Pith. sign in

REVIEW 3 major objections 5 minor 4 cited by

OmniAD: Detect and Understand Industrial Anomaly via Multimodal Reasoning

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read OmniAD unifies defect detection and explanation in a single vision-language model, reporting 79.1% average accuracy on MMAD as a 0-shot model.

desk verdict Solid detection results and a neat text-as-mask integration, but the MMAD "0-shot" score is contaminated by training on the benchmark itself. read the letter →

arxiv 2505.22039 v2 pith:CBOSOJFX submitted 2025-05-28 cs.CV

Shifang Zhao , Yiheng Lin , Lu Han , Yao Zhao , Yunchao Wei This is my paper
classification cs.CV
keywords industrialanomalydetectionunderstandingmultimodallargelanguagemodelsText-as-MaskencodingGRPOreinforcementlearningfew-shotgeneralizationMMADbenchmarkthreshold-free
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

OmniAD sets out to prove that one vision-language model can do both halves of industrial inspection: pinpoint a defect and then explain what it is, why it matters, and how it affects the object. The paper claims that encoding anomaly segmentation as text, rather than as pixel masks or bounding boxes, lets the model detect defects without hand-picked thresholds, and that coupling supervised fine-tuning with GRPO reinforcement learning makes this work from very few examples. On the MMAD anomaly-understanding benchmark, the model reports 79.1 average accuracy, above GPT-4o's 74.9 and well above its Qwen2.5-VL-7B base's 64.0. On four anomaly-detection datasets (MVTec-AD, VisA, MVTec-LOCO, GoodsAD), it reports image-level F1 scores in the high 80s to 90s with no threshold selection. If true, this makes defect understanding a natural sequel to defect detection rather than a separate expert pipeline.

What carries the argument

The load-bearing mechanism is Text-as-Mask Encoding: the image is cut into fixed patches (24×24 in the experiments), normal patches are dropped, and the remaining anomalous patches are compressed row-wise with run-length encoding into a short string of coordinates such as '(2,4), (3,2)-(3,4)'. The model produces this string inside segmentation tags, turning segmentation into next-token prediction and thereby removing any external threshold. That textual mask then feeds Visual Guided Textual Reasoning, where the model must emit the segmentation, then a thinking step, then the answer, so localization and explanation share one generation pass. The second mechanism is the SFT+GRPO training recipe with three rewards (format, F1-based detection accuracy, and answer accuracy), and the ablations show that only the combined recipe yields the large reported gains.

What would settle it

Retrain OmniAD with all MMAD images removed from both SFT and GRPO, keeping the other training images and the same rewards, and re-evaluate on MMAD: if the average falls from 79.1 toward the 64.0 of the Qwen2.5-VL-7B base, the claimed few-shot understanding gain is an artifact of benchmark examples in training rather than of the multimodal-reasoning mechanism.

Watch

Extended reading notes

Core claim

The paper's central claim is that anomaly understanding improves when detection is inside the language model instead of bolted on. OmniAD encodes the anomaly mask as a short run-length-encoded string of patch coordinates, so the model emits a segmentation string before it reasons, and then reasons with that segmentation in view; this visual guided textual reasoning is what lets the same 7B model both localize and explain. Training mixes SFT for format and domain knowledge with GRPO under three verifiable rewards, namely format, detection F1, and answer accuracy, and the paper reports that the combination is needed: SFT alone gives 69.8 average on MMAD, GRPO alone gives 55.7, and SFT+GRPO gives 79.1. The headline results are 79.1 average accuracy on MMAD in the 0-shot setting, 79.9 in the 1-shot setting, and an average image-level detection F1 of 92.2 in the 0-shot setting across the four detection datasets, all obtained without any threshold selection.

Load-bearing premise

The result depends on the assumption that including one example from every MMAD category in the training set does not materially teach the model the benchmark's question style, so that the 79.1 score is a fair zero-shot comparison against models that were not trained on MMAD.

Editorial extensions

If this is right

  • OmniAD performs anomaly segmentation and image-level detection in both 0-shot and 1-shot settings without threshold selection, reporting average image-level F1 of 92.2 (0-shot) and 92.0 (1-shot) across MVTec-AD, VisA, MVTec-LOCO, and GoodsAD.
  • Training only with SFT yields 69.8 average on MMAD and only with GRPO yields 55.7, while SFT+GRPO yields 79.1, so the paper's recipe, not either stage alone, carries the gain.
  • Ablations removing visual and textual reasoning drop average understanding from 79.1 to 68.0, with defect localization falling from 73.5 to 64.9, showing that emitting the mask before answering is what improves localization.
  • On structural and logical anomaly datasets MVTec-LOCO and GoodsAD, where CLIP-based baselines typically classify almost everything as normal, OmniAD reports image-level F1 around 90 to 93, indicating the text-as-mask approach generalizes beyond texture defects.
  • On MMAD, 0-shot OmniAD outperforms GPT-4o by 4.2 points and its own Qwen2.5-VL-7B base by 15.1 points, and it reaches or exceeds ordinary-human accuracy on Defect Analysis and Object Classification.

Reading between the lines

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

  • An untested consequence of the one-example-per-category MMAD data is that the 79.1 score may be a lower bound on what the method would achieve on genuinely unseen question formats; removing MMAD from training entirely would separate the mechanism's gain from benchmark-format leakage.
  • Because the mask is emitted as a compact text string, the same model could be extended to conversational inspection, asking follow-up questions about a defect and getting revised segmentation updates, without changing the architecture; the paper does not explore this.
  • The F1-based detection reward and the 24×24 patch grid are likely to interact with defect size; a testable extension is to measure how the patch count and scale factor affect localization of very small versus spread-out anomalies.
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 / 5 minor

Summary. The paper introduces OmniAD, a multimodal large language model framework that unifies industrial anomaly detection and anomaly understanding. Detection is performed via a Text-as-Mask encoding that converts anomaly segmentation into text generation, avoiding manual thresholds; understanding is performed through a Visual Guided Textual Reasoning step that produces a structured <seg>, <think>, <answer> output. The model is built on Qwen2.5-VL-7B and trained with an integrated SFT+GRPO pipeline using three reward functions. The authors report 79.1 average accuracy on the MMAD benchmark, surpassing GPT-4o (74.9) and the base Qwen2.5-VL-7B (64.0), and report strong threshold-free detection results on MVTec-AD, VisA, MVTec-LOCO, and GoodsAD. The training data includes one example per MMAD category in both SFT and GRPO stages, and the evaluation is performed on the full MMAD benchmark.

Significance. The proposed unification of detection and understanding in a single text-generation MLLM is a useful direction, and the Text-as-Mask encoding is a sensible adaptation of segmentation-as-text-generation to sparse, irregular anomaly masks. The detection results on the four external benchmarks, if reproducible, would be a meaningful step toward threshold-free few-shot anomaly detection. However, the headline understanding result on MMAD is compromised by the inclusion of MMAD training examples in the fine-tuning data, so the claimed superiority over GPT-4o is not currently established. The paper also gives no code, seeds, or error bars, which further limits confidence in the reported margins.

major comments (3)
  1. [Section 3.3, Table 2] The training set includes one example per MMAD category in both the SFT data (1.6K images) and the GRPO data (293 images), yet Table 2 reports OmniAD(0-shot) as 79.1 on the full MMAD benchmark. All comparison models, including GPT-4o and Qwen2.5-VL-7B, were not fine-tuned on MMAD. This makes the headline comparison non-zero-shot for OmniAD and potentially inflates the result by providing category-level priors about the question types, anomaly classes, and object categories in MMAD. The authors must provide a control experiment that removes all MMAD examples from training, or evaluates only on held-out MMAD categories/instances, and report the resulting accuracy. Without such a control, the claim that OmniAD surpasses GPT-4o by 4.2 points is not supported.
  2. [Section 4.1, Tables 3-5] All experimental results are reported as single runs with no code, random seeds, or error bars. The differences between OmniAD and baselines in Tables 3 and 4 are often small (e.g., 0.3-1.0 points in F1-score), and the improvement over AnomalyR1 in Table 2 is also within a plausible run-to-run variation for MLLM fine-tuning. Please report at least three seeds for the main MMAD result and the four detection benchmarks, or provide a significance test, so that the claimed improvements can be assessed statistically.
  3. [Section 3.2, Eq. (1)] The Detection Accuracy Reward defines a special case for normal images (FG = ∅), but the paper does not describe how the model is trained or encouraged to emit an empty <seg></seg> sequence for normal images, nor how empty predictions are parsed at inference. Since the Text-as-Mask encoding removes normal patches and relies on generating coordinates for anomaly patches only, the treatment of normal test images is critical for the threshold-free detection claim. Please clarify the training data construction for normal images and provide examples of empty-segmentation outputs and their reward assignment.
minor comments (5)
  1. [Table 2] The method name is misspelled as 'AnomlyR1'; it should be 'AnomalyR1'.
  2. [Tables 3-5 captions] The phrase 'Detection results are report as' should be 'Detection results are reported as'.
  3. [Section 3.3] The data preparation paragraph says 'one example per category from several industrial anomaly detection datasets' but does not list which datasets; please enumerate them.
  4. [Section 5, Limitations] The limitations paragraph mentions edge errors and inference efficiency but does not acknowledge the MMAD training/evaluation overlap; this should be disclosed and addressed.
  5. [Section 4.1] The scale factor alpha in Eq. (1) is set to 2, but no sensitivity analysis is provided. Please either provide a small ablation or state that the result is insensitive to this value.

Circularity Check

1 steps flagged · score 5.0 of 10

The headline MMAD 79.1 result is only partially independent because OmniAD was fine-tuned on one example per MMAD category before being evaluated on MMAD and labeled 0-shot.

  1. fitted input called prediction [Section 3.3, Data preparation; Table 2; Inference Setting]
    "Following the approach in [ 9], we add one example per category from MMAD to avoid domain gaps between datasets. For the GRPO training stage, reasoning annotations are not necessary, so we also use one example per category from MMAD."

    The paper's headline understanding result is the MMAD average accuracy: Table 2 reports 'OmniAD(0-shot) 7B ... 79.1'. But Section 3.3 shows that the training data includes one example per category from MMAD in both the SFT set (1.6K images) and the GRPO set (293 images). The paper defines '0-shot' only with respect to whether a reference normal image is provided at inference ('A 0-shot setting means only a query image is provided'), not with respect to whether the model was fine-tuned on MMAD. The comparison models, including GPT-4o and Qwen2.5-VL-7B, were not fine-tuned on MMAD examples. Thus the claimed 4.2-point advantage is not an independent zero-shot comparison: the model has been fitted to the evaluation benchmark's categories, question formats, and answer types.

full rationale

The central derivation chain for the anomaly-detection claims is self-contained: Text-as-Mask Encoding converts masks to text, the detection reward is computed from predicted and ground-truth masks, and the detection evaluations on MVTec-AD, VisA, MVTec-LOCO, and GoodsAD are external benchmarks not included in the training data. No load-bearing self-citation, imported uniqueness theorem, or ansatz-via-citation was found. The only substantial circularity concern is the MMAD understanding comparison: one example per MMAD category is deliberately added to both SFT and GRPO training, and the same MMAD benchmark is then used to report the headline 79.1 average accuracy and the claim of surpassing GPT-4o. Because the paper labels this row '0-shot' while the comparison models received no MMAD fine-tuning, the result is not fully independent. The inflation magnitude is not quantified, so this is not a complete reduction to a fitted value, but it is a partial circularity in the central claim. Score 5 reflects that the detection results remain external and valuable while the headline understanding comparison is compromised.

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

The experimental claims rest on the pretrained base model, the benchmark's validity, and hand-picked hyperparameters. The most consequential ledger entry is the use of MMAD categories in training, which is an assumption about acceptable benchmark usage.

free parameters (4)
  • alpha (Detection Accuracy Reward scale factor) = 2
    Scale factor in Eq. (1) to balance the F1-score reward against other rewards; chosen by hand, affects RL training.
  • Text-as-Mask patch grid = 24x24
    Image is divided into a 24x24 patch grid for encoding; chosen by hand, affects localization granularity.
  • Answer accuracy reward for incorrect answers = 0.1
    Incorrect answers receive a small positive reward of 0.1 to encourage exploration; set by hand.
  • Learning rates and batch size = SFT lr=1e-4, GRPO lr=1e-6, batch=16
    Standard hyperparameters chosen by hand; not central but listed for completeness.
assumptions (4)
  • domain assumption Qwen2.5-VL-7B provides a suitable base for anomaly detection and understanding after fine-tuning.
    The entire method builds on this pretrained model; if the base model's visual perception is insufficient, the results would not transfer.
  • domain assumption The MMAD benchmark accurately measures anomaly understanding and its categories align with the training data.
    The paper uses MMAD both as a source of one training example per category and as the evaluation benchmark, assuming this does not invalidate the comparison.
  • domain assumption F1-score is an appropriate reward signal for anomaly detection under class imbalance.
    Used in Eq. (1) as the detection accuracy reward; the paper assumes it is a good proxy for detection quality.
  • ad hoc to paper Text4Seg's text-generation segmentation paradigm can be extended to sparse, irregular anomaly masks by removing normal patches.
    This is a design choice specific to OmniAD; if the encoding loses too much spatial information, the unified reasoning fails.

how reviews work

0 comments
Cite this review

Pith. "Pith review of OmniAD: Detect and Understand Industrial Anomaly via Multimodal Reasoning." pith.science (2026). https://pith.science/paper/CBOSOJFX

@misc{pith2026250522039,
  author       = {Pith},
  title        = {Pith review of: OmniAD: Detect and Understand Industrial Anomaly via Multimodal Reasoning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CBOSOJFX}},
  note         = {Machine review of arXiv:2505.22039}
}
read the original abstract

While anomaly detection has made significant progress, generating detailed analyses that incorporate industrial knowledge remains a challenge. To address this gap, we introduce OmniAD, a novel framework that unifies anomaly detection and understanding for fine-grained analysis. OmniAD is a multimodal reasoner that combines visual and textual reasoning processes. The visual reasoning provides detailed inspection by leveraging Text-as-Mask Encoding to perform anomaly detection through text generation without manually selected thresholds. Following this, Visual Guided Textual Reasoning conducts comprehensive analysis by integrating visual perception. To enhance few-shot generalization, we employ an integrated training strategy that combines supervised fine-tuning (SFT) with reinforcement learning (GRPO), incorporating three sophisticated reward functions. Experimental results demonstrate that OmniAD achieves a performance of 79.1 on the MMAD benchmark, surpassing models such as Qwen2.5-VL-7B and GPT-4o. It also shows strong results across multiple anomaly detection benchmarks. These results highlight the importance of enhancing visual perception for effective reasoning in anomaly understanding. All codes and models will be publicly available.

Figures

Figures reproduced from arXiv: 2505.22039 by the authors.

Figure 1
Figure 1. OmniAD unifies anomaly detection and understanding through Multimodal Reasoning. It [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. An illustration of Text-as-Mask encoding process. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Qualitative results on MMAD [25]. The multimodal reasoning process facilitates accurate anomaly detection and question analysis. The ground truth mask and selected choice are indicated by the blue line. of 16 for full parameter fine-tuning. For SFT training, the initial learning rate is set to 1e-4, with a weight decay of 0.03. For GRPO training, the initial learning rate is set to 1e-6, the weight decay is 0.1, and… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Case Study of Multimodal Reasoning. Anomaly Detection. To evaluate the effective￾ness and potential of our multimodal reasoning, not only for anomaly understanding but also for accurate detection, we compare the anomaly de￾tection performance with six baseline models i…
Figure 5
Figure 5. Figure 5: An illustration of results in MVTec-AD dataset. [PITH_FULL_IMAGE:figures/full_fig_p017_5.png]
Figure 6
Figure 6. Figure 6: An illustration of results in VisA dataset. [PITH_FULL_IMAGE:figures/full_fig_p018_6.png]
Figure 7
Figure 7. Figure 7: An illustration of results in MVTec-LOCO dataset. [PITH_FULL_IMAGE:figures/full_fig_p019_7.png]
Figure 8
Figure 8. Figure 8: An illustration of results in GoodsAD dataset. [PITH_FULL_IMAGE:figures/full_fig_p020_8.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 4 Pith papers

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

  1. O-VAD: Industrial Video Anomaly Detection through Object-Centric Tracking and Reasoning

    cs.CV 2026-07 conditional novelty 6.0 of 10

    An object-centric, training-free agentic pipeline that tracks object state changes and reasons over them with a vision-language model achieves strong video-level AUROC on Phys-AD, LiquidAD, and IPAD, while producing i...

  2. Global Logic and Local Search: Dual-Stream Multimodal In-Context Learning for Verifiable Industrial Anomaly Detection

    cs.CV 2026-07 conditional novelty 6.0 of 10

    A training-free dual-stream multimodal framework (PVLA + SAM 3 global logic + MCTS local search) improves verifiable industrial anomaly QA without defective training samples.

  3. AD-FM: Multimodal LLMs for Anomaly Detection via Multi-Stage Reasoning and Fine-Grained Reward Optimization

    cs.CV 2025-08 conditional novelty 6.0 of 10

    AD-FM combines multi-stage reasoning with localization-aware rewards to fine-tune MLLMs for anomaly detection, improving average accuracy by about 22 percentage points over the base model.

  4. EMIT: Enhancing MLLMs for Industrial Anomaly Detection via Difficulty-Aware GRPO

    cs.CV 2025-07 conditional novelty 6.0 of 10

    A difficulty-aware GRPO training scheme with response resampling, advantage reweighting, GPT-generated text samples, and heatmap-guided contrastive embeddings improves InternVL3-8B by 7.77 percentage points on the MMA...

Reference graph

Works this paper leans on

59 extracted references · 19 canonical work pages · cited by 4 Pith papers

  1. [1]

    Anomalib: A deep learning library for anomaly detection

    Samet Akcay, Dick Ameln, Ashwin Vaidya, Barath Lakshmanan, Nilesh Ahuja, and Utku Genc. Anomalib: A deep learning library for anomaly detection. In2022 IEEE International Conference on Image Processing (ICIP), pages 1706–1710. IEEE, 2022

    work page

  2. [2]

    Cableinspect-ad: An expert-annotated anomaly detection dataset.Advances in Neural Information Processing Systems, 37:64703–64716, 2024

    Akshatha Arodi, Margaux Luck, Jean-Luc Bedwani, Aldo Zaimi, Ge Li, Nicolas Pouliot, Julien Beaudry, and Gaétan M Caron. Cableinspect-ad: An expert-annotated anomaly detection dataset.Advances in Neural Information Processing Systems, 37:64703–64716, 2024

    work page 2024

  3. [3]

    Shuai Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, Sibo Song, Kai Dang, Peng Wang, Shijie Wang, Jun Tang, et al. Qwen2. 5-vl technical report.arXiv preprint arXiv:2502.13923, 2025

    arXiv 2025

  4. [4]

    Beyond dents and scratches: Logical constraints in unsupervised anomaly detection and localization.International Journal of Computer Vision, 130(4):947–969, 2022

    Paul Bergmann, Kilian Batzner, Michael Fauser, David Sattlegger, and Carsten Steger. Beyond dents and scratches: Logical constraints in unsupervised anomaly detection and localization.International Journal of Computer Vision, 130(4):947–969, 2022

    work page 2022

  5. [5]

    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. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9592–9600, 2019

    2019

  6. [6]

    Segment any anomaly without training via hybrid prompt regularization.arXiv preprint arXiv:2305.10724, 2023

    Yunkang Cao, Xiaohao Xu, Chen Sun, Yuqi Cheng, Zongwei Du, Liang Gao, and Weiming Shen. Segment any anomaly without training via hybrid prompt regularization.arXiv preprint arXiv:2305.10724, 2023

    arXiv 2023

  7. [7]

    Towards generic anomaly detection and understanding: Large-scale visual-linguistic model (gpt-4v) takes the lead.arXiv preprint arXiv:2311.02782, 2023

    Yunkang Cao, Xiaohao Xu, Chen Sun, Xiaonan Huang, and Weiming Shen. Towards generic anomaly detection and understanding: Large-scale visual-linguistic model (gpt-4v) takes the lead.arXiv preprint arXiv:2311.02782, 2023

    arXiv 2023

  8. [8]

    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. InEuropean Conference on Computer Vision, pages 55–72. Springer, 2024

    2024

Show all 59 references
  1. [9]

    Anomalyr1: A grpo-based end-to-end mllm for industrial anomaly detection.arXiv preprint arXiv:2504.11914, 2025

    Yuhao Chao, Jie Liu, Jie Tang, and Gangshan Wu. Anomalyr1: A grpo-based end-to-end mllm for industrial anomaly detection.arXiv preprint arXiv:2504.11914, 2025

    2025 arXiv

  2. [10]

    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, and Yong Liu. Clip-ad: A language-guided staged dual-path model for zero-shot anomaly detection. InInternational Joint Conference on Artificial Intelligence, pages 17–33. Springer, 2024

    2024

  3. [11]

    Internvl: Scaling up vision foundation models and aligning for generic visual-linguistic tasks

    Zhe Chen, Jiannan Wu, Wenhai Wang, Weijie Su, Guo Chen, Sen Xing, Muyan Zhong, Qinglong Zhang, Xizhou Zhu, Lewei Lu, et al. Internvl: Scaling up vision foundation models and aligning for generic visual-linguistic tasks. InProceedings of the IEEE/CVF conference on computer visi...

    2024

  4. [12]

    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

    2025 arXiv

  5. [13]

    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. InInternational conference on pattern recognition, pages 475–489. Springer, 2021

    2021

  6. [14]

    Revitalizing reconstruction models for multi-class anomaly detection via class-aware contrastive learning.arXiv preprint arXiv:2412.04769, 2024

    Lei Fan, Junjie Huang, Donglin Di, Anyang Su, Maurice Pagnucco, and Yang Song. Revitalizing reconstruction models for multi-class anomaly detection via class-aware contrastive learning.arXiv preprint arXiv:2412.04769, 2024

    2024 arXiv

  7. [15]

    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. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 17481–17490, 2023

    2023

  8. [16]

    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. InProceedings of the AAAI conference on artificial intelligence, volume 38, pages 1932–1940, 2024. 10

    1932

  9. [17]

    Univad: A training-free unified model for few-shot visual anomaly detection.arXiv preprint arXiv:2412.03342, 2024

    Zhaopeng Gu, Bingke Zhu, Guibo Zhu, Yingying Chen, Ming Tang, and Jinqiao Wang. Univad: A training-free unified model for few-shot visual anomaly detection.arXiv preprint arXiv:2412.03342, 2024

    2024 arXiv

  10. [18]

    Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning.arXiv preprint arXiv:2501.12948, 2025

    Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, et al. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning.arXiv preprint arXiv:2501.12948, 2025

    2025 arXiv

  11. [19]

    Dinomaly: The less is more philosophy in multi-class unsupervised anomaly detection.arXiv preprint arXiv:2405.14325, 2024

    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 preprint arXiv:2405.14325, 2024

    2024 arXiv

  12. [20]

    Mambaad: Exploring state space models for multi-class unsupervised anomaly detection.arXiv preprint arXiv:2404.06564, 2024

    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.arXiv preprint arXiv:2404.06564, 2024

    2024 arXiv

  13. [21]

    Surface defect saliency of magnetic tile.The Visual Computer, 36(1):85–96, 2020

    Yibin Huang, Congying Qiu, and Kui Yuan. Surface defect saliency of magnetic tile.The Visual Computer, 36(1):85–96, 2020

    2020

  14. [22]

    Reconpatch: Contrastive patch representation learning for industrial anomaly detection

    Jeeho Hyun, Sangyun Kim, Giyoung Jeon, Seung Hwan Kim, Kyunghoon Bae, and Byung Jun Kang. Reconpatch: Contrastive patch representation learning for industrial anomaly detection. InProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 2052–2061, 2024

    2024

  15. [23]

    Openai o1 system card.arXiv preprint arXiv:2412.16720, 2024

    Aaron Jaech, Adam Kalai, Adam Lerer, Adam Richardson, Ahmed El-Kishky, Aiden Low, Alec Hel- yar, Aleksander Madry, Alex Beutel, Alex Carney, et al. Openai o1 system card.arXiv preprint arXiv:2412.16720, 2024

    2024 arXiv

  16. [24]

    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. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 19606–19616, 2023

    2023

  17. [25]

    MMAD: A comprehensive benchmark for multimodal large language models in industrial anomaly detection

    Xi Jiang, Jian Li, Hanqiu Deng, Yong Liu, Bin-Bin Gao, Yifeng Zhou, Jialin Li, Chengjie Wang, and Feng Zheng. MMAD: A comprehensive benchmark for multimodal large language models in industrial anomaly detection. InThe Thirteenth International Conference on Learning Representat...

    2025

  18. [26]

    Softpatch: Unsupervised anomaly detection with noisy data.Advances in Neural Information Processing Systems, 35:15433–15445, 2022

    Xi Jiang, Jianlin Liu, Jinbao Wang, Qiang Nie, Kai Wu, Yong Liu, Chengjie Wang, and Feng Zheng. Softpatch: Unsupervised anomaly detection with noisy data.Advances in Neural Information Processing Systems, 35:15433–15445, 2022

    2022

  19. [27]

    Fabgpt: An efficient large multimodal model for complex wafer defect knowledge queries

    Yuqi Jiang, Xudong Lu, Qian Jin, Qi Sun, Hanming Wu, and Cheng Zhuo. Fabgpt: An efficient large multimodal model for complex wafer defect knowledge queries. InProceedings of the 43rd IEEE/ACM International Conference on Computer-Aided Design, pages 1–8, 2024

    2024

  20. [28]

    Logicad: Explainable anomaly detection via vlm-based text feature extraction

    Er Jin, Qihui Feng, Yongli Mou, Gerhard Lakemeyer, Stefan Decker, Oliver Simons, and Johannes Stegmaier. Logicad: Explainable anomaly detection via vlm-based text feature extraction. InProceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 4129–4137, 2025

    2025

  21. [29]

    Segment anything

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer Whitehead, Alexander C Berg, Wan-Yen Lo, et al. Segment anything. InProceedings of the IEEE/CVF international conference on computer vision, pages 4015–4026, 2023

    2023

  22. [30]

    Text4seg: Reimagining image segmentation as text generation.arXiv preprint arXiv:2410.09855, 2024

    Mengcheng Lan, Chaofeng Chen, Yue Zhou, Jiaxing Xu, Yiping Ke, Xinjiang Wang, Litong Feng, and Wayne Zhang. Text4seg: Reimagining image segmentation as text generation.arXiv preprint arXiv:2410.09855, 2024

    2024 arXiv

  23. [31]

    Llava-onevision: Easy visual task transfer.arXiv preprint arXiv:2408.03326, 2024

    Bo Li, Yuanhan Zhang, Dong Guo, Renrui Zhang, Feng Li, Hao Zhang, Kaichen Zhang, Yanwei Li, Ziwei Liu, and Chunyuan Li. Llava-onevision: Easy visual task transfer.arXiv preprint arXiv:2408.03326, 2024

    2024 arXiv

  24. [32]

    Lad-reasoner: Tiny multimodal models are good reasoners for logical anomaly detection.arXiv preprint arXiv:2504.12749, 2025

    Weijia Li, Guanglei Chu, Jiong Chen, Guo-Sen Xie, Caifeng Shan, and Fang Zhao. Lad-reasoner: Tiny multimodal models are good reasoners for logical anomaly detection.arXiv preprint arXiv:2504.12749, 2025

    2025 arXiv

  25. [33]

    Myriad: Large multimodal model by applying vision experts for industrial anomaly detection.arXiv preprint arXiv:2310.19070, 2023

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

    arXiv

  26. [34]

    Triad: Empowering lmm-based anomaly detection with vision expert-guided visual tokenizer and manufacturing process.arXiv preprint arXiv:2503.13184, 2025

    Yuanze Li, Shihao Yuan, Haolin Wang, Qizhang Li, Ming Liu, Chen Xu, Guangming Shi, and Wangmeng Zuo. Triad: Empowering lmm-based anomaly detection with vision expert-guided visual tokenizer and manufacturing process.arXiv preprint arXiv:2503.13184, 2025. 11

    2025 arXiv

  27. [35]

    Improved baselines with visual instruction tuning

    Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. Improved baselines with visual instruction tuning. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 26296–26306, 2024

    2024

  28. [36]

    Llava-next: Improved reasoning, ocr, and world knowledge, January 2024

    Haotian Liu, Chunyuan Li, Yuheng Li, Bo Li, Yuanhan Zhang, Sheng Shen, and Yong Jae Lee. Llava-next: Improved reasoning, ocr, and world knowledge, January 2024

    2024

  29. [37]

    Seg-zero: Reasoning- chain guided segmentation via cognitive reinforcement.arXiv preprint arXiv:2503.06520, 2025

    Yuqi Liu, Bohao Peng, Zhisheng Zhong, Zihao Yue, Fanbin Lu, Bei Yu, and Jiaya Jia. Seg-zero: Reasoning- chain guided segmentation via cognitive reinforcement.arXiv preprint arXiv:2503.06520, 2025

    2025 arXiv

  30. [38]

    Visual-rft: Visual reinforcement fine-tuning.arXiv preprint arXiv:2503.01785, 2025

    Ziyu Liu, Zeyi Sun, Yuhang Zang, Xiaoyi Dong, Yuhang Cao, Haodong Duan, Dahua Lin, and Jiaqi Wang. Visual-rft: Visual reinforcement fine-tuning.arXiv preprint arXiv:2503.01785, 2025

    2025 arXiv

  31. [39]

    Aa-clip: Enhancing zero-shot anomaly detection via anomaly-aware clip.arXiv preprint arXiv:2503.06661, 2025

    Wenxin Ma, Xu Zhang, Qingsong Yao, Fenghe Tang, Chenxu Wu, Yingtai Li, Rui Yan, Zihang Jiang, and S Kevin Zhou. Aa-clip: Enhancing zero-shot anomaly detection via anomaly-aware clip.arXiv preprint arXiv:2503.06661, 2025

    2025 arXiv

  32. [40]

    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. InInternational conference on machine learning, pag...

    2021

  33. [41]

    Deepspeed: System optimizations enable training deep learning models with over 100 billion parameters

    Jeff Rasley, Samyam Rajbhandari, Olatunji Ruwase, and Yuxiong He. Deepspeed: System optimizations enable training deep learning models with over 100 billion parameters. InProceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining, pages 35...

    2020

  34. [42]

    Towards total recall in industrial anomaly detection

    Karsten Roth, Latha Pemula, Joaquin Zepeda, Bernhard Schölkopf, Thomas Brox, and Peter Gehler. Towards total recall in industrial anomaly detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 14318–14328, 2022

    2022

  35. [43]

    Deepseekmath: Pushing the limits of mathematical reasoning in open language models.arXiv preprint arXiv:2402.03300, 2024

    Zhihong Shao, Peiyi Wang, Qihao Zhu, Runxin Xu, Junxiao Song, Xiao Bi, Haowei Zhang, Mingchuan Zhang, YK Li, Y Wu, et al. Deepseekmath: Pushing the limits of mathematical reasoning in open language models.arXiv preprint arXiv:2402.03300, 2024

    2024 arXiv

  36. [44]

    Vlm-r1: A stable and generalizable r1-style large vision-language model.arXiv preprint arXiv:2504.07615, 2025

    Haozhan Shen, Peng Liu, Jingcheng Li, Chunxin Fang, Yibo Ma, Jiajia Liao, Qiaoli Shen, Zilun Zhang, Kangjia Zhao, Qianqian Zhang, et al. Vlm-r1: A stable and generalizable r1-style large vision-language model.arXiv preprint arXiv:2504.07615, 2025

    2025 arXiv

  37. [45]

    Expanding rl with verifiable rewards across diverse domains.arXiv preprint arXiv:2503.23829, 2025

    Yi Su, Dian Yu, Linfeng Song, Juntao Li, Haitao Mi, Zhaopeng Tu, Min Zhang, and Dong Yu. Expanding rl with verifiable rewards across diverse domains.arXiv preprint arXiv:2503.23829, 2025

    2025 arXiv

  38. [46]

    Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context.arXiv preprint arXiv:2403.05530, 2024

    Gemini Team, Petko Georgiev, Ving Ian Lei, Ryan Burnell, Libin Bai, Anmol Gulati, Garrett Tanzer, Damien Vincent, Zhufeng Pan, Shibo Wang, et al. Gemini 1.5: Unlocking multimodal understanding across millions of tokens of context.arXiv preprint arXiv:2403.05530, 2024

    2024 arXiv

  39. [47]

    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. InProceedings of the IEEE/CVF Conference on Compute...

    2024

  40. [48]

    3cad: A large-scale real-world 3c product dataset for unsupervised anomaly.arXiv preprint arXiv:2502.05761, 2025

    Enquan Yang, Peng Xing, Hanyang Sun, Wenbo Guo, Yuanwei Ma, Zechao Li, and Dan Zeng. 3cad: A large-scale real-world 3c product dataset for unsupervised anomaly.arXiv preprint arXiv:2502.05761, 2025

    2025 arXiv

  41. [49]

    Defect spectrum: a granular look of large-scale defect datasets with rich semantics

    Shuai Yang, Zhifei Chen, Pengguang Chen, Xi Fang, Yixun Liang, Shu Liu, and Yingcong Chen. Defect spectrum: a granular look of large-scale defect datasets with rich semantics. InEuropean Conference on Computer Vision, pages 187–203. Springer, 2024

    2024

  42. [50]

    Minicpm-v: A gpt-4v level mllm on your phone.arXiv preprint arXiv:2408.01800, 2024

    Yuan Yao, Tianyu Yu, Ao Zhang, Chongyi Wang, Junbo Cui, Hongji Zhu, Tianchi Cai, Haoyu Li, Weilin Zhao, Zhihui He, et al. Minicpm-v: A gpt-4v level mllm on your phone.arXiv preprint arXiv:2408.01800, 2024

    2024 arXiv

  43. [51]

    Draem-a discriminatively trained reconstruction embedding for surface anomaly detection

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

    2021

  44. [52]

    Diffusionad: Norm-guided one-step denoising diffusion for anomaly detection.arXiv preprint arXiv:2303.08730, 2023

    Hui Zhang, Zheng Wang, Zuxuan Wu, and Yu-Gang Jiang. Diffusionad: Norm-guided one-step denoising diffusion for anomaly detection.arXiv preprint arXiv:2303.08730, 2023. 12

    2023 arXiv

  45. [53]

    Pku-goodsad: A supermarket goods dataset for unsupervised anomaly detection and segmentation.IEEE Robotics and Automation Letters, 9(3):2008– 2015, 2024

    Jian Zhang, Runwei Ding, Miaoju Ban, and Linhui Dai. Pku-goodsad: A supermarket goods dataset for unsupervised anomaly detection and segmentation.IEEE Robotics and Automation Letters, 9(3):2008– 2015, 2024

    2008

  46. [54]

    Realnet: A feature selection network with realistic synthetic anomaly for anomaly detection

    Ximiao Zhang, Min Xu, and Xiuzhuang Zhou. Realnet: A feature selection network with realistic synthetic anomaly for anomaly detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 16699–16708, 2024

    2024

  47. [55]

    Logicode: an llm-driven framework for logical anomaly detection.IEEE Transactions on Automation Science and Engineering, 2024

    Yiheng Zhang, Yunkang Cao, Xiaohao Xu, and Weiming Shen. Logicode: an llm-driven framework for logical anomaly detection.IEEE Transactions on Automation Science and Engineering, 2024

    2024

  48. [56]

    Industrial anomaly detection with domain shift: A real-world dataset and masked multi-scale reconstruction.Computers in Industry, 151:103990, 2023

    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

  49. [57]

    Anomalyclip: Object-agnostic prompt learning for zero-shot anomaly detection.arXiv preprint arXiv:2310.18961, 2023

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

    2023

  50. [58]

    Do llms understand visual anomalies? uncovering llm’s capabilities in zero-shot anomaly detection

    Jiaqi Zhu, Shaofeng Cai, Fang Deng, Beng Chin Ooi, and Junran Wu. Do llms understand visual anomalies? uncovering llm’s capabilities in zero-shot anomaly detection. InProceedings of the 32nd ACM International Conference on Multimedia, pages 48–57, 2024

    2024

  51. [59]

    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. 13 A Effects of Expert Model. Table 8: Pe...

    2022

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.