mlsanityData to 7 Oct 2026

OPD-IAD: From Language Judgment to Industrial Anomaly Detection via On-Policy Self-Distillation

arXiv 2607.18850 ↗First posted 21 Jul 2026cs.CV cs.AIcited by 1

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

Abstract (arXiv)

Large vision-language models (LVLMs) have shown strong potential for industrial anomaly detection (IAD) by providing image-level anomaly judgments and interpretable reasoning. However, reliably translating generated judgments into precise pixel-level localization remains challenging. We propose Mixed-Trajectory On-Policy Distillation for Language-Guided Industrial Anomaly Detection (MOPDA), the first framework to introduce on-policy self-distillation into LVLM-based IAD. For judgment learning, method introduces Mixed-Trajectory Supervision, combining student-generated on-policy trajectories with evidence-conditioned teacher trajectories under a shared token-level distillation objective. Student trajectories preserve supervision on deployment-relevant response paths, while teacher trajectories provide complementary evidence-conditioned supervision. For dense localization, Language-guided Visual Anchoring uses the final judgment as a compact semantic condition to construct image-specific normal and abnormal anchors, which are contrasted with dense visual features to produce anomaly maps. This keeps language as semantic guidance while grounding pixel-level responses in visual evidence. Under a strict cross-dataset zero-shot protocol on five IAD benchmarks, method outperforms the evaluated LVLM-based baselines on most detection, localization, and judgment metrics while remaining competitive with CLIP-based methods. Ablations further validate both mixed-trajectory supervision and final-judgment conditioning.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Placed in Chart understanding and anomaly detection, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 20 of its 29 references that are in the map. First posted on arXiv 21 Jul 2026.

Builds on

20 of 29 references.

Closest ancestors · 7

Same direction, newest first

  1. 2026Reasoning-Driven Anomaly Detection and Localization with Image-Level Supervision28 Mar
  2. 2026AG-VAS: Anchor-Guided Zero-Shot Visual Anomaly Segmentation with Large Multimodal Models1 Mar
  3. 2025Towards Zero-Shot Anomaly Detection and Reasoning with Multimodal Large Language Models11 Feb
  4. 2024MMAD: A Comprehensive Benchmark for Multimodal Large Language Models in Industrial Anomaly Detection12 Oct
  5. 2024VMAD: Visual-enhanced Multimodal Large Language Model for Zero-Shot Anomaly Detection30 Sep
  6. 2023Myriad: Large Multimodal Model by Applying Vision Experts for Industrial Anomaly Detection29 Oct
  7. 2023AnomalyGPT: Detecting Industrial Anomalies Using Large Vision-Language Models29 Aug

Other references · 13

  1. 2016V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentationcited by 2,284
  2. 2023On-Policy Distillation of Language Models: Learning from Self-Generated Mistakescited by 840
  3. 2023WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentationcited by 311
  4. 2023AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detectioncited by 270
  5. 2019Segmentation-Based Deep-Learning Approach for Surface-Defect Detectioncited by 145
  6. 2024AdaCLIP: Adapting CLIP with Hybrid Learnable Prompts for Zero-Shot Anomaly Detectioncited by 126
  7. 2021Mixed supervision for surface-defect detection: from weakly to fully supervised learningcited by 101
  8. 2024A Unified Anomaly Synthesis Strategy with Gradient Ascent for Industrial Anomaly Detection and Localizationcited by 74
  9. 2024MuSc: Zero-Shot Industrial Anomaly Classification and Segmentation with Mutual Scoring of the Unlabeled Imagescited by 52
  10. 2025AnomalyR1: A GRPO-based End-to-end MLLM for Industrial Anomaly Detectioncited by 28
  11. 2023PKU-GoodsAD: A Supermarket Goods Dataset for Unsupervised Anomaly Detection and Segmentationcited by 21
  12. 2025IAD-R1: Reinforcing Consistent Reasoning in Industrial Anomaly Detectioncited by 17
1 more
  1. 2026VisualAD: Language-Free Zero-Shot Anomaly Detection via Vision Transformercited by 10

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026ADOPD: Reference-Privileged On-Policy Distillation for MLLM-Based Industrial Anomaly Detectioncited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Industrial Anomaly Detection via Defect-Grounded Reasoning in Visual Latent Space
  2. 2026AD-Copilot: A Vision-Language Assistant for Industrial Anomaly Detection via Visual In-context Comparison
  3. 2026A Structured Benchmark for Text-Guided Anomaly Detection: When Language Stops Conditioning the Decision
  4. 2025IADGPT: Unified LVLM for Few-Shot Industrial Anomaly Detection, Localization, and Reasoning via In-Context Learning
  5. 2025IAD-GPT: Advancing Visual Knowledge in Multimodal Large Language Model for Industrial Anomaly Detection
  6. 2025LR-IAD:Mask-Free Industrial Anomaly Detection with Logical Reasoning
  7. 2026GenAU: Language-Grounded Industrial Anomaly Understanding with Vision-Language Models
  8. 2024Exploring Large Vision-Language Models for Robust and Efficient Industrial Anomaly Detection