mlsanityData to 7 Oct 2026

VD-DeepStack: Bridging Visual Comparison and Language Reasoning for Few-Shot Anomaly Detection

arXiv 2609.34949 ↗First posted 28 Sep 2026cs.AIcited by 0

On the map: LLM reasoning → Multimodal reasoning

Abstract (arXiv)

Few-shot visual anomaly detection is fundamentally a visual comparison task, requiring fine-grained inspection of a query against normal references. Many recent methods based on large vision-language models (LVLMs) emphasize comparative reasoning through language chain-of-thought. Yet discrete, abstract descriptions may underrepresent dense, fine-grained visual differences, leaving a gap between visual comparison and its expression in language. To address this gap, we propose Visual Difference DeepStack (VD-DeepStack), which explicitly conditions language reasoning on query-reference visual differences. Specifically, we fuse DINO features with the LVLM visual hierarchy to strengthen fine-grained representations, then construct dense difference evidence from residuals between query features and softly matched reference features. The difference-evidence path injects spatially weighted difference vectors into query-image states at multiple decoder depths, while an auxiliary visual-context path provides fine-grained appearance information to support their interpretation. Experiments on 4 industrial and 2 medical anomaly benchmarks demonstrate substantial improvements in few-shot anomaly detection over baselines relying on textual comparative reasoning. These results support mitigating the visual comparison-reasoning gap through the joint design of comparison representations and their integration into the decoder. Code will be released upon acceptance.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Multimodal reasoning, one of 17 directions in LLM reasoning. Builds on 30 of its 37 references that are in the map. First posted on arXiv 28 Sep 2026.

Builds on

30 of 37 references.

Closest ancestors · 1

Same direction, newest first

  1. 2026Through the Lens of Contrast: Self-Improving Visual Reasoning in VLMs3 Mar

Other references · 29

  1. 2024Cambrian-1: A Fully Open, Vision-Centric Exploration of Multimodal LLMscited by 822
  2. 2024Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMscited by 719
  3. 2022SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentationcited by 397
  4. 2023WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentationcited by 311
  5. 2020Multiresolution Knowledge Distillation for Anomaly Detectioncited by 285
  6. 2023AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detectioncited by 270
  7. 2021The MVTec 3D-AD Dataset for Unsupervised 3D Anomaly Detection and Localizationcited by 174
  8. 2024Real-IAD: A Real-World Multi-View Dataset for Benchmarking Versatile Industrial Anomaly Detectioncited by 138
  9. 2022Registration based Few-Shot Anomaly Detectioncited by 133
  10. 2024AdaCLIP: Adapting CLIP with Hybrid Learnable Prompts for Zero-Shot Anomaly Detectioncited by 126
  11. 2024PromptAD: Learning Prompts with only Normal Samples for Few-Shot Anomaly Detectioncited by 97
  12. 2024Toward Generalist Anomaly Detection via In-context Residual Learning with Few-shot Sample Promptscited by 88
17 more
  1. 2024A Unified Anomaly Synthesis Strategy with Gradient Ascent for Industrial Anomaly Detection and Localizationcited by 74
  2. 2024DeepStack: Deeply Stacking Visual Tokens is Surprisingly Simple and Effective for LMMscited by 72
  3. 2024FiLo: Zero-Shot Anomaly Detection by Fine-Grained Description and High-Quality Localizationcited by 51
  4. 2024MMAD: A Comprehensive Benchmark for Multimodal Large Language Models in Industrial Anomaly Detectioncited by 46
  5. 2024MANTA: A Large-Scale Multi-View and Visual-Text Anomaly Detection Dataset for Tiny Objectscited by 26
  6. 2023PKU-GoodsAD: A Supermarket Goods Dataset for Unsupervised Anomaly Detection and Segmentationcited by 21
  7. 2022MIAD: A Maintenance Inspection Dataset for Unsupervised Anomaly Detectioncited by 19
  8. 2025IAD-R1: Reinforcing Consistent Reasoning in Industrial Anomaly Detectioncited by 17
  9. 2025Real-IAD D3: A Real-World 2D/Pseudo-3D/3D Dataset for Industrial Anomaly Detectioncited by 12
  10. 2025MetaUAS: Universal Anomaly Segmentation with One-Prompt Meta-Learningcited by 12
  11. 20253CAD: A Large-Scale Real-World 3C Product Dataset for Unsupervised Anomalycited by 10
  12. 2025AD-FM: Multimodal LLMs for Anomaly Detection via Multi-Stage Reasoning and Fine-Grained Reward Optimizationcited by 8
  13. 2025FiLo++: Zero-/Few-Shot Anomaly Detection by Fused Fine-Grained Descriptions and Deformable Localizationcited by 7
  14. 2024CableInspect-AD: An Expert-Annotated Anomaly Detection Datasetcited by 5
  15. 2026JUDO: A Juxtaposed Domain-Oriented Multimodal Reasoner for Industrial Anomaly QAcited by 4
  16. 2024Texture-AD: An Anomaly Detection Dataset and Benchmark for Real Algorithm Developmentcited by 2
  17. 2026MMR-AD: A Large-Scale Multimodal Dataset for Benchmarking General Anomaly Detection with Multimodal Large Language Modelscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025From Benchmarks to Reality: Advancing Visual Anomaly Detection by the VAND 3.0 Challenge
  2. 2026VLM-SubtleBench: How Far Are VLMs from Human-Level Subtle Comparative Reasoning?
  3. 2024Do LLMs Understand Visual Anomalies? Uncovering LLM's Capabilities in Zero-shot Anomaly Detection
  4. 2026H2VLR: Heterogeneous Hypergraph Vision-Language Reasoning for Few-Shot Anomaly Detection
  5. 2026AnomalyClaw: A Universal Visual Anomaly Detection Agent via Tool-Grounded Refutation
  6. 2025Towards Fine-Grained Vision-Language Alignment for Few-Shot Anomaly Detection
  7. 2026VisualAD: Language-Free Zero-Shot Anomaly Detection via Vision Transformer
  8. 2024Vision-Language Models Assisted Unsupervised Video Anomaly Detection