pith. sign in

arxiv: 2606.28268 · v1 · pith:XDSTU37Xnew · submitted 2026-06-26 · 💻 cs.CV · cs.AI

Learning Topology-Aware Representations via Test-Time Adaptation for Anomaly Segmentation

Ali Zia , Usman Ali , Abdul Rehman , Umer Ramzan , Kang Han , Muhammad Faheem , Shahnawaz Qureshi , Wei Xiang This is my paper

Pith reviewed 2026-06-29 04:01 UTC · model grok-4.3

classification 💻 cs.CV cs.AI
keywords test-time adaptationanomaly segmentationpersistent homologytopological data analysiscubical complexpseudo-labelsdistribution shiftdefect detection
0
0 comments X

The pith

TopoTTA applies multi-level cubical filtration to anomaly maps to create topological pseudo-labels that guide test-time adaptation.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces TopoTTA to improve anomaly segmentation under distribution shifts. Current test-time methods use pixel-level rules like confidence thresholds that break structural consistency when noise or texture changes occur. TopoTTA instead runs persistent homology through multi-level cubical complex filtration on the anomaly score maps. This step produces topological pseudo-labels that steer a small test-time classifier while the main model stays frozen. On six benchmarks the method records an average 15 percent F1 gain, with bigger lifts on defects that have intricate geometry.

Core claim

By applying multi-level cubical complex filtration to anomaly score maps, TopoTTA derives robust topological pseudo-labels that guide a lightweight test-time classifier, enhancing segmentation quality without retraining the backbone model. The approach avoids reliance on method-specific raw-score thresholding for mask binarisation, preserves connectivity, and generalises across both 2D and 3D modalities.

What carries the argument

Multi-level cubical complex filtration drawn from persistent homology, applied directly to anomaly score maps to extract topological pseudo-labels that steer adaptation.

If this is right

  • The method preserves connectivity and structural coherence that pixel heuristics lose under noise.
  • It removes the need for dataset-specific raw-score thresholds when creating final masks.
  • Performance gains appear on both 2D and 3D data without changing the backbone.
  • Improvements are largest for anomalies that show complex geometric structure.
  • Topological reasoning supplies a route to structure-aware generalisation during adaptation.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same filtration step could be tested on other test-time tasks such as semantic segmentation under domain shift.
  • Combining the pseudo-labels with persistence diagrams rather than raw labels might yield even richer guidance signals.
  • In factory inspection pipelines the approach could lower the cost of collecting new labels when camera or lighting conditions drift.
  • Adaptive choice of filtration levels based on the complexity of each anomaly map remains an open direction.

Load-bearing premise

That cubical complex filtration on anomaly score maps will produce pseudo-labels more robust to noise and texture variation than simple pixel-level rules.

What would settle it

An experiment that adds controlled texture noise to the score maps of one benchmark and measures whether the reported F1 advantage over pixel-heuristic TTA methods vanishes or reverses.

Figures

Figures reproduced from arXiv: 2606.28268 by Abdul Rehman, Ali Zia, Kang Han, Muhammad Faheem, Shahnawaz Qureshi, Umer Ramzan, Usman Ali, Wei Xiang.

Figure 1
Figure 1. Figure 1: Progressive refinement of anomaly segmentation using multi-level filtrations on cubical complexes. Each row shows a 2D or 3D test image with (left [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Elementary cubes across dimensions and a cubical complex. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Overview of the TopoTTA architecture. Given a test image I, an AD&S method produces an anomaly score map Ψ. A pre-trained feature extractor g generates dense feature maps F = g(I). Topological pseudo-labels are extracted by applying multi-level cubical complex filtrations (both sublevel and superlevel) to Ψ, producing structurally meaningful binary masks via persistent homology. These masks are fused using… view at source ↗
Figure 4
Figure 4. Figure 4: Sublevel and superlevel filtrations on a 2D grayscale image. Given an image [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: Qualitative comparison on the 3D MVTec AD dataset [89]. TopoTTA improves surface-level defect localization and preserves geometric continuity in 3D reconstructions. respectively, to adapt the test-time module, after which evalua￾tion is performed on the remaining test images of that class. This setting remains unsupervised, as no anomalous examples, class labels, or pixel-level ground-truth masks are used … view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative comparison across MVTec AD, MVTec LOCO, VisA, and Real-IAD. Columns: RGB, Ground Truth (GT), anomaly heat map, simple thresholding (THR), TTT4AS, and our TopoTTA. TopoTTA produces sharper, topologically consistent anomaly segmentations across diverse categories and datasets. 3D MVTec AD [89] RGB PC GT Heat Map THR [55] TTT4AS [78] TopoTTA Bagel Cookie Peach [PITH_FULL_IMAGE:figures/full_fig_p0… view at source ↗
Figure 7
Figure 7. Figure 7: Qualitative comparison under input perturbations. Representative examples under sequential perturbations comprising brightness shift, speckle noise, and Gaussian noise. Compared with TTT4AS, whose masks become increasingly fragmented and noisy, TopoTTA preserves more coherent anomaly structure and yields visually cleaner segmentations. enforcing structural consistency during refinement. This shows that the… view at source ↗
Figure 8
Figure 8. Figure 8: Qualitative comparison on texture-heavy categories (carpet, grid, wood). [PITH_FULL_IMAGE:figures/full_fig_p014_8.png] view at source ↗
read the original abstract

Test-time adaptation (TTA) has emerged as a promising paradigm for mitigating distribution shifts in deep models. However, existing TTA approaches for anomaly segmentation remain limited by their reliance on pixel-level heuristics, such as confidence thresholding or entropy minimisation, which fail to preserve structural consistency under noise and texture variation. Moreover, they typically treat anomaly maps as flat intensity fields, ignoring the higher-order spatial relationships that characterise complex defect geometries. We introduce TopoTTA (Topological Test-Time Adaptation), a novel framework that integrates persistent homology, a tool from topological data analysis, into the TTA pipeline to enforce geometric and structural coherence during adaptation. By applying multi-level cubical complex filtration to anomaly score maps, TopoTTA derives robust topological pseudo-labels that guide a lightweight test-time classifier, enhancing segmentation quality without retraining the backbone model. The approach avoids reliance on method-specific raw-score thresholding for mask binarisation, preserves connectivity, and generalises across both 2D and 3D modalities. Extensive experiments across six standard benchmarks (MVTec AD, VisA, Real-IAD, MVTec 3D-AD, AnomalyShapeNet, and MVTec LOCO) demonstrate an average 15% F1 improvement over state-of-the-art unsupervised anomaly detection and segmentation methods, with the largest gains on anomalies exhibiting complex geometric or structural variations. These findings suggest that integrating topological reasoning into test-time adaptation provides a principled route to structure-aware generalisation, bridging the gap between geometric learning and robust adaptation.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 0 minor

Summary. The paper introduces TopoTTA, a test-time adaptation method for anomaly segmentation that applies persistent homology through multi-level cubical complex filtration to anomaly score maps in order to derive topological pseudo-labels. These labels then supervise a lightweight test-time classifier, with the goal of improving structural consistency over pixel-level heuristics such as confidence or entropy thresholding. The central empirical claim is an average 15% F1-score gain across six benchmarks (MVTec AD, VisA, Real-IAD, MVTec 3D-AD, AnomalyShapeNet, MVTec LOCO) relative to prior unsupervised anomaly detection and segmentation methods, with larger gains on geometrically complex anomalies.

Significance. If the topological component can be shown to produce pseudo-labels that are demonstrably more robust than standard thresholding on identical score maps, the work would provide a concrete bridge between topological data analysis and test-time adaptation, offering a principled mechanism for preserving connectivity and higher-order geometry under distribution shift without backbone retraining.

major comments (2)
  1. [Experiments / Abstract] The abstract and experimental claims assert that multi-level cubical filtration yields pseudo-labels superior to pixel-level heuristics for guiding adaptation, yet no ablation replaces the filtration step with standard thresholding or entropy minimization applied to the identical anomaly score maps. Without this isolation, the reported 15% F1 improvement cannot be attributed specifically to the topological reasoning rather than the lightweight classifier or other TTA elements.
  2. [Method / Experiments] The motivation states that the approach is robust to noise and texture variation, but the manuscript supplies no controlled experiments that inject synthetic noise or texture perturbations into the score maps and measure stability of the derived topological pseudo-labels versus baseline heuristics. This test is load-bearing for the central claim that cubical filtration preserves structural consistency where pixel-level methods fail.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. The comments highlight important aspects of experimental validation that we will address to strengthen the attribution of our results to the topological component. We respond to each major comment below.

read point-by-point responses

  1. Referee: [Experiments / Abstract] The abstract and experimental claims assert that multi-level cubical filtration yields pseudo-labels superior to pixel-level heuristics for guiding adaptation, yet no ablation replaces the filtration step with standard thresholding or entropy minimization applied to the identical anomaly score maps. Without this isolation, the reported 15% F1 improvement cannot be attributed specifically to the topological reasoning rather than the lightweight classifier or other TTA elements.

    Authors: We agree that the current manuscript lacks an explicit ablation that applies standard thresholding or entropy minimization directly to the identical anomaly score maps (i.e., holding the backbone and score maps fixed) before feeding them to the lightweight classifier. Such an isolation experiment would more cleanly attribute gains to the cubical filtration step. We will add this ablation study in the revised manuscript, reporting F1 scores for both the topological pseudo-labels and the baseline heuristics on the same six benchmarks. revision: yes

  2. Referee: [Method / Experiments] The motivation states that the approach is robust to noise and texture variation, but the manuscript supplies no controlled experiments that inject synthetic noise or texture perturbations into the score maps and measure stability of the derived topological pseudo-labels versus baseline heuristics. This test is load-bearing for the central claim that cubical filtration preserves structural consistency where pixel-level methods fail.

    Authors: The referee correctly notes the absence of controlled synthetic perturbation experiments on the anomaly score maps. While the six benchmarks contain natural texture and geometric variations, they do not constitute the requested synthetic injection protocol. We will incorporate these experiments in the revision by adding controlled noise (e.g., Gaussian, salt-and-pepper) and texture perturbations to the score maps and quantifying stability metrics (e.g., topological feature persistence, F1 variance) for both our method and the pixel-level baselines. revision: yes

Circularity Check

0 steps flagged

No circularity: procedural method description with no self-referential reductions

full rationale

The paper presents TopoTTA as a framework that applies multi-level cubical complex filtration to anomaly score maps to obtain topological pseudo-labels for guiding a test-time classifier. This is a definitional description of the proposed pipeline rather than a derivation chain in which any claimed prediction or result reduces by construction to fitted inputs or self-citations. No equations appear in the abstract, and the central claims concern empirical F1 gains on benchmarks rather than mathematical predictions that loop back to the method's own parameters. The reader's assessment of score 1.0 aligns with the absence of any load-bearing step matching the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

Ledger is necessarily sparse because only the abstract is available; full paper may contain additional fitted parameters or unstated modeling choices.

axioms (1)
  • domain assumption Persistent homology computed on multi-level cubical filtrations of anomaly score maps yields robust topological pseudo-labels that preserve connectivity and generalize across modalities.
    This premise is invoked to justify replacing pixel-level heuristics with topological guidance.
invented entities (1)
  • TopoTTA framework no independent evidence
    purpose: To integrate topological reasoning into test-time adaptation for anomaly segmentation
    Newly proposed method whose effectiveness is asserted in the abstract.

pith-pipeline@v0.9.1-grok · 5829 in / 1331 out tokens · 67149 ms · 2026-06-29T04:01:40.865188+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

98 extracted references · 3 linked inside Pith

  1. [1]

    Test-time training with self-supervision for generalization under distribution shifts,

    Y . Sun, X. Wang, Z. Liu, J. Miller, A. Efros, and M. Hardt, “Test-time training with self-supervision for generalization under distribution shifts,” inInternational conference on machine learning. PMLR, 2020, pp. 9229–9248

    2020

  2. [2]

    Domain generalization for cross-domain fault diagnosis: An application-oriented perspective and a benchmark study,

    C. Zhao, E. Zio, and W. Shen, “Domain generalization for cross-domain fault diagnosis: An application-oriented perspective and a benchmark study,”Reliability Engineering & System Safety, p. 109964, 2024

    2024

  3. [3]

    Deep learning for unsupervised anomaly localization in industrial images: A survey,

    X. Tao, X. Gong, X. Zhang, S. Yan, and C. Adak, “Deep learning for unsupervised anomaly localization in industrial images: A survey,” IEEE Transactions on Instrumentation and Measurement, vol. 71, pp. 1–21, 2022

    2022

  4. [4]

    A survey on visual anomaly detection: Challenge, approach, and prospect,

    Y . Cao, X. Xu, J. Zhang, Y . Cheng, X. Huang, G. Pang, and W. Shen, “A survey on visual anomaly detection: Challenge, approach, and prospect,” arXiv preprint arXiv:2401.16402, 2024

    arXiv 2024

  5. [5]

    Enhanced multi-scale features mutual mapping fusion based on reverse knowledge distillation for industrial anomaly detection and localization,

    G. Tong, Q. Li, and Y . Song, “Enhanced multi-scale features mutual mapping fusion based on reverse knowledge distillation for industrial anomaly detection and localization,”IEEE Transactions on Big Data, vol. 10, no. 4, pp. 498–513, 2024

    2024

  6. [6]

    Aekd: Unsupervised auto-encoder knowledge distillation for industrial anomaly detection,

    Q. Wu, H. Li, C. Tian, L. Wen, and X. Li, “Aekd: Unsupervised auto-encoder knowledge distillation for industrial anomaly detection,” Journal of Manufacturing Systems, vol. 73, pp. 159–169, 2024

    2024

  7. [7]

    Msflow: Multiscale flow-based framework for unsupervised anomaly detection,

    Y . Zhou, X. Xu, J. Song, F. Shen, and H. T. Shen, “Msflow: Multiscale flow-based framework for unsupervised anomaly detection,”IEEE Transactions on Neural Networks and Learning Systems, 2024

    2024

  8. [8]

    Supervised anomaly detection for complex industrial images,

    A. Baitieva, D. Hurych, V . Besnier, and O. Bernard, “Supervised anomaly detection for complex industrial images,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 17 754–17 762

    2024

  9. [9]

    Anomalydiffusion: Few-shot anomaly image generation with diffusion model,

    T. Hu, J. Zhang, R. Yi, Y . Du, X. Chen, L. Liu, Y . Wang, and C. Wang, “Anomalydiffusion: Few-shot anomaly image generation with diffusion model,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 38, 2024, pp. 8526–8534

    2024

  10. [10]

    Anomaly heterogeneity learning for open-set supervised anomaly detection,

    J. Zhu, C. Ding, Y . Tian, and G. Pang, “Anomaly heterogeneity learning for open-set supervised anomaly detection,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 17 616–17 626

    2024

  11. [11]

    Catching both gray and black swans: Open-set supervised anomaly detection,

    C. Ding, G. Pang, and C. Shen, “Catching both gray and black swans: Open-set supervised anomaly detection,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 7388–7398

    2022

  12. [12]

    Texems: Texture exemplars for defect detection on random textured surfaces,

    X. Xie and M. Mirmehdi, “Texems: Texture exemplars for defect detection on random textured surfaces,”IEEE transactions on pattern analysis and machine intelligence, vol. 29, no. 8, pp. 1454–1464, 2007

    2007

  13. [13]

    A high-efficiency fully convolutional networks for pixel-wise surface defect detection,

    L. Qiu, X. Wu, and Z. Yu, “A high-efficiency fully convolutional networks for pixel-wise surface defect detection,”IEEE Access, vol. 7, pp. 15 884–15 893, 2019

    2019

  14. [14]

    A theory of learning from different domains,

    S. Ben-David, J. Blitzer, K. Crammer, A. Kulesza, F. Pereira, and J. W. Vaughan, “A theory of learning from different domains,”Machine learning, vol. 79, pp. 151–175, 2010

    2010

  15. [15]

    Self-supervised aggregation of diverse experts for test-agnostic long-tailed recognition,

    Y . Zhang, B. Hooi, L. Hong, and J. Feng, “Self-supervised aggregation of diverse experts for test-agnostic long-tailed recognition,”Advances in Neural Information Processing Systems, vol. 35, pp. 34 077–34 090, 2022

    2022

  16. [16]

    Deep face detector adaptation without negative transfer or catastrophic forgetting,

    M. A. Jamal, H. Li, and B. Gong, “Deep face detector adaptation without negative transfer or catastrophic forgetting,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 5608–5618

    2018

  17. [17]

    On the road to online adaptation for semantic image segmentation,

    R. V olpi, P. De Jorge, D. Larlus, and G. Csurka, “On the road to online adaptation for semantic image segmentation,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 19 184–19 195

    2022

  18. [18]

    Transductive conformal inference with adaptive scores,

    U. Gazin, G. Blanchard, and E. Roquain, “Transductive conformal inference with adaptive scores,” inInternational Conference on Artificial Intelligence and Statistics. PMLR, 2024, pp. 1504–1512

    2024

  19. [19]

    Topological deep learning: a review of an emerging paradigm,

    A. Zia, A. Khamis, J. Nichols, U. B. Tayab, Z. Hayder, V . Rolland, E. Stone, and L. Petersson, “Topological deep learning: a review of an emerging paradigm,”Artificial Intelligence Review, vol. 57, no. 4, p. 77, 2024

    2024

  20. [20]

    Icml 2023 topological deep learning challenge: design and results,

    M. Papillon, M. Hajij, A. Myers, F. Frantzen, G. Zamzmi, H. Jenne, J. Mathe, J. Hoppe, M. Schaub, T. Papamarkouet al., “Icml 2023 topological deep learning challenge: design and results,” inTopological, Algebraic and Geometric Learning Workshops 2023. PMLR, 2023, pp. 3–8

    2023

  21. [21]

    Mambaad: Exploring state space models for multi- class unsupervised anomaly detection,

    H. He, Y . Bai, J. Zhang, Q. He, H. Chen, Z. Gan, C. Wang, X. Li, G. Tian, and L. Xie, “Mambaad: Exploring state space models for multi- class unsupervised anomaly detection,”arXiv preprint arXiv:2404.06564, 2024

    arXiv 2024

  22. [22]

    2d-3d feature fusion via cross-modal latent synthesis and attention-guided restoration for industrial anomaly detection,

    U. Ali, A. Zia, A. Rehman, U. Ramzan, Z. Hassan, T. Sattar, J. Wang, and W. Xiang, “2d-3d feature fusion via cross-modal latent synthesis and attention-guided restoration for industrial anomaly detection,” in 2025 International Conference on Digital Image Computing: Techniques and Applications (DICTA), 2025, pp. 1–8

    2025

  23. [23]

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

    Z. Fang, X. Wang, H. Li, J. Liu, Q. Hu, and J. Xiao, “Fastrecon: Few- shot industrial anomaly detection via fast feature reconstruction,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 17 481–17 490

    2023

  24. [25]

    A reconstruction-based feature adaptation for anomaly detection with self-supervised multi-scale aggregation,

    Z. Zuo, Z. Wu, B. Chen, and X. Zhong, “A reconstruction-based feature adaptation for anomaly detection with self-supervised multi-scale aggregation,” inICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2024, pp. 5840–5844

    2024

  25. [26]

    R3d-ad: Reconstruction via diffusion for 3d anomaly detection,

    Z. Zhou, L. Wang, N. Fang, Z. Wang, L. Qiu, and S. Zhang, “R3d-ad: Reconstruction via diffusion for 3d anomaly detection,” inEuropean conference on computer vision. Springer, 2024, pp. 91–107

    2024

  26. [27]

    Multi-feature reconstruction network using crossed-mask restoration for unsupervised anomaly detection,

    J. Wang, G. Xu, C. Li, G. Gao, and Y . Cheng, “Multi-feature reconstruction network using crossed-mask restoration for unsupervised anomaly detection,”arXiv preprint arXiv:2404.13273, 2024

    arXiv 2024

  27. [28]

    Superpixel masking and inpainting for self-supervised anomaly detection

    Z. Li, N. Li, K. Jiang, Z. Ma, X. Wei, X. Hong, and Y . Gong, “Superpixel masking and inpainting for self-supervised anomaly detection.” inBmvc, 2020

    2020

  28. [29]

    Fixing the train-test objective discrepancy: Iterative image inpainting for unsupervised anomaly detection,

    H. Nakanishi, M. Suzuki, and Y . Matsuo, “Fixing the train-test objective discrepancy: Iterative image inpainting for unsupervised anomaly detection,”Journal of Information Processing, vol. 30, pp. 495–504, 2022

    2022

  29. [30]

    Reconstruction by inpainting for visual anomaly detection,

    V . Zavrtanik, M. Kristan, and D. Sko ˇcaj, “Reconstruction by inpainting for visual anomaly detection,”Pattern Recognition, vol. 112, p. 107706, 2021. 16

    2021

  30. [31]

    Inpainting transformer for anomaly detection,

    J. Pirnay and K. Chai, “Inpainting transformer for anomaly detection,” in International Conference on Image Analysis and Processing. Springer, 2022, pp. 394–406

    2022

  31. [32]

    Ami-net: Adaptive mask inpainting network for industrial anomaly detection and localization,

    W. Luo, H. Yao, W. Yu, and Z. Li, “Ami-net: Adaptive mask inpainting network for industrial anomaly detection and localization,”IEEE Transactions on Automation Science and Engineering, 2024

    2024

  32. [33]

    Glad: Towards better reconstruction with global and local adaptive diffusion models for unsupervised anomaly detection,

    H. Yao, M. Liu, Z. Yin, Z. Yan, X. Hong, and W. Zuo, “Glad: Towards better reconstruction with global and local adaptive diffusion models for unsupervised anomaly detection,” inEuropean Conference on Computer Vision. Springer, 2024, pp. 1–17

    2024

  33. [34]

    Transfusion–a transparency- based diffusion model for anomaly detection,

    M. Fu ˇcka, V . Zavrtanik, and D. Sko ˇcaj, “Transfusion–a transparency- based diffusion model for anomaly detection,” inEuropean conference on computer vision. Springer, 2024, pp. 91–108

    2024

  34. [35]

    Cagen: Controllable anomaly generator using diffusion model,

    B. Jiang, Y . Xie, J. Li, N. Li, Y . Jiang, and S.-T. Xia, “Cagen: Controllable anomaly generator using diffusion model,” inICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2024, pp. 3110–3114

    2024

  35. [36]

    Neural network training strategy to enhance anomaly detection performance: A perspective on reconstruction loss amplification,

    Y . Park, S. Kang, M. J. Kim, H. Jeong, H. Park, H. S. Kim, and J. Yi, “Neural network training strategy to enhance anomaly detection performance: A perspective on reconstruction loss amplification,” in ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2024, pp. 5165–5169

    2024

  36. [38]

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

    T. Defard, A. Setkov, A. Loesch, and R. Audigier, “Padim: a patch distribution modeling framework for anomaly detection and localization,” inInternational Conference on Pattern Recognition. Springer, 2021, pp. 475–489

    2021

  37. [39]

    Structural teacher-student normality learning for multi-class anomaly detection and localization,

    H. Deng and X. Li, “Structural teacher-student normality learning for multi-class anomaly detection and localization,”arXiv preprint arXiv:2402.17091, 2024

    arXiv 2024

  38. [41]

    Destseg: Segmentation guided denoising student-teacher for anomaly detection,

    X. Zhang, S. Li, X. Li, P. Huang, J. Shan, and T. Chen, “Destseg: Segmentation guided denoising student-teacher for anomaly detection,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 3914–3923

    2023

  39. [42]

    Masked feature regeneration based asymmetric student–teacher network for anomaly detection,

    H. Gu, G. Li, and Z. Liu, “Masked feature regeneration based asymmetric student–teacher network for anomaly detection,”Multimedia Tools and Applications, vol. 83, no. 42, pp. 90 573–90 594, 2024

    2024

  40. [43]

    Asymmetric student-teacher networks for industrial anomaly detection,

    M. Rudolph, T. Wehrbein, B. Rosenhahn, and B. Wandt, “Asymmetric student-teacher networks for industrial anomaly detection,” inProceed- ings of the IEEE/CVF winter conference on applications of computer vision, 2023, pp. 2592–2602

    2023

  41. [44]

    Hierarchical gaussian mixture normalizing flow modeling for unified anomaly detection,

    X. Yao, R. Li, Z. Qian, L. Wang, and C. Zhang, “Hierarchical gaussian mixture normalizing flow modeling for unified anomaly detection,” in European Conference on Computer Vision. Springer, 2024, pp. 92–108

    2024

  42. [45]

    Cflow-ad: Real-time unsupervised anomaly detection with localization via conditional normalizing flows,

    D. Gudovskiy, S. Ishizaka, and K. Kozuka, “Cflow-ad: Real-time unsupervised anomaly detection with localization via conditional normalizing flows,” inProceedings of the IEEE/CVF winter conference on applications of computer vision, 2022, pp. 98–107

    2022

  43. [46]

    Pyramidflow: High-resolution defect contrastive localization using pyramid normalizing flow,

    J. Lei, X. Hu, Y . Wang, and D. Liu, “Pyramidflow: High-resolution defect contrastive localization using pyramid normalizing flow,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2023, pp. 14 143–14 152

    2023

  44. [47]

    Sanflow: semantic-aware normalizing flow for anomaly detection and localization,

    D. Kim, S. Baik, and T. H. Kim, “Sanflow: semantic-aware normalizing flow for anomaly detection and localization,” inProceedings of the 37th International Conference on Neural Information Processing Systems, 2023, pp. 75 434–75 454

    2023

  45. [48]

    Zero-shot versus many- shot: Unsupervised texture anomaly detection,

    T. Aota, L. T. T. Tong, and T. Okatani, “Zero-shot versus many- shot: Unsupervised texture anomaly detection,” inProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2023, pp. 5564–5572

    2023

  46. [49]

    Towards scalable 3d anomaly detection and localization: A benchmark via 3d anomaly synthesis and a self-supervised learning network,

    W. Li, X. Xu, Y . Gu, B. Zheng, S. Gao, and Y . Wu, “Towards scalable 3d anomaly detection and localization: A benchmark via 3d anomaly synthesis and a self-supervised learning network,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 22 207–22 216

    2024

  47. [50]

    Anomalyx- fusion: Multi-modal anomaly synthesis with diffusion,

    J. Hu, Y . Huang, Y . Lu, G. Xie, G. Jiang, and Y . Zheng, “Anomalyx- fusion: Multi-modal anomaly synthesis with diffusion,”arXiv preprint arXiv:2404.19444, 2024

    arXiv 2024

  48. [51]

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

    Q. Chen, H. Luo, C. Lv, and Z. Zhang, “A unified anomaly synthesis strategy with gradient ascent for industrial anomaly detection and localization,” inEuropean Conference on Computer Vision. Springer, 2024, pp. 37–54

    2024

  49. [52]

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

    J. Guo, S. Lu, W. Zhang, F. Chen, H. Li, and H. Liao, “Dinomaly: The less is more philosophy in multi-class unsupervised anomaly detection,” inProceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp. 20 405–20 415

    2025

  50. [53]

    Mambaad: Exploring state space models for multi- class unsupervised anomaly detection,

    H. He, Y . Bai, J. Zhang, Q. He, H. Chen, Z. Gan, C. Wang, X. Li, G. Tian, and L. Xie, “Mambaad: Exploring state space models for multi- class unsupervised anomaly detection,”Advances in Neural Information Processing Systems, vol. 37, pp. 71 162–71 187, 2024

    2024

  51. [54]

    Multimodal industrial anomaly detection via hybrid fusion,

    Y . Wang, J. Peng, J. Zhang, R. Yi, Y . Wang, and C. Wang, “Multimodal industrial anomaly detection via hybrid fusion,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 8032–8041

    2023

  52. [55]

    Multimodal industrial anomaly detection by crossmodal feature mapping,

    A. Costanzino, P. Z. Ramirez, G. Lisanti, and L. Di Stefano, “Multimodal industrial anomaly detection by crossmodal feature mapping,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 17 234–17 243

    2024

  53. [56]

    Po3ad: Predicting point offsets toward better 3d point cloud anomaly detection,

    J. Ye, W. Zhao, X. Yang, G. Cheng, and K. Huang, “Po3ad: Predicting point offsets toward better 3d point cloud anomaly detection,” in Proceedings of the Computer Vision and Pattern Recognition Conference, 2025, pp. 1353–1362

    2025

  54. [57]

    Classification of hepatic lesions using the matching metric,

    A. Adcock, D. Rubin, and G. Carlsson, “Classification of hepatic lesions using the matching metric,”Computer vision and image understanding, vol. 121, pp. 36–42, 2014

    2014

  55. [58]

    Functional sum- maries of persistence diagrams,

    E. Berry, Y .-C. Chen, J. Cisewski-Kehe, and B. T. Fasy, “Functional sum- maries of persistence diagrams,”Journal of Applied and Computational Topology, vol. 4, no. 2, pp. 211–262, 2020

    2020

  56. [59]

    Predicting clinical outcomes in glioblastoma: an application of topolog- ical and functional data analysis,

    L. Crawford, A. Monod, A. X. Chen, S. Mukherjee, and R. Rabadán, “Predicting clinical outcomes in glioblastoma: an application of topolog- ical and functional data analysis,”Journal of the American Statistical Association, vol. 115, no. 531, pp. 1139–1150, 2020

    2020

  57. [60]

    Topological data analysis of high resolution diabetic retinopathy images,

    K. Garside, R. Henderson, I. Makarenko, and C. Masoller, “Topological data analysis of high resolution diabetic retinopathy images,”PloS one, vol. 14, no. 5, p. e0217413, 2019

    2019

  58. [61]

    A topological representation of branching neuronal morphologies,

    L. Kanari, P. Dłotko, M. Scolamiero, R. Levi, J. Shillcock, K. Hess, and H. Markram, “A topological representation of branching neuronal morphologies,”Neuroinformatics, vol. 16, pp. 3–13, 2018

    2018

  59. [62]

    Persistent homology-a survey,

    H. Edelsbrunner, J. Hareret al., “Persistent homology-a survey,” Contemporary mathematics, vol. 453, no. 26, pp. 257–282, 2008

    2008

  60. [63]

    Uncovering the topology of time-varying fmri data using cubical persistence,

    B. Rieck, T. Yates, C. Bock, K. Borgwardt, G. Wolf, N. Turk-Browne, and S. Krishnaswamy, “Uncovering the topology of time-varying fmri data using cubical persistence,”Advances in neural information processing systems, vol. 33, pp. 6900–6912, 2020

    2020

  61. [64]

    The persistent homology of dual digital image constructions,

    B. Bleile, A. Garin, T. Heiss, K. Maggs, and V . Robins, “The persistent homology of dual digital image constructions,” inResearch in Computational Topology 2. Springer, 2022, pp. 1–26

    2022

  62. [65]

    Stability of persistence diagrams,

    D. Cohen-Steiner, H. Edelsbrunner, and J. Harer, “Stability of persistence diagrams,”Discrete & Computational Geometry, vol. 37, no. 1, pp. 103–120, 2007

    2007

  63. [66]

    A comprehensive survey on test-time adaptation under distribution shifts,

    J. Liang, R. He, and T. Tan, “A comprehensive survey on test-time adaptation under distribution shifts,”International Journal of Computer Vision, vol. 133, no. 1, pp. 31–64, 2025

    2025

  64. [68]

    To adapt or not to adapt? real-time adaptation for semantic segmentation,

    M. B. Colomer, P. L. Dovesi, T. Panagiotakopoulos, J. F. Carvalho, L. Härenstam-Nielsen, H. Azizpour, H. Kjellström, D. Cremers, and M. Poggi, “To adapt or not to adapt? real-time adaptation for semantic segmentation,” inProceedings of the IEEE/CVF International Confer- ence on Computer Vision, 2023, pp. 16 548–16 559

    2023

  65. [69]

    Ev-tta: Test-time adaptation for event- based object recognition,

    J. Kim, I. Hwang, and Y . M. Kim, “Ev-tta: Test-time adaptation for event- based object recognition,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 17 745–17 754

    2022

  66. [70]

    Evaluating prediction-time batch normalization for robustness under covariate shift,

    Z. Nado, S. Padhy, D. Sculley, A. D’Amour, B. Lakshminarayanan, and J. Snoek, “Evaluating prediction-time batch normalization for robustness under covariate shift,”arXiv preprint arXiv:2006.10963, 2020

    arXiv 2006

  67. [71]

    Tipi: Test time adaptation with transformation invariance,

    A. T. Nguyen, T. Nguyen-Tang, S.-N. Lim, and P. H. Torr, “Tipi: Test time adaptation with transformation invariance,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 24 162–24 171

    2023

  68. [72]

    Sita: Single image test-time adaptation,

    A. Khurana, S. Paul, P. Rai, S. Biswas, and G. Aggarwal, “Sita: Single image test-time adaptation,”arXiv preprint arXiv:2112.02355, 2021

    arXiv 2021

  69. [73]

    Test-time training with self-supervision for generalization under distribution shifts,

    Y . Sun, X. Wang, Z. Liu, J. Miller, A. Efros, and M. Hardt, “Test-time training with self-supervision for generalization under distribution shifts,” inInternational conference on machine learning. PMLR, 2020, pp. 9229–9248. 17

    2020

  70. [74]

    Ttt++: When does self-supervised test-time training fail or thrive?

    Y . Liu, P. Kothari, B. Van Delft, B. Bellot-Gurlet, T. Mordan, and A. Alahi, “Ttt++: When does self-supervised test-time training fail or thrive?”Advances in Neural Information Processing Systems, vol. 34, pp. 21 808–21 820, 2021

    2021

  71. [75]

    Tent: Fully test-time adaptation by entropy minimization,

    D. Wang, E. Shelhamer, S. Liu, B. Olshausen, and T. Darrell, “Tent: Fully test-time adaptation by entropy minimization,”arXiv preprint arXiv:2006.10726, 2020

    Pith/arXiv arXiv 2006

  72. [76]

    Efficient test-time model adaptation without forgetting,

    S. Niu, J. Wu, Y . Zhang, Y . Chen, S. Zheng, P. Zhao, and M. Tan, “Efficient test-time model adaptation without forgetting,” inInternational conference on machine learning. PMLR, 2022, pp. 16 888–16 905

    2022

  73. [77]

    Continual test-time domain adaptation,

    Q. Wang, O. Fink, L. Van Gool, and D. Dai, “Continual test-time domain adaptation,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 7201–7211

    2022

  74. [78]

    Test time training for industrial anomaly segmentation,

    A. Costanzino, P. Z. Ramirez, M. Del Moro, A. Aiezzo, G. Lisanti, S. Salti, and L. Di Stefano, “Test time training for industrial anomaly segmentation,” inProceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition, 2024, pp. 3910–3920

    2024

  75. [79]

    Stage: Segmentation-oriented industrial anomaly synthesis via graded diffusion with explicit mask alignment,

    X. Xu, Y . Wang, J. Wang, Q. Zhang, X. Lei, G. Xie, G. Jiang, and Z. Lu, “Stage: Segmentation-oriented industrial anomaly synthesis via graded diffusion with explicit mask alignment,”arXiv preprint arXiv:2509.06693, 2025

    arXiv 2025

  76. [80]

    Diffusion-tta: Test-time adaptation of discriminative models via gener- ative feedback,

    M. Prabhudesai, T.-W. Ke, A. Li, D. Pathak, and K. Fragkiadaki, “Diffusion-tta: Test-time adaptation of discriminative models via gener- ative feedback,”Advances in Neural Information Processing Systems, vol. 36, pp. 17 567–17 583, 2023

    2023

  77. [81]

    Topo-r1: Detecting topological anomalies via vision- language models,

    M. Xu, Q. Hu, X. Hu, S. Abousamra, X. Yu, W. Lyu, K. Qi, D. Samaras, and C. Chen, “Topo-r1: Detecting topological anomalies via vision- language models,”arXiv preprint arXiv:2603.13054, 2026

    Pith/arXiv arXiv 2026

  78. [82]

    A survey of vectorization methods in topological data analysis,

    D. Ali, A. Asaad, M.-J. Jimenez, V . Nanda, E. Paluzo-Hidalgo, and M. Soriano-Trigueros, “A survey of vectorization methods in topological data analysis,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 12, pp. 14 069–14 080, 2023

    2023

  79. [83]

    Momentum contrast for unsupervised visual representation learning,

    K. He, H. Fan, Y . Wu, S. Xie, and R. Girshick, “Momentum contrast for unsupervised visual representation learning,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020, pp. 9729–9738

    2020

  80. [84]

    A simple framework for contrastive learning of visual representations,

    T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A simple framework for contrastive learning of visual representations,” inInternational conference on machine learning. PmLR, 2020, pp. 1597–1607

    2020

Showing first 80 references.