mlsanityData to 14 Oct 2026

ADOPD: Reference-Privileged On-Policy Distillation for MLLM-Based Industrial Anomaly Detection

arXiv 2608.09789 ↗First posted 10 Aug 2026cs.CVcited by 0

On the map: Knowledge distillation → Detection and dense prediction distillation

Abstract (arXiv)

Industrial anomaly detection (IAD) requires identifying fine-grained deviations from normal visual patterns. Multimodal large language models (MLLMs) can improve recognition accuracy by comparing query images with references at inference time, but these benefits rely on additional retrieval and processing. We investigate whether the benefits of reference comparison can instead be internalized in the model parameters. Access to references during training allows a reference-aware teacher to supervise a query-only student. However, the teacher may favor plausible responses based on query cues or language priors rather than valid visual information. We propose ADOPD, a reference-privileged on-policy distillation framework. The teacher evaluates student-generated rollouts under matched and mismatched references. The matched-reference teacher-to-student log-ratio defines the token-level learning direction, specifying what the student should learn. The likelihood gap between the two reference views estimates reference-specific support and calibrates the sequence-level weight. ADOPD achieves 77.31% average accuracy on the MMAD benchmark under zero-shot inference, improving the Qwen3-VL-4B backbone by 6.14 points and outperforming its one-shot setting by 2.64 points. Experiments show that ADOPD learns a fine-grained anomaly inspection strategy from reference comparison. The project will be available at https://github.com/withTai/ADOPD.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Detection and dense prediction distillation, one of 12 directions in Knowledge distillation. Builds on 25 of its 31 references that are in the map. First posted on arXiv 10 Aug 2026.

Builds on

25 of 31 references.

Closest ancestors · 1

Same direction, newest first

  1. 2026OPD-IAD: From Language Judgment to Industrial Anomaly Detection via On-Policy Self-Distillation21 Jul

Other references · 24

  1. 2025Qwen2.5-VL Technical Reportcited by 5,587
  2. 2023Improved Baselines with Visual Instruction Tuningcited by 4,418
  3. 2025Qwen3-VL Technical Reportcited by 2,793
  4. 2024HybridFlow: A Flexible and Efficient RLHF Frameworkcited by 2,057
  5. 2025InternVL3: Exploring Advanced Training and Test-Time Recipes for Open-Source Multimodal Modelscited by 1,834
  6. 2021Towards Total Recall in Industrial Anomaly Detectioncited by 776
  7. 2025GLM-4.5V and GLM-4.1V-Thinking: Towards Versatile Multimodal Reasoning with Scalable Reinforcement Learningcited by 466
  8. 2022SPot-the-Difference Self-Supervised Pre-training for Anomaly Detection and Segmentationcited by 400
  9. 2023AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detectioncited by 271
  10. 2023AnomalyGPT: Detecting Industrial Anomalies Using Large Vision-Language Modelscited by 171
  11. 2026Revisiting On-Policy Distillation: Empirical Failure Modes and Simple Fixescited by 145
  12. 2024Real-IAD: A Real-World Multi-View Dataset for Benchmarking Versatile Industrial Anomaly Detectioncited by 139
12 more
  1. 2026Self-Distilled RLVRcited by 131
  2. 2026Privileged Information Distillation for Language Modelscited by 90
  3. 2026Scaling Reasoning Efficiently via Relaxed On-Policy Distillationcited by 82
  4. 2026Vision-OPD: Learning to See Fine Details for Multimodal LLMs via On-Policy Self-Distillationcited by 77
  5. 2024MMAD: A Comprehensive Benchmark for Multimodal Large Language Models in Industrial Anomaly Detectioncited by 47
  6. 2025AnomalyR1: A GRPO-based End-to-end MLLM for Industrial Anomaly Detectioncited by 28
  7. 2023PKU-GoodsAD: A Supermarket Goods Dataset for Unsupervised Anomaly Detection and Segmentationcited by 21
  8. 2025IAD-R1: Reinforcing Consistent Reasoning in Industrial Anomaly Detectioncited by 17
  9. 2026V-Zero: Answer-Label-Free On-Policy Distillation with Contrastive Evidence Gating for Fine-Grained Visual Reasoningcited by 11
  10. 2026Visual Contrastive Self-Distillationcited by 10
  11. 2026Towards Explainable Industrial Anomaly Detection via Knowledge-Guided Latent Reasoningcited by 6
  12. 2026AD-Copilot: A Vision-Language Assistant for Industrial Anomaly Detection via Visual In-context Comparisoncited by 5

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Can Multimodal Large Language Models be Guided to Improve Industrial Anomaly Detection?
  2. 2025EMIT: Enhancing MLLMs for Industrial Anomaly Detection via Difficulty-Aware GRPO
  3. 2026InspectorGPT: A Comparative Reasoning Enhanced VLM for Comprehensive Industrial Anomaly Detection
  4. 2025IAD-GPT: Advancing Visual Knowledge in Multimodal Large Language Model for Industrial Anomaly Detection
  5. 2024VMAD: Visual-enhanced Multimodal Large Language Model for Zero-Shot Anomaly Detection
  6. 2026Industrial Anomaly Detection via Defect-Grounded Reasoning in Visual Latent Space
  7. 2026Global Logic and Local Search: Dual-Stream Multimodal In-Context Learning for Verifiable Industrial Anomaly Detection
  8. 2026A Structured Benchmark for Text-Guided Anomaly Detection: When Language Stops Conditioning the Decision