mlsanityData to 14 Oct 2026

TED:Text-Axis Evidence Decomposition for Prompted Anomaly Localization

arXiv 2609.39033 ↗First posted 30 Sep 2026cs.CV cs.AIcited by 0

On the map: Multimodal LLMs / vision-language models → Chart understanding and anomaly detection

Abstract (arXiv)

CLIP is a powerful vision-language model, but it was not designed for fine-grained defect localization; CLIP-based anomaly detectors therefore adapt it with prompts or lightweight modules to increase defect sensitivity. We show that stronger sensitivity does not necessarily make local evidence reliable: under domain shift, adapted CLIP-AD models often assign high anomaly scores to both true defects and visually complex normal regions. The issue is not simply missing defect information, but a local scoring rule that decodes defect and hard-normal evidence, having the same anomaly evidence. We propose TED (Text-Axis Evidence Decomposition), a post-hoc scoring method that asks whether each ambiguous response is better supported by source defect patches or by source normal patches mistaken as anomalous. TED compares these supports under the host's normal-versus-anomaly text response, leaves the backbone and prompts unchanged, and requires no target-domain training. It works as a train-free score for raw VLM backbones or as a source-calibrated residual correction for adapted CLIP-AD hosts. Across frozen VLM backbones, TED substantially improves pixel-level localization over raw prompt similarity; across adapted hosts, it improves most pixel-level settings over P-AUROC, P-PRO, and P-AP. Gains are largest under stronger hard-FP competition, with mean localization gain increasing from +5.0 in low-competition regimes to about +10.9 in mid/high-competition regimes. These results suggest that recoverable defect evidence can already exist in pretrained multimodal representations, but reliable localization requires decoding it against hard-normal competitors. Code will be released at TED GitHub repository.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Chart understanding and anomaly detection, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 35 of its 41 references that are in the map. First posted on arXiv 30 Sep 2026.

Builds on

35 of 41 references.

Closest ancestors · 3

Same direction, newest first

  1. 2025AD-FM: Multimodal LLMs for Anomaly Detection via Multi-Stage Reasoning and Fine-Grained Reward Optimization6 Aug
  2. 2025Towards Zero-Shot Anomaly Detection and Reasoning with Multimodal Large Language Models11 Feb
  3. 2024MMAD: A Comprehensive Benchmark for Multimodal Large Language Models in Industrial Anomaly Detection12 Oct

Other references · 32

  1. 2021Learning Transferable Visual Models From Natural Language Supervisioncited by 33,744
  2. 2021Learning to Prompt for Vision-Language Modelscited by 2,281
  3. 2023ImageBind: One Embedding Space To Bind Them Allcited by 1,175
  4. 2023Vision-Language Models for Vision Tasks: A Surveycited by 839
  5. 2020PaDiM: a Patch Distribution Modeling Framework for Anomaly Detection and Localizationcited by 594
  6. 2021DRAEM -- A discriminatively trained reconstruction embedding for surface anomaly detectioncited by 514
  7. 2022Anomaly Detection via Reverse Distillation from One-Class Embeddingcited by 421
  8. 2022SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentationcited by 400
  9. 2019Uninformed Students: Student-Teacher Anomaly Detection with Discriminative Latent Embeddingscited by 386
  10. 2021CFLOW-AD: Real-Time Unsupervised Anomaly Detection with Localization via Conditional Normalizing Flowscited by 321
  11. 2023WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentationcited by 315
  12. 2020Sub-Image Anomaly Detection with Deep Pyramid Correspondencescited by 300
20 more
  1. 2021DeepViT: Towards Deeper Vision Transformercited by 289
  2. 2023AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detectioncited by 271
  3. 2021FastFlow: Unsupervised Anomaly Detection and Localization via 2D Normalizing Flowscited by 227
  4. 2021VT-ADL: A Vision Transformer Network for Image Anomaly Detection and Localizationcited by 223
  5. 2023Deep Industrial Image Anomaly Detection: A Surveycited by 148
  6. 2024AdaCLIP: Adapting CLIP with Hybrid Learnable Prompts for Zero-Shot Anomaly Detectioncited by 127
  7. 2021Self-Supervised Predictive Convolutional Attentive Block for Anomaly Detectioncited by 119
  8. 2021Fully Convolutional Cross-Scale-Flows for Image-based Defect Detectioncited by 117
  9. 2024RealNet: A Feature Selection Network with Realistic Synthetic Anomaly for Anomaly Detectioncited by 116
  10. 2020Modeling the Distribution of Normal Data in Pre-Trained Deep Features for Anomaly Detectioncited by 115
  11. 2021Divide-and-Assemble: Learning Block-wise Memory for Unsupervised Anomaly Detectioncited by 83
  12. 2025AA-CLIP: Enhancing Zero-shot Anomaly Detection via Anomaly-Aware CLIPcited by 73
  13. 2025The MVTec AD 2 Dataset: Advanced Scenarios for Unsupervised Anomaly Detectioncited by 30
  14. 2025AdaptCLIP: Adapting CLIP for Universal Visual Anomaly Detectioncited by 29
  15. 2025AnomalyR1: A GRPO-based End-to-end MLLM for Industrial Anomaly Detectioncited by 28
  16. 2024Fine-grained Abnormality Prompt Learning for Zero-shot Anomaly Detectioncited by 22
  17. 2022A Survey on Unsupervised Anomaly Detection Algorithms for Industrial Imagescited by 21
  18. 2025OmniAD: Detect and Understand Industrial Anomaly via Multimodal Reasoningcited by 18
  19. 2025Continual-MEGA: A Large-scale Benchmark for Generalizable Continual Anomaly Detectioncited by 4
  20. 2025GenCLIP: Generalizing CLIP Prompts for Zero-shot Anomaly Detectioncited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Defect-aware Hybrid Prompt Optimization via Progressive Tuning for Zero-Shot Multi-type Anomaly Detection and Segmentation
  2. 2024Do LLMs Understand Visual Anomalies? Uncovering LLM's Capabilities in Zero-shot Anomaly Detection
  3. 2026TIPS Over Tricks: Simple Prompts for Effective Zero-shot Anomaly Detection
  4. 2026Proximity-CLIP: Text-Guided Semantic Proximity Learning for Zero-Shot Anomaly Detection
  5. 2024Dual-Image Enhanced CLIP for Zero-Shot Anomaly Detection
  6. 2026DevPrompt: Deviation-Based Prompt Learning for One-Normal ShotImage Anomaly Detection
  7. 2026Reasoning-Driven Anomaly Detection and Localization with Image-Level Supervision
  8. 2026VisualAD: Language-Free Zero-Shot Anomaly Detection via Vision Transformer