Research map / Detection and segmentation / Open-set and few-shot detection
Prototypical Residual Networks for Anomaly Detection and Localization
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 multi-scale prototypes (ablation) | PRN · MVTec AD | Pixel-AUROC: 91.7 -> 98.5 | Helped |
| Add multi-size self-attention (ablation) | PRN with multi-scale prototypes and fusion · MVTec AD | Pixel-AP: 77.0 -> 78.6 | Helped |
| Add multi-scale fusion (ablation) | PRN with multi-scale prototypes and self-attention · MVTec AD | Pixel-AUROC: 98.5 -> 99.0 | Helped |
| Add extended anomalies (ablation) | PRN anomaly generation · MVTec AD | Image-AUROC: 98.7 -> 99.4 | Helped |
| Add heterologous anomalies (ablation) | PRN anomaly generation with EA and TA · MVTec AD | Pixel-AUROC: 97.2 -> 98.4 | Helped |
| Add homologous anomalies (ablation) | PRN anomaly generation with EA and TA · MVTec AD | PRO: 93.4 -> 95.7 | Helped |
| Add target-area constraints (ablation) | PRN anomaly generation with EA, HEA and HOA · MVTec AD | Image-AUROC: 98.4 -> 99.4 | Helped |
| Add extended anomalies (ablation) | PRN anomaly generation with HEA, HOA and TA · MVTec AD | Image-AUROC: 98.7 -> 99.4 | Helped |
| Set prototype proportion to 5% instead of 10% (ablation) | PRN · MVTec AD | Image-AUROC: 99.4 -> 99.2 | Hurt |
| Set prototype proportion to 20% instead of 10% (ablation) | PRN · MVTec AD | Image-AUROC: 99.4 -> 99.2 | Hurt |
| Use every normal sample as a prototype (100%) instead of 10% (ablation) | PRN · MVTec AD | Image-AUROC: 99.4 -> 86.2 | Hurt |
| Use fewer seen anomalies (1 instead of 10) (ablation) | PRN · MVTec AD | Image-AUROC: 99.4 -> 98.8 | Hurt |
Closest papers by meaning
- RealNet: A Feature Selection Network with Realistic Synthetic Anomaly for Anomaly Detection (CVPR 2024 · similarity 0.89)
- ADPretrain: Advancing Industrial Anomaly Detection via Anomaly Representation Pretraining (NeurIPS 2025 · similarity 0.88)
- ResAD: A Simple Framework for Class Generalizable Anomaly Detection (NeurIPS 2024 · similarity 0.88)
- AnomalyNCD: Towards Novel Anomaly Class Discovery in Industrial Scenarios (CVPR 2025 · similarity 0.88)
- Normal-Abnormal Guided Generalist Anomaly Detection (NeurIPS 2025 · similarity 0.87)
- AnomalyGPT: Detecting Industrial Anomalies Using Large Vision-Language Models (AAAI 2024 · similarity 0.86)
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.