via Dictionary Lookup},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Y5FO4ENP}},
  note         = {Machine review of arXiv:2508.13560}
}
read the original abstract

Recent vision-language models (e.g., CLIP) have demonstrated remarkable class-generalizable ability to unseen classes in few-shot anomaly segmentation (FSAS), leveraging supervised prompt learning or fine-tuning on seen classes. However, their cross-category generalization largely depends on prior knowledge of real seen anomaly samples. In this paper, we propose a novel framework, namely DictAS, which enables a unified model to detect visual anomalies in unseen object categories without any retraining on the target data, only employing a few normal reference images as visual prompts. The insight behind DictAS is to transfer dictionary lookup capabilities to the FSAS task for unseen classes via self-supervised learning, instead of merely memorizing the normal and abnormal feature patterns from the training set. Specifically, DictAS mainly consists of three components: (1) Dictionary Construction - to simulate the index and content of a real dictionary using features from normal reference images. (2) Dictionary Lookup - to retrieve queried region features from the dictionary via a sparse lookup strategy. When a query feature cannot be retrieved, it is classified as an anomaly. (3) Query Discrimination Regularization - to enhance anomaly discrimination by making abnormal features harder to retrieve from the dictionary. To achieve this, Contrastive Query Constraint and Text Alignment Constraint are further proposed. Extensive experiments on seven public industrial and medical datasets demonstrate that DictAS consistently outperforms state-of-the-art FSAS methods.

Discussion (0). Sign in 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. STAGE: Segmentation-oriented Industrial Anomaly Synthesis via Graded Diffusion with Explicit Mask Alignment

    cs.CV 2025-09 conditional novelty 6.0 of 10

    A graded diffusion framework with explicit mask alignment synthesizes realistic, mask-aligned industrial anomalies and improves downstream anomaly segmentation on MVTec and BTAD.

Reference graph

Works this paper leans on

44 extracted references · 35 canonical work pages · cited by 1 Pith paper

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

  2. [2]

    Medical image analysis using convolutional neural networks: a review

    Syed Muhammad Anwar, Muhammad Majid, Adnan Qayyum, Muhammad Awais, Majdi Alnowami, and Muhammad Khurram Khan. Medical image analysis using convolutional neural networks: a review. Journal of medical systems, 42: 0 1--13, 2018

    work page 2018

  3. [3]

    Fewsome: One-class few shot anomaly detection with siamese networks

    Niamh Belton, Misgina Tsighe Hagos, Aonghus Lawlor, and Kathleen M Curran. Fewsome: One-class few shot anomaly detection with siamese networks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2977--2986, 2023

    work page 2023

  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

    work page 2019

  5. [5]

    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. arXiv preprint arXiv:2112.09045, 2021

    arXiv 2021

  6. [6]

    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, pages 55--72. Springer, 2024

    work page 2024

  7. [7]

    Center-aware residual anomaly synthesis for multiclass industrial anomaly detection

    Qiyu Chen, Huiyuan Luo, Haiming Yao, Wei Luo, Zhen Qu, Chengkan Lv, and Zhengtao Zhang. Center-aware residual anomaly synthesis for multiclass industrial anomaly detection. IEEE Transactions on Industrial Informatics, pages 1--11, 2025

    work page 2025

  8. [8]

    April-gan: A zero-/few-shot anomaly classification and segmentation method for cvpr 2023 vand workshop challenge tracks 1&2: 1st place on zero-shot ad and 4th place on few-shot ad

    Xuhai Chen, Yue Han, and Jiangning Zhang. April-gan: A zero-/few-shot anomaly classification and segmentation method for cvpr 2023 vand workshop challenge tracks 1&2: 1st place on zero-shot ad and 4th place on few-shot ad. arXiv preprint arXiv:2305.17382, 2023

    arXiv 2023

Show all 44 references
  1. [9]

    Sub-image anomaly detection with deep pyramid correspondences

    Niv Cohen and Yedid Hoshen. Sub-image anomaly detection with deep pyramid correspondences. CoRR, abs/2005.02357, 2020

    2005 arXiv

  2. [10]

    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 International conference on pattern recognition, pages 475--489. Springer, 2021

    2021

  3. [11]

    The pascal visual object classes (voc) challenge

    Mark Everingham, Luc Van Gool, Christopher KI Williams, John Winn, and Andrew Zisserman. The pascal visual object classes (voc) challenge. International journal of computer vision, 88: 0 303--338, 2010

    2010

  4. [12]

    Fastrecon: Few-shot industrial anomaly detection via fast feature reconstruction

    Zheng Fang, Xiaoyang Wang, Haocheng Li, Jiejie Liu, Qiugui Hu, and Jimin Xiao. Fastrecon: Few-shot industrial anomaly detection via fast feature reconstruction. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 17481--17490, 2023

    2023

  5. [13]

    Metauas: Universal anomaly segmentation with one-prompt meta-learning

    Bin-Bin Gao. Metauas: Universal anomaly segmentation with one-prompt meta-learning. Advances in Neural Information Processing Systems, 37: 0 39812--39836, 2025

    2025

  6. [14]

    Anomalygpt: Detecting industrial anomalies using large vision-language models

    Zhaopeng Gu, Bingke Zhu, Guibo Zhu, Yingying Chen, Ming Tang, and Jinqiao Wang. Anomalygpt: Detecting industrial anomalies using large vision-language models. In Proceedings of the AAAI Conference on Artificial Intelligence, pages 1932--1940, 2024

    1932

  7. [15]

    Automated segmentation of macular edema in oct using deep neural networks

    Junjie Hu, Yuanyuan Chen, and Zhang Yi. Automated segmentation of macular edema in oct using deep neural networks. Medical image analysis, 55: 0 216--227, 2019

    2019

  8. [16]

    Registration based few-shot anomaly detection

    Chaoqin Huang, Haoyan Guan, Aofan Jiang, Ya Zhang, Michael Spratling, and Yan-Feng Wang. Registration based few-shot anomaly detection. In European Conference on Computer Vision, pages 303--319. Springer, 2022

    2022

  9. [17]

    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 Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 19606--19616, 2023

    2023

  10. [18]

    Deep learning-based defect detection of metal parts: evaluating current methods in complex conditions

    Stepan Jezek, Martin Jonak, Radim Burget, Pavel Dvorak, and Milos Skotak. Deep learning-based defect detection of metal parts: evaluating current methods in complex conditions. In 2021 13th International congress on ultra modern telecommunications and control systems and works...

    2021

  11. [19]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  12. [20]

    Deep learning

    Yann LeCun, Yoshua Bengio, and Geoffrey Hinton. Deep learning. nature, 521 0 (7553): 0 436--444, 2015

    2015

  13. [21]

    Promptad: Learning prompts with only normal samples for few-shot anomaly detection

    Xiaofan Li, Zhizhong Zhang, Xin Tan, Chengwei Chen, Yanyun Qu, Yuan Xie, and Lizhuang Ma. Promptad: Learning prompts with only normal samples for few-shot anomaly detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 16838--16848, 2024

    2024

  14. [22]

    Coft-ad: Contrastive fine-tuning for few-shot anomaly detection

    Jingyi Liao, Xun Xu, Manh Cuong Nguyen, Adam Goodge, and Chuan Sheng Foo. Coft-ad: Contrastive fine-tuning for few-shot anomaly detection. arXiv preprint arXiv:2402.18998, 2024

    2024 arXiv

  15. [23]

    Medical image classification using generalized zero shot learning

    Dwarikanath Mahapatra, Behzad Bozorgtabar, and Zongyuan Ge. Medical image classification using generalized zero shot learning. In Proceedings of the IEEE/CVF international conference on computer vision, pages 3344--3353, 2021

    2021

  16. [24]

    From softmax to sparsemax: A sparse model of attention and multi-label classification

    Andre Martins and Ramon Astudillo. From softmax to sparsemax: A sparse model of attention and multi-label classification. In International conference on machine learning, pages 1614--1623. PMLR, 2016

    2016

  17. [25]

    The multimodal brain tumor image segmentation benchmark (brats)

    Bjoern H Menze, Andras Jakab, Stefan Bauer, Jayashree Kalpathy-Cramer, Keyvan Farahani, Justin Kirby, Yuliya Burren, Nicole Porz, Johannes Slotboom, Roland Wiest, et al. The multimodal brain tumor image segmentation benchmark (brats). IEEE transactions on medical imaging, 34 0...

    1993

  18. [26]

    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. In 2021 IEEE 30th International Symposium on Industrial Electronics (ISIE), pages 01--06. IEEE, 2021

    2021

  19. [27]

    Lscad: A large-small model collaboration framework for unsupervised industrial anomaly detection

    Shichen Qu, Xian Tao, Xinyi Gong, Zhen Qu, Mukesh Prasad, Fei Shen, Zhengtao Zhang, and Guiguang Ding. Lscad: A large-small model collaboration framework for unsupervised industrial anomaly detection. IEEE Transactions on Instrumentation and Measurement, 2025 a

    2025

  20. [28]

    Investigating shift equivalence of convolutional neural networks in industrial defect segmentation

    Zhen Qu, Xian Tao, Fei Shen, Zhengtao Zhang, and Tao Li. Investigating shift equivalence of convolutional neural networks in industrial defect segmentation. IEEE Transactions on Instrumentation and Measurement, 72: 0 1--17, 2023

    2023

  21. [29]

    Vcp-clip: A visual context prompting model for zero-shot anomaly segmentation

    Zhen Qu, Xian Tao, Mukesh Prasad, Fei Shen, Zhengtao Zhang, Xinyi Gong, and Guiguang Ding. Vcp-clip: A visual context prompting model for zero-shot anomaly segmentation. In European Conference on Computer Vision, pages 301--317. Springer, 2024

    2024

  22. [30]

    Bayesian prompt flow learning for zero-shot anomaly detection

    Zhen Qu, Xian Tao, Xinyi Gong, Shichen Qu, Qiyu Chen, Zhengtao Zhang, Xingang Wang, and Guiguang Ding. Bayesian prompt flow learning for zero-shot anomaly detection. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 30398--30408, 2025 b

    2025

  23. [31]

    Learning transferable visual models from natural language supervision

    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 supervision. In International conference on machine learning, pa...

    2021

  24. [32]

    Towards total recall in industrial anomaly detection

    Karsten Roth, Latha Pemula, Joaquin Zepeda, Bernhard Sch \"o lkopf, Thomas Brox, and Peter Gehler. Towards total recall in industrial anomaly detection. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 14318--14328, 2022

    2022

  25. [33]

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

    Marco Rudolph, Bastian Wandt, and Bodo Rosenhahn. Same same but differnet: Semi-supervised defect detection with normalizing flows. In Proceedings of the IEEE/CVF winter conference on applications of computer vision, pages 1907--1916, 2021

    1907

  26. [34]

    Maeday: Mae for few-and zero-shot anomaly-detection

    Eli Schwartz, Assaf Arbelle, Leonid Karlinsky, Sivan Harary, Florian Scheidegger, Sivan Doveh, and Raja Giryes. Maeday: Mae for few-and zero-shot anomaly-detection. Computer Vision and Image Understanding, page 103958, 2024

    2024

  27. [35]

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

    Shelly Sheynin, Sagie Benaim, and Lior Wolf. A hierarchical transformation-discriminating generative model for few shot anomaly detection. In Proceedings of the IEEE/CVF international conference on computer vision, pages 8495--8504, 2021

    2021

  28. [36]

    Learning unsupervised metaformer for anomaly detection

    Jhih-Ciang Wu, Ding-Jie Chen, Chiou-Shann Fuh, and Tyng-Luh Liu. Learning unsupervised metaformer for anomaly detection. In Proceedings of the IEEE/CVF international conference on computer vision, pages 4369--4378, 2021 a

    2021

  29. [37]

    Learning unsupervised metaformer for anomaly detection

    Jhih-Ciang Wu, Ding-Jie Chen, Chiou-Shann Fuh, and Tyng-Luh Liu. Learning unsupervised metaformer for anomaly detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 4369--4378, 2021 b

    2021

  30. [38]

    Pushing the limits of fewshot anomaly detection in industry vision: Graphcore

    Guoyang Xie, Jinbao Wang, Jiaqi Liu, Feng Zheng, and Yaochu Jin. Pushing the limits of fewshot anomaly detection in industry vision: Graphcore. arXiv preprint arXiv:2301.12082, 2023

    2023 arXiv

  31. [39]

    Resad: A simple framework for class generalizable anomaly detection

    Xincheng Yao, Zixin Chen, Chao Gao, Guangtao Zhai, and Chongyang Zhang. Resad: A simple framework for class generalizable anomaly detection. Advances in Neural Information Processing Systems, 37: 0 125287--125311, 2025

    2025

  32. [40]

    Draem-a discriminatively trained reconstruction embedding for surface anomaly detection

    Vitjan Zavrtanik, Matej Kristan, and Danijel Sko c aj. Draem-a discriminatively trained reconstruction embedding for surface anomaly detection. In Proceedings of the IEEE/CVF international conference on computer vision, pages 8330--8339, 2021

    2021

  33. [41]

    Scene parsing through ade20k dataset

    Bolei Zhou, Hang Zhao, Xavier Puig, Sanja Fidler, Adela Barriuso, and Antonio Torralba. Scene parsing through ade20k dataset. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 633--641, 2017

    2017

  34. [42]

    Anomaly CLIP : Object-agnostic prompt learning for zero-shot anomaly detection

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

    2024

  35. [43]

    Toward generalist anomaly detection via in-context residual learning with few-shot sample prompts

    Jiawen Zhu and Guansong Pang. Toward generalist anomaly detection via in-context residual learning with few-shot sample prompts. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 17826--17836, 2024

    2024

  36. [44]

    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 European Conference on Computer Vision, pages 392--408. Springer, 2022

    2022

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.