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

arXiv: https://arxiv.org/abs/2609.34949 · HTML: https://mlsanity.com/paper/2609.34949/

## Key facts

- Cited by 1 papers in the map (data through 14 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.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)

## Builds on (most cited first, 30 of 30)

- [Cambrian-1: A Fully Open, Vision-Centric Exploration of Multimodal LLMs](https://mlsanity.com/paper/2406.16860/) (2024)
- [Eyes Wide Shut? Exploring the Visual Shortcomings of Multimodal LLMs](https://mlsanity.com/paper/2401.06209/) (2024)
- [SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentation](https://mlsanity.com/paper/2207.14315/) (2022)
- [WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation](https://mlsanity.com/paper/2303.14814/) (2023)
- [Multiresolution Knowledge Distillation for Anomaly Detection](https://mlsanity.com/paper/2011.11108/) (2020)
- [AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detection](https://mlsanity.com/paper/2310.18961/) (2023)
- [The MVTec 3D-AD Dataset for Unsupervised 3D Anomaly Detection and Localization](https://mlsanity.com/paper/2112.09045/) (2021)
- [Real-IAD: A Real-World Multi-View Dataset for Benchmarking Versatile Industrial Anomaly Detection](https://mlsanity.com/paper/2403.12580/) (2024)
- [Registration based Few-Shot Anomaly Detection](https://mlsanity.com/paper/2207.07361/) (2022)
- [AdaCLIP: Adapting CLIP with Hybrid Learnable Prompts for Zero-Shot Anomaly Detection](https://mlsanity.com/paper/2407.15795/) (2024)
- [PromptAD: Learning Prompts with only Normal Samples for Few-Shot Anomaly Detection](https://mlsanity.com/paper/2404.05231/) (2024)
- [Toward Generalist Anomaly Detection via In-context Residual Learning with Few-shot Sample Prompts](https://mlsanity.com/paper/2403.06495/) (2024)
- [DeepStack: Deeply Stacking Visual Tokens is Surprisingly Simple and Effective for LMMs](https://mlsanity.com/paper/2406.04334/) (2024)
- [A Unified Anomaly Synthesis Strategy with Gradient Ascent for Industrial Anomaly Detection and Localization](https://mlsanity.com/paper/2407.09359/) (2024)
- [FiLo: Zero-Shot Anomaly Detection by Fine-Grained Description and High-Quality Localization](https://mlsanity.com/paper/2404.13671/) (2024)
- [MMAD: A Comprehensive Benchmark for Multimodal Large Language Models in Industrial Anomaly Detection](https://mlsanity.com/paper/2410.09453/) (2024)
- [MANTA: A Large-Scale Multi-View and Visual-Text Anomaly Detection Dataset for Tiny Objects](https://mlsanity.com/paper/2412.04867/) (2024)
- [PKU-GoodsAD: A Supermarket Goods Dataset for Unsupervised Anomaly Detection and Segmentation](https://mlsanity.com/paper/2307.04956/) (2023)
- [MIAD: A Maintenance Inspection Dataset for Unsupervised Anomaly Detection](https://mlsanity.com/paper/2211.13968/) (2022)
- [IAD-R1: Reinforcing Consistent Reasoning in Industrial Anomaly Detection](https://mlsanity.com/paper/2508.09178/) (2025)
- [MetaUAS: Universal Anomaly Segmentation with One-Prompt Meta-Learning](https://mlsanity.com/paper/2505.09265/) (2025)
- [Real-IAD D3: A Real-World 2D/Pseudo-3D/3D Dataset for Industrial Anomaly Detection](https://mlsanity.com/paper/2504.14221/) (2025)
- [3CAD: A Large-Scale Real-World 3C Product Dataset for Unsupervised Anomaly](https://mlsanity.com/paper/2502.05761/) (2025)
- [Through the Lens of Contrast: Self-Improving Visual Reasoning in VLMs](https://mlsanity.com/paper/2603.02556/) (2026)
- [FiLo++: Zero-/Few-Shot Anomaly Detection by Fused Fine-Grained Descriptions and Deformable Localization](https://mlsanity.com/paper/2501.10067/) (2025)
- [AD-FM: Multimodal LLMs for Anomaly Detection via Multi-Stage Reasoning and Fine-Grained Reward Optimization](https://mlsanity.com/paper/2508.04175/) (2025)
- [CableInspect-AD: An Expert-Annotated Anomaly Detection Dataset](https://mlsanity.com/paper/2409.20353/) (2024)
- [JUDO: A Juxtaposed Domain-Oriented Multimodal Reasoner for Industrial Anomaly QA](https://mlsanity.com/paper/2605.20284/) (2026)
- [Texture-AD: An Anomaly Detection Dataset and Benchmark for Real Algorithm Development](https://mlsanity.com/paper/2409.06367/) (2024)
- [MMR-AD: A Large-Scale Multimodal Dataset for Benchmarking General Anomaly Detection with Multimodal Large Language Models](https://mlsanity.com/paper/2604.10971/) (2026)

## Built on by (most cited first, 1 of 1)

- [Anchor and Adapt: Asymmetric Prompt Adaptation for Few-Shot Industrial Anomaly Detection](https://mlsanity.com/paper/2610.07016/) (2026)

## 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.