Pith. sign in

REVIEW 5 major objections 6 minor 2 cited by

AnomalyControl: Learning Cross-modal Semantic Features for Controllable Anomaly Synthesis

T0 review · 5 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read AnomalyControl uses cross-modal semantic features from a text-image reference to generate realistic, controllable synthetic anomalies without retraining the diffusion model.

desk verdict Solid method paper with a useful new prompt formulation, but the SOTA claim rests on unreplicated single-run numbers and a Tg chosen on the test metrics. read the letter →

arxiv 2412.06510 v4 pith:FDOCL6CL submitted 2024-12-09 cs.CV cs.AI

Shidan He , Lei Liu , Xiujun Shu , Bo Wang , Yuanhao Feng , Shen Zhao This is my paper · ORCID
classification cs.CVcs.AI
keywords anomalysynthesiscross-modalsemanticfeaturescontrollablegenerationdiffusionmodeldetectionvision-languageattentionguidancedataaugmentation
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

AnomalyControl sets out to prove that synthetic defects for industrial and medical anomaly inspection can be generated with enough realism and control to stand in for real abnormal data. It does this by extracting cross-modal semantic features from a non-matching text-image reference prompt — a crop of a defect plus a short phrase such as 'red stains' — using a frozen vision-language model, and injecting those features into a frozen Stable Diffusion base through a lightweight adapter. On the MVTec AD benchmark the framework reports the highest generation-quality scores among anomaly-synthesis methods (Inception Score 1.84, and IC-LPIPS 0.35, a perceptual similarity-and-generalization metric) and improves downstream pixel-level anomaly-localization AUC to 99.5 and AP to 87.4. If the claim holds, factories and clinics short on defect data could generate training samples on demand, with the anomaly type, location, and surface controlled by the user.

What carries the argument

The carrying mechanism is the cross-modal semantic feature $F_c$, produced by a frozen vision-language model from the text-image reference prompt. ASEA extracts the anomaly-relevant part of the VLM's attention by averaging the cross-attention map over the anomaly tokens, masking it to the anomaly region $M_A$, and optimizing a trainable guidance variable $e_g$ for a few steps to minimize an energy function that penalizes attention outside the mask. SGA then injects $F_c$ into Stable Diffusion using decoupled cross-attention: for each of the 16 cross-attention layers, two trainable projection matrices ($W'_k$, $W'_v$) are added to condition generation on the cross-modal features, while the base model stays frozen. Because the VLM is frozen and the features depend only on the reference prompt, they can be cached, which makes the added computation during training small.

What would settle it

Generate the same anomaly type (e.g., a scratch) on a metal surface twice, once with a reference image of a scratch on leather and once with a reference image of a scratch on the same metal. If the cross-material reference yields visibly worse synthesis or makes downstream metal-defect detection drop, the claimed generalization of cross-modal semantic features is not holding; the comparison can be scored with the same IS, IC-LPIPS, and detection metrics used in the paper.

Watch

Extended reading notes

Core claim

The paper's central claim is that a non-matching text-image reference prompt — a visual anomaly descriptor (a magnified crop of the defect) and a textual anomaly descriptor (e.g., 'This is an image with red stains') that need not match the target surface or material — supplies enough cross-modal semantic information to guide a frozen diffusion model toward high-fidelity, generalizable anomaly synthesis. AnomalyControl implements this with three modules: Cross-modal Semantic Modeling (CSM) fuses the visual and textual descriptors through a frozen BLIP-2 vision-language model; Anomaly-Semantic Enhanced Attention (ASEA) refines the VLM's cross-attention map by iteratively optimizing a trainable guidance variable so that attention concentrates on the anomaly region; and the Semantic Guided Adapter (SGA) routes the resulting cross-modal features $F_c$ into Stable Diffusion through a decoupled cross-attention branch with only two trainable projection matrices per layer. The framework reports state-of-the-art generation quality on MVTec AD (IS 1.84, IC-LPIPS 0.35) and the best downstream anomaly-localization results among compared methods (pixel-level AUC 99.5, AP 87.4), and the qualitative figures show anomaly types transferring across wood, leather, metal, and other surfaces.

Load-bearing premise

The load-bearing premise is that cross-modal semantic features extracted by a frozen vision-language model from a small defect crop and a short phrase encode the anomaly type itself, rather than the specific surface, material, or background of the reference image.

Editorial extensions

If this is right

  • Anomaly-free manufacturing lines can create pixel-aligned defect samples for any new product category on demand, converting unsupervised anomaly detection into a supervised task without collecting real defective items.
  • Synthetic training data produced by AnomalyControl improve downstream localization performance beyond the best previous anomaly-synthesis baselines: pixel-level AUC reaches 99.5 and AP 87.4 on MVTec AD.
  • A single anomaly reference (image plus phrase) can synthesize that anomaly type across different objects and materials seen only at test time, enabling out-of-distribution defect generation beyond the training set.
  • Because only two projection matrices per cross-attention layer are trained and the VLM features are cacheable, the method can be adapted to new anomaly types with small training cost and without retraining the diffusion backbone.

Reading between the lines

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

  • The same cross-modal transfer idea could be applied to medical imaging, where a single radiograph of a pathology (e.g., a fracture) could be turned into controllable synthetic training examples on unseen anatomies, easing the chronic shortage of annotated abnormal medical images.
  • The ASEA refinement loop (three gradient steps per sample) could be replaced by a small feed-forward attention predictor that regresses the energy-minimizing attention map directly, removing the per-prompt optimization and further reducing inference latency.
  • Because the guidance strength $\gamma$ and the number of guidance steps $T_g$ are free hyperparameters, an adaptive scheme that increases $T_g$ for high-entropy attention maps could push realism further on hard categories such as grid and screw, where the reported scores are lowest.
  • A controlled evaluation that varies only the reference surface (same anomaly type on many materials) would isolate how much of the generalization gain comes from cross-modal features versus from the frozen diffusion model's own priors.
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

5 major / 6 minor

Summary. The paper proposes AnomalyControl, a diffusion-based anomaly synthesis framework that augments Stable Diffusion with a Cross-modal Semantic Modeling (CSM) module, an Anomaly-Semantic Enhanced Attention (ASEA) mechanism, and a Semantic Guided Adapter (SGA). CSM uses a frozen BLIP-2 VLM to extract cross-modal features from a non-matching text-image reference pair (a visual anomaly descriptor and a textual anomaly descriptor), ASEA refines the VLM attention toward the anomaly region via a trainable guidance variable, and SGA injects the resulting features into the denoising U-Net through a decoupled cross-attention layer. The authors report state-of-the-art results on MVTec AD, MPDD, and ViSA for generation quality (IS, IC-LPIPS) and downstream anomaly detection/localization (AUC, AP, F1), and they provide qualitative demonstrations of controllable and cross-material anomaly synthesis.

Significance. If the empirical claims hold, the work would provide a practical plug-and-play adapter for controllable anomaly synthesis, building on frozen diffusion and vision-language models, with released code and a caching mechanism for efficiency. The use of non-matching text-image reference prompts is a useful idea for separating anomaly type from surface appearance. However, at present the headline state-of-the-art claim rests on single-run evaluations without statistical uncertainty, on hyperparameters selected using the same test metrics as the final comparison, and on comparisons that omit a closely related multi-modal baseline. These issues must be addressed before the significance claim is trustworthy.

major comments (5)
  1. [Table 4 and Experimental Setup] The ablation in Table 4 is used to set Tg=3 by selecting the row with the best IS, IL, AUC-P, and AUC-I, and these same metrics are subsequently reported in Tables 2 and 3 as the main evidence for the state-of-the-art claim. Because the test metrics are used for hyperparameter selection, the reported numbers constitute test-set tuning. Please provide a validation split for selecting Tg, or report the full range of Tg in the main comparisons and show that the conclusions are insensitive to this choice.
  2. [Tables 2, 3, and 5] All quantitative results are reported as single runs without error bars, confidence intervals, or seed counts. The differences over AnoDiff are small in several key metrics (e.g., MVTec AD IS 1.84 vs 1.80; AUC-I 99.3 vs 99.2), so with one seed the comparison cannot be distinguished from noise. Please re-run the experiments with at least three seeds, report mean and standard deviation, and perform significance tests (e.g., paired tests) where appropriate.
  3. [Experimental Results, Comparisons] AnomalyXFusion (Hu et al. 2024a) is cited in the introduction and related work as a multi-modal anomaly synthesis method, and the MVTec AD captions used in the paper come from this work, yet AnomalyXFusion is absent from all comparisons in Tables 2, 3, and 5. This omission is load-bearing for the state-of-the-art claim. Please add AnomalyXFusion under the same protocol, or provide a clear justification for its exclusion.
  4. [Cross-modal Semantic Modeling and Figure 8] The central generalization claim—that cross-modal semantic features transfer anomaly patterns across different materials and object surfaces—is supported only by qualitative examples in Figure 8. This leaves open the possibility that the VLM features are surface-specific rather than anomaly-type-specific. Please provide a quantitative evaluation of cross-material transfer, for instance by using a reference prompt from one material to synthesize anomalies on another material and measuring downstream detection performance or perceptual similarity, so that the generalization claim is grounded in quantitative evidence.
  5. [Metrics, Experimental Setup] The primary generation metrics, Inception Score (IS) and IC-LPIPS, are weak proxies for the realism of synthesized anomalies: IS reflects a pre-trained classifier's label distribution rather than photorealism, and IC-LPIPS can be inflated by intra-cluster repetition or memorization. Because the paper's claim of 'realism' relies heavily on these metrics, please supplement them with a human evaluation or additional measures such as FID on the generated set, and report how these correlate with downstream detection performance.
minor comments (6)
  1. [Table 5] In the ViSA row, the IL (IC-LPIPS) values for AnoDiff (1.73) and Ours (2.12) appear inconsistent with the usual [0,1] range of LPIPS-based similarity metrics; please clarify the scaling or correct the values.
  2. [References] Ojha et al. appears twice in the reference list, as 2021a and 2021b, with identical titles; these should be merged or properly distinguished.
  3. [Table 1] The venue label 'W ACV'21' in Table 1 should be 'WACV'21'.
  4. [Methods, Overall Framework and Figure 2] The text says 'Our framework consists of two main modules: CSM and SGA,' but Figure 2 and the abstract describe three modules (CSM, ASEA, SGA). Clarify that ASEA is an inner mechanism of CSM rather than a separate module.
  5. [Equation (6)] Equation (6) contains a stray comma in 'concat(zt, MA, zm, )'; please remove it.
  6. [Appendix B, Limitation] The limitation that some anomalies are difficult to describe textually, and that image references partially compensate, appears only in Appendix B; consider stating this limitation in the main paper's discussion.

Circularity Check

0 steps flagged · score 2.0 of 10

No constructional circularity: the cross-modal features are extracted from the supplied reference prompts, and the SOTA claim is benchmarked, not defined by the inputs; only a minor test-set hyperparameter-selection caveat remains.

full rationale

The paper's derivation chain is not circular under the enumerated patterns. CSM extracts cross-modal semantic features Fc from the user-supplied visual and textual anomaly descriptors (Ia and Ta) using a frozen VLM plus attention guidance (Eqs. 2-4); ASEA only steers attention toward the mask; SGA is trained with the denoising objective in Eq. 6 to map Fc and the targeted text prompt into the target image. The reported IS, IL, and downstream AUROC/AP values are evaluations against held-out test anomalies, not quantities algebraically equal to the model's inputs. The AnoDiff protocol and mask-generation method are cited as external prior work (Hu et al. 2024b) with no author overlap, so the self-citation and uniqueness-import patterns do not apply. The one legitimate concern is that Tg=3 is selected from Table 4 using the same metrics later used for the headline comparison; this is test-set hyperparameter tuning rather than constructional circularity, because the final numbers are not forced to equal the ablation rows by any equation. For these reasons, no specific circular step can be quoted.

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

The method stacks several pre-existing components: Stable Diffusion 1.5, BLIP-2, IP-Adapter-style decoupled cross-attention, and cross-attention guidance energy from Chen et al. 2024. The hand-set hyperparameters include gamma, T_g, alpha (unreported), and the inference guidance scale. No new physical entities are introduced. The main domain assumptions are that frozen VLM features transfer anomaly semantics across surfaces and that the generated masks from textual inversion are valid.

free parameters (4)
  • gamma (cross-modal attention weight in SGA) = 1
    Set to 1 in the implementation (Eq. 5) with no sensitivity analysis reported; it controls how strongly the cross-modal features influence generation.
  • T_g (ASEA guidance steps) = 3
    Chosen from the values {1,2,3,4,5} because it maximizes IS, IL and AUC in Table 4; no separate validation split is described, so this selection may be tuned to the test metrics.
  • alpha (ASEA attention guidance learning rate) = not reported
    Appears in Eq. 4 for the iterative update of eg; the value is not stated, which makes exact replication difficult.
  • Inference classifier-free guidance scale = 7.5
    Used in the 30-step DDIM sampler; set by hand and not swept in the paper.
assumptions (4)
  • domain assumption Pre-trained Stable Diffusion 1.5 and BLIP-2 provide sufficiently rich representations for industrial anomaly patterns.
    The framework freezes both models and relies on their joint generalization to unseen defect types; this is invoked in the Methods section (Preliminaries and Implementation Details) and is not empirically verified beyond the reported datasets.
  • domain assumption A short text phrase 'This is an image with [anomaly]' plus a small anomaly crop Ia forms a consistent cross-modal description of the defect.
    Used by CSM and ASEA; the masked anomaly region in Ia is assumed to correspond to the anomaly keyword in Ta. This enters in the 'Cross-modal Semantic Modeling' paragraph and the caption of Figure 1.
  • domain assumption The anomaly masks produced by learned textual inversion (mask embedding em, following AnoDiff) are valid and diverse proxies for real anomaly masks.
    Used to generate the 1,000 synthetic anomaly samples per category and to define MA in the energy function; see Appendix B, Mask Generation.
  • domain assumption The cross-attention guidance energy function of Chen et al. (2024), designed for Stable Diffusion cross-attention, transfers directly to BLIP-2 cross-attention layers.
    ASEA applies Eq. 3 to BLIP-2 attention maps without modification; the transfer of this guidance mechanism is assumed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AnomalyControl: Learning Cross-modal Semantic Features for Controllable Anomaly Synthesis." pith.science (2026). https://pith.science/paper/FDOCL6CL

@misc{pith2026241206510,
  author       = {Pith},
  title        = {Pith review of: AnomalyControl: Learning Cross-modal Semantic Features for Controllable Anomaly Synthesis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FDOCL6CL}},
  note         = {Machine review of arXiv:2412.06510}
}
read the original abstract

Anomaly synthesis is a crucial approach to augment abnormal data for advancing anomaly inspection. Based on the knowledge from the large-scale pre-training, existing text-to-image anomaly synthesis methods predominantly focus on textual information or coarse-aligned visual features to guide the entire generation process. However, these methods often lack sufficient descriptors to capture the complicated characteristics of realistic anomalies (e.g., the fine-grained visual pattern of anomalies), limiting the realism and generalization of the generation process. To this end, we propose a novel anomaly synthesis framework called AnomalyControl to learn cross-modal semantic features as guidance signals, which could encode the generalized anomaly cues from text-image reference prompts and improve the realism of synthesized abnormal samples. Specifically, AnomalyControl adopts a flexible and non-matching prompt pair (i.e., a text-image reference prompt and a targeted text prompt), where a Cross-modal Semantic Modeling (CSM) module is designed to extract cross-modal semantic features from the textual and visual descriptors. Then, an Anomaly-Semantic Enhanced Attention (ASEA) mechanism is formulated to allow CSM to focus on the specific visual patterns of the anomaly, thus enhancing the realism and contextual relevance of the generated anomaly features. Treating cross-modal semantic features as the prior, a Semantic Guided Adapter (SGA) is designed to encode effective guidance signals for the adequate and controllable synthesis process. Extensive experiments indicate that AnomalyControl can achieve state-of-the-art results in anomaly synthesis compared with existing methods while exhibiting superior performance for downstream tasks.

Figures

Figures reproduced from arXiv: 2412.06510 by the authors.

Figure 1
Figure 1. Our method learns cross-modal semantic features as the priors to achieve both [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Pipeline of the proposed AnomalyControl. Our framework consists of three modules to enhance controllability in [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Display of anomaly synthesis on 15 objects from the MVTec AD dataset. The first column shows the normal image, [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (6 more)
Figure 5
Figure 5. Figure 5: Effect of guidance steps Tg in ASEA. As Tg in￾creases, the loss value decreases, where ASEA progressively refines VLM’s attention on anomaly regions. The generated images exhibit improved anomaly realism, confirming the effectiveness of ASEA in focusing on specified an…
Figure 6
Figure 6. Figure 6: Comparison of anomaly synthesis results between AnoDiff and our method. The images show the same normal image [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]
Figure 7
Figure 7. Figure 7: Qualitative Evaluation. Our method could exhibit large improvements in realism and generalization. (a) Comparison with anomaly synthesis methods. Each row showcases anomalies across various object categories, where binary-value masks indicate the position of the genera…
Figure 8
Figure 8. Figure 8: Controllable anomaly synthesis across different object categories and anomaly types. Given a normal image (wood [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 9
Figure 9. Figure 9: Display of anomaly synthesis on 6 objects from the MPDD dataset. The first column shows the normal image, the [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]
Figure 10
Figure 10. Figure 10: Display of anomaly synthesis on 12 objects from the ViSA dataset. The first column shows the normal image, the [PITH_FULL_IMAGE:figures/full_fig_p016_10.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. DeCo: Zero-Shot Industrial Anomaly Generation through Decoupling and Recoupling

    cs.CV 2026-08 conditional novelty 6.0 of 10

    DeCo decouples the shape of an industrial defect from its source product and recouples it with a new product, improving downstream defect detection by 5-8% pixel AP.

  2. NexViTAD: Few-shot Unsupervised Cross-Domain Defect Detection via Vision Foundation Models and Multi-Task Learning

    cs.CV 2025-07 reject novelty 4.0 of 10

    NexViTAD fuses Hiera and DINOv2 features with a multi-task decoder and Sinkhorn K-means memory bank to detect industrial defects across domains, reporting MVTec AD target AUC of 97.5%.

Reference graph

Works this paper leans on

41 extracted references · 28 canonical work pages · cited by 2 Pith papers

  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]

    Blended latent diffusion

    Omri Avrahami, Ohad Fried, and Dani Lischinski. Blended latent diffusion. ACM transactions on graphics (TOG), 42 0 (4): 0 1--11, 2023

    work page 2023

  3. [3]

    ediff-i: Text-to-image diffusion models with an ensemble of expert denoisers

    Yogesh Balaji, Seungjun Nah, Xun Huang, Arash Vahdat, Jiaming Song, Qinsheng Zhang, Karsten Kreis, Miika Aittala, Timo Aila, Samuli Laine, et al. ediff-i: Text-to-image diffusion models with an ensemble of expert denoisers. arXiv preprint arXiv:2211.01324, 2022

    arXiv 2022

  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 (CVPR), pages 9592--9600, 2019

    work page 2019

  5. [5]

    The mvtec anomaly detection dataset: a comprehensive real-world dataset for unsupervised anomaly detection

    Paul Bergmann, Kilian Batzner, Michael Fauser, David Sattlegger, and Carsten Steger. The mvtec anomaly detection dataset: a comprehensive real-world dataset for unsupervised anomaly detection. International Journal of Computer Vision (IJCV), 129 0 (4): 0 1038--1059, 2021

    work page 2021

  6. [6]

    Demystifying mmd gans

    Miko aj Bi \'n kowski, Danica J Sutherland, Michael Arbel, and Arthur Gretton. Demystifying mmd gans. arXiv preprint arXiv:1801.01401, 2018

    arXiv 2018

  7. [7]

    Training-free layout control with cross-attention guidance

    Minghao Chen, Iro Laina, and Andrea Vedaldi. Training-free layout control with cross-attention guidance. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), pages 5343--5353, 2024

    work page 2024

  8. [8]

    Easynet: An easy network for 3d industrial anomaly detection

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

    work page 2023

Show all 41 references
  1. [9]

    Few-shot defect image generation via defect-aware feature manipulation

    Yuxuan Duan, Yan Hong, Li Niu, and Liqing Zhang. Few-shot defect image generation via defect-aware feature manipulation. In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), pages 571--578, 2023

    2023

  2. [10]

    An image is worth one word: Personalizing text-to-image generation using textual inversion

    Rinon Gal, Yuval Alaluf, Yuval Atzmon, Or Patashnik, Amit H Bermano, Gal Chechik, and Daniel Cohen-Or. An image is worth one word: Personalizing text-to-image generation using textual inversion. arXiv preprint arXiv:2208.01618, 2022

    2022 arXiv

  3. [11]

    Gans trained by a two time-scale update rule converge to a local nash equilibrium

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. Gans trained by a two time-scale update rule converge to a local nash equilibrium. Advances in Neural Information Processing Systems (NeurIPS), 30, 2017

    2017

  4. [12]

    Classifier-free diffusion guidance

    Jonathan Ho and Tim Salimans. Classifier-free diffusion guidance. arXiv preprint arXiv:2207.12598, 2022

    2022 arXiv

  5. [13]

    Anomalyxfusion: Multi-modal anomaly synthesis with diffusion

    Jie Hu, Yawen Huang, Yilin Lu, Guoyang Xie, Guannan Jiang, and Yefeng Zheng. Anomalyxfusion: Multi-modal anomaly synthesis with diffusion. arXiv preprint arXiv:2404.19444, 2024 a

    2024 arXiv

  6. [14]

    Anomalydiffusion: Few-shot anomaly image generation with diffusion model

    Teng Hu, Jiangning Zhang, Ran Yi, Yuzhen Du, Xu Chen, Liang Liu, Yabiao Wang, and Chengjie Wang. Anomalydiffusion: Few-shot anomaly image generation with diffusion model. In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), pages 8526--8534, 2024 b

    2024

  7. [15]

    Adapting visual-language models for generalizable anomaly detection in medical images

    Chaoqin Huang, Aofan Jiang, Jinghao Feng, Ya Zhang, Xinchao Wang, and Yanfeng Wang. Adapting visual-language models for generalizable anomaly detection in medical images. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 11375--...

    2024

  8. [16]

    Auto-encoding variational bayes

    Diederik P Kingma. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114, 2013

    2013 arXiv

  9. [17]

    Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models

    Junnan Li, Dongxu Li, Silvio Savarese, and Steven Hoi. Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models. In International Conference on Machine Learning (ICML), pages 19730--19742, 2023

    2023

  10. [18]

    Deep industrial image anomaly detection: A survey

    Jiaqi Liu, Guoyang Xie, Jinbao Wang, Shangnian Li, Chengjie Wang, Feng Zheng, and Yaochu Jin. Deep industrial image anomaly detection: A survey. Machine Intelligence Research, 21 0 (1): 0 104--135, 2024

    2024

  11. [19]

    Decoupled weight decay regularization

    I Loshchilov. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101, 2017

    2017 arXiv

  12. [20]

    T2i-adapter: Learning adapters to dig out more controllable ability for text-to-image diffusion models

    Chong Mou, Xintao Wang, Liangbin Xie, Yanze Wu, Jian Zhang, Zhongang Qi, and Ying Shan. T2i-adapter: Learning adapters to dig out more controllable ability for text-to-image diffusion models. In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI), pages 4296--...

    2024

  13. [21]

    Glide: Towards photorealistic image generation and editing with text-guided diffusion models

    Alex Nichol, Prafulla Dhariwal, Aditya Ramesh, Pranav Shyam, Pamela Mishkin, Bob McGrew, Ilya Sutskever, and Mark Chen. Glide: Towards photorealistic image generation and editing with text-guided diffusion models. arXiv preprint arXiv:2112.10741, 2021

    2021 arXiv

  14. [22]

    Defect image sample generation with gan for improving defect recognition

    Shuanlong Niu, Bin Li, Xinggang Wang, and Hui Lin. Defect image sample generation with gan for improving defect recognition. IEEE Transactions on Automation Science and Engineering, 17 0 (3): 0 1611--1622, 2020

    2020

  15. [23]

    Few-shot image generation via cross-domain correspondence

    Utkarsh Ojha, Yijun Li, Jingwan Lu, Alexei A Efros, Yong Jae Lee, Eli Shechtman, and Richard Zhang. Few-shot image generation via cross-domain correspondence. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10743--10752, 2021 a

    2021

  16. [24]

    Few-shot image generation via cross-domain correspondence

    Utkarsh Ojha, Yijun Li, Jingwan Lu, Alexei A Efros, Yong Jae Lee, Eli Shechtman, and Richard Zhang. Few-shot image generation via cross-domain correspondence. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10743--10752, 2021 b

    2021

  17. [25]

    Diffusers: State-of-the-art diffusion models

    Patrick von Platen, Suraj Patil, Anton Lozhkov, Pedro Cuenca, Nathan Lambert, Kashif Rasul, Mishig Davaadorj, and Thomas Wolf . Diffusers: State-of-the-art diffusion models . https://github.com/huggingface/diffusers, 2022

    2022

  18. [26]

    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 (IC...

    2021

  19. [27]

    Hierarchical text-conditional image generation with clip latents

    Aditya Ramesh, Prafulla Dhariwal, Alex Nichol, Casey Chu, and Mark Chen. Hierarchical text-conditional image generation with clip latents. arXiv preprint arXiv:2204.06125, 1 0 (2): 0 3, 2022

    2022 arXiv

  20. [28]

    High-resolution image synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj \"o rn Ommer. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10684--10695, 2022

    2022

  21. [29]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In Medical image computing and computer-assisted intervention (MICCAI), pages 234--241. Springer, 2015

    2015

  22. [30]

    Photorealistic text-to-image diffusion models with deep language understanding

    Chitwan Saharia, William Chan, Saurabh Saxena, Lala Li, Jay Whang, Emily L Denton, Kamyar Ghasemipour, Raphael Gontijo Lopes, Burcu Karagol Ayan, Tim Salimans, et al. Photorealistic text-to-image diffusion models with deep language understanding. Advances in Neural Information...

    2022

  23. [31]

    Improved techniques for training gans

    Tim Salimans, Ian Goodfellow, Wojciech Zaremba, Vicki Cheung, Alec Radford, and Xi Chen. Improved techniques for training gans. Advances in Neural Information Processing Systems (NeurIPS), 29, 2016

    2016

  24. [32]

    Cut: A controllable, universal, and training-free visual anomaly generation framework

    Han Sun, Yunkang Cao, and Olga Fink. Cut: A controllable, universal, and training-free visual anomaly generation framework. arXiv preprint arXiv:2406.01078, 2024

    2024 arXiv

  25. [33]

    Raphael: Text-to-image generation via large mixture of diffusion paths

    Zeyue Xue, Guanglu Song, Qiushan Guo, Boxiao Liu, Zhuofan Zong, Yu Liu, and Ping Luo. Raphael: Text-to-image generation via large mixture of diffusion paths. Advances in Neural Information Processing Systems (NeurIPS), 36, 2024

    2024

  26. [34]

    Ip-adapter: Text compatible image prompt adapter for text-to-image diffusion models

    Hu Ye, Jun Zhang, Sibo Liu, Xiao Han, and Wei Yang. Ip-adapter: Text compatible image prompt adapter for text-to-image diffusion models. arXiv preprint arXiv:2308.06721, 2023

    2023 arXiv

  27. [35]

    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 (ICCV), pages 8330--8339, 2021

    2021

  28. [36]

    Defect-gan: High-fidelity defect synthesis for automated defect inspection

    Gongjie Zhang, Kaiwen Cui, Tzu-Yi Hung, and Shijian Lu. Defect-gan: High-fidelity defect synthesis for automated defect inspection. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), pages 2524--2534, 2021

    2021

  29. [37]

    Prototypical residual networks for anomaly detection and localization

    Hui Zhang, Zuxuan Wu, Zheng Wang, Zhineng Chen, and Yu-Gang Jiang. Prototypical residual networks for anomaly detection and localization. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 16281--16291, 2023 a

    2023

  30. [38]

    Adding conditional control to text-to-image diffusion models

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. Adding conditional control to text-to-image diffusion models. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 3836--3847, 2023 b

    2023

  31. [39]

    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. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 16699--16708, 2024

    2024

  32. [40]

    Differentiable augmentation for data-efficient gan training

    Shengyu Zhao, Zhijian Liu, Ji Lin, Jun-Yan Zhu, and Song Han. Differentiable augmentation for data-efficient gan training. Advances in Neural Information Processing Systems (NeurIPS), 33: 0 7559--7570, 2020

    2020

  33. [41]

    Uni-controlnet: All-in-one control to text-to-image diffusion models

    Shihao Zhao, Dongdong Chen, Yen-Chun Chen, Jianmin Bao, Shaozhe Hao, Lu Yuan, and Kwan-Yee K Wong. Uni-controlnet: All-in-one control to text-to-image diffusion models. Advances in Neural Information Processing Systems (NeurIPS), 36, 2023

    2023

Pith tools

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