Research map / Label-efficient and robust learning / OOD and anomaly detection
FastRef: Fast Prototype Refinement for Few-shot Industrial Anomaly Detection
Yufei Li, Long Tian, Yuyang Dai, Wenchao Chen, Liang Bao, Xiyang Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add FastRef prototype refinement | PatchCore · VisA 2-shot | image AUROC: 80.0 -> 87.1 | Improved |
| add FastRef | PatchCore · MPDD 2-shot | image AUROC: 71.4 -> 78.2 | Improved |
| add FastRef | WinCLIP · MVTec 2-shot | image AUROC: 93.7 -> 93.9 | No clear effect |
| add FastRef | WinCLIP · MVTec 2-shot | pixel AUROC: 93.8 -> 96.2 | Improved |
| add FastRef | AnomalyDINO · MVTec 2-shot | image AUROC: 96.7 -> 97.2 | No clear effect |
| add FastRef | AnomalyDINO · MPDD 2-shot | image AUROC: 75.2 -> 78.4 | Improved |
| add FastRef | WinCLIP · MPDD 4-shot | image AUROC: 75.0 -> 82.0 | Improved |
| add FastRef | PatchCore · MPDD 4-shot pixel | pixel AUROC: 97.2 -> 97.2 | No clear effect |
| add characteristic transfer W* (ablation) | WinCLIP+ · MPDD 2-shot | image AUROC: 72.5 -> 74.9 | Improved |
| add anomaly suppression T* on top of W* (ablation) | WinCLIP+ · MPDD 2-shot | image AUROC: 74.9 -> 76.0 | Improved |
| add anomaly suppression T* (ablation) | WinCLIP+ · MVTec/VisA/MPDD 2-shot | image AUROC: 93.8 -> 93.9; 84.0 -> 84.1; 74.9 -> 76.0 | Improved |
| use OT regularizer instead of mean-prototype Gaussian assumption (FastRecon form) (ablation) | FastRef · MPDD 2-shot | image AUROC: 76.4 -> 78.2 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PromptAD — PromptAD: Learning Prompts with only Normal Samples for Few-Shot Anomaly Detection (CVPR 2024) · better on 3 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- FastRecon: Few-shot Industrial Anomaly Detection via Fast Feature Reconstruction (ICCV 2023 · similarity 0.93)
- ReMP-AD: Retrieval-enhanced Multi-modal Prompt Fusion for Few-Shot Industrial Visual Anomaly Detection (ICCV 2025 · similarity 0.89)
- ReinAD: Towards Real-world Industrial Anomaly Detection with a Comprehensive Contrastive Dataset (NeurIPS 2025 · similarity 0.89)
- Exploring Intrinsic Normal Prototypes within a Single Image for Universal Anomaly Detection (CVPR 2025 · similarity 0.88)
- Real-IAD: A Real-World Multi-View Dataset for Benchmarking Versatile Industrial Anomaly Detection (CVPR 2024 · similarity 0.88)
- AnomalyGPT: Detecting Industrial Anomalies Using Large Vision-Language Models (AAAI 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
