mlsanityData to 7 Oct 2026

MuSc-V2: Zero-Shot Multimodal Industrial Anomaly Classification and Segmentation with Mutual Scoring of Unlabeled Samples

arXiv 2511.10047 ↗First posted 13 Nov 2025cs.CVcited by 3

Not on the map. Why?

Abstract (arXiv)

Zero-shot anomaly classification (AC) and segmentation (AS) methods aim to identify and outline defects without using any labeled samples. In this paper, we reveal a key property that is overlooked by existing methods: normal image patches across industrial products typically find many other similar patches, not only in 2D appearance but also in 3D shapes, while anomalies remain diverse and isolated. To explicitly leverage this discriminative property, we propose a Mutual Scoring framework (MuSc-V2) for zero-shot AC/AS, which flexibly supports single 2D/3D or multimodality. Specifically, our method begins by improving 3D representation through Iterative Point Grouping (IPG), which reduces false positives from discontinuous surfaces. Then we use Similarity Neighborhood Aggregation with Multi-Degrees (SNAMD) to fuse 2D/3D neighborhood cues into more discriminative multi-scale patch features for mutual scoring. The core comprises a Mutual Scoring Mechanism (MSM) that lets samples within each modality to assign score to each other, and Cross-modal Anomaly Enhancement (CAE) that fuses 2D and 3D scores to recover modality-specific missing anomalies. Finally, Re-scoring with Constrained Neighborhood (RsCon) suppresses false classification based on similarity to more representative samples. Our framework flexibly works on both the full dataset and smaller subsets with consistently robust performance, ensuring seamless adaptability across diverse product lines. In aid of the novel framework, MuSc-V2 achieves significant performance improvements: a +23.7% AP gain on the MVTec 3D-AD dataset and a +19.3% boost on the Eyecandies dataset, surpassing previous zero-shot benchmarks and even outperforming most few-shot methods. The code will be available at The code will be available at hrefhttps://github.com/HUST-SLOW/MuSc-V2https://github.com/HUST-SLOW/MuSc-V2.

Key facts. Cited by 3 papers in the map (data through 7 Oct 2026). Builds on 52 of its 72 references that are in the map. First posted on arXiv 13 Nov 2025.

Builds on

52 of 72 references.

References · 52

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,456
  2. 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
  3. 2021Swin Transformer: Hierarchical Vision Transformer using Shifted Windowscited by 12,082
  4. 2021Emerging Properties in Self-Supervised Vision Transformerscited by 6,893
  5. 2016PointNet: Deep Learning on Point Sets for 3D Classification and Segmentationcited by 5,223
  6. 2020Point Transformercited by 1,462
  7. 2021Towards Total Recall in Industrial Anomaly Detectioncited by 772
  8. 2020PCT: Point cloud transformercited by 730
  9. 2021Point-BERT: Pre-training 3D Point Cloud Transformers with Masked Point Modelingcited by 647
  10. 2023Point Transformer V3: Simpler, Faster, Strongercited by 579
  11. 2022Masked Autoencoders for Point Cloud Self-supervised Learningcited by 456
  12. 2022SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentationcited by 397
38 more
  1. 2022Point Transformer V2: Grouped Vector Attention and Partition-based Poolingcited by 384
  2. 2023WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentationcited by 311
  3. 2023AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detectioncited by 270
  4. 2022ULIP: Learning a Unified Representation of Language, Images, and Point Clouds for 3D Understandingcited by 262
  5. 2021VT-ADL: A Vision Transformer Network for Image Anomaly Detection and Localizationcited by 222
  6. 2023ULIP-2: Towards Scalable Multimodal Pre-training for 3D Understandingcited by 199
  7. 2022PointCLIP V2: Prompting CLIP and GPT for Powerful 3D Open-world Learningcited by 186
  8. 2021The MVTec 3D-AD Dataset for Unsupervised 3D Anomaly Detection and Localizationcited by 174
  9. 2023Multimodal Industrial Anomaly Detection via Hybrid Fusioncited by 138
  10. 2023APRIL-GAN: A Zero-/Few-Shot Anomaly Classification and Segmentation Method for CVPR 2023 VAND Workshop Challenge Tracks 1&2: 1st Place on Zero-shot AD and 4th Place on Few-shot ADcited by 128
  11. 2024AdaCLIP: Adapting CLIP with Hybrid Learnable Prompts for Zero-Shot Anomaly Detectioncited by 126
  12. 2022VLT: Vision-Language Transformer and Query Generation for Referring Segmentationcited by 110
  13. 2024PromptAD: Learning Prompts with only Normal Samples for Few-Shot Anomaly Detectioncited by 97
  14. 2022Back to the Feature: Classical 3D Features are (Almost) All You Need for 3D Anomaly Detectioncited by 92
  15. 2024Toward Generalist Anomaly Detection via In-context Residual Learning with Few-shot Sample Promptscited by 88
  16. 2023Pushing the Limits of Fewshot Anomaly Detection in Industry Vision: Graphcorecited by 75
  17. 2024A Unified Anomaly Synthesis Strategy with Gradient Ascent for Industrial Anomaly Detection and Localizationcited by 74
  18. 2024Adapting Visual-Language Models for Generalizable Anomaly Detection in Medical Imagescited by 69
  19. 2023Towards Scalable 3D Anomaly Detection and Localization: A Benchmark via 3D Anomaly Synthesis and A Self-Supervised Learning Networkcited by 67
  20. 2023Collaborative Discrepancy Optimization for Reliable Image Anomaly Localizationcited by 56
  21. 2022The Eyecandies Dataset for Unsupervised Multimodal Anomaly Detection and Localizationcited by 55
  22. 2024MuSc: Zero-Shot Industrial Anomaly Classification and Segmentation with Mutual Scoring of the Unlabeled Imagescited by 52
  23. 2024FiLo: Zero-Shot Anomaly Detection by Fine-Grained Description and High-Quality Localizationcited by 51
  24. 2024VCP-CLIP: A visual context prompting model for zero-shot anomaly segmentationcited by 49
  25. 2023DiffusionAD: Norm-guided One-step Denoising Diffusion for Anomaly Detectioncited by 46
  26. 2024R3D-AD: Reconstruction via Diffusion for 3D Anomaly Detectioncited by 45
  27. 2023EasyNet: An Easy Network for 3D Industrial Anomaly Detectioncited by 43
  28. 2023Multimodal Industrial Anomaly Detection by Crossmodal Feature Mappingcited by 41
  29. 2023Vote2Cap-DETR++: Decoupling Localization and Describing for End-to-End 3D Dense Captioningcited by 38
  30. 2022Self-Supervised Masked Convolutional Transformer Block for Anomaly Detectioncited by 34
  31. 2024PointAD: Comprehending 3D Anomalies from Points and Pixels for Zero-shot 3D Anomaly Detectioncited by 29
  32. 2024M3DM-NR: RGB-3D Noisy-Resistant Industrial Anomaly Detection via Multimodal Denoisingcited by 24
  33. 2023Target before Shooting: Accurate Anomaly Detection and Localization under One Millisecond via Cascade Patch Retrievalcited by 22
  34. 2024Fine-grained Abnormality Prompt Learning for Zero-shot Anomaly Detectioncited by 22
  35. 2022Flattening-Net: Deep Regular 2D Representation for 3D Point Cloud Analysiscited by 19
  36. 2024Prior Normality Prompt Transformer for Multi-class Industrial Image Anomaly Detectioncited by 19
  37. 2024Towards Zero-shot 3D Anomaly Localizationcited by 8
  38. 2025Center-aware Residual Anomaly Synthesis for Multi-class Industrial Anomaly Detectioncited by 7

50 most cited of 52 All 52 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026Open-Set Supervised 3D Anomaly Detection: An Industrial Dataset and a Generalisable Framework for Unknown Defectscited by 0
  2. 2026CoGeoAD: Hierarchical Color-Geometric Fusion with Multi-View Attention for Zero-Shot 3D Anomaly Detectioncited by 0
  3. 2026When Point Clouds Outperform Pixels: Rethinking Zero-Shot Multimodal Anomaly Detectioncited by 0