Research map / Label-efficient and robust learning / OOD and anomaly detection
Towards Real Unsupervised Anomaly Detection Via Confident Meta-Learning
Muhammad Aqeel, Shakiba Sharifi, Marco Cristani, Francesco Setti
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 CoMet confident meta-learning to SimpleNet under 10% noisy training data (ablation) | SimpleNet · MVTec AD, 10% noise | image AUROC (%): 98.5 -> 99.4 | Improved |
| add CoMet to SimpleNet under 10% noisy training data (ablation) | SimpleNet · MVTec AD, 10% noise | pixel AUROC (%): 94.9 -> 96.2 | Improved |
| add CoMet to SimpleNet with clean training data (0% noise) (ablation) | SimpleNet · MVTec AD, 0% noise | image AUROC (%): 99.6 -> 99.7 | No clear effect |
| raise training noise from 0% to 10% without CoMet (ablation) | SimpleNet · MVTec AD | image AUROC (%): 99.6 -> 98.5 | Hurt |
| raise training noise from 0% to 10% without CoMet (ablation) | SimpleNet · MVTec AD | pixel AUROC (%): 98.1 -> 94.9 | Hurt |
| raise training noise from 0% to 10% with CoMet (ablation) | CoMet-SN · MVTec AD | image AUROC (%): 99.7 -> 99.4 | Hurt |
| add CoMet to SimpleNet (ablation) | SimpleNet · MVTec AD | precision and recall (%): 98.1 -> 99.3 precision, 98.9 -> 99.8 recall | Improved |
| add CoMet to SimpleNet (ablation) | SimpleNet · VIADUCT | precision and recall (%): 84.1 -> 87.4 precision, 93.6 -> 96.5 recall | Improved |
| replace DifferNet with CoMet-NF | DifferNet · MVTec AD | image AUROC (%): 94.9 -> 99.2 | Improved |
| replace DifferNet with CoMet-NF | DifferNet · MVTec AD | precision and recall (%): 95.6 -> 92.5 precision, 76.4 -> 93.4 recall | Mixed |
| replace DifferNet with CoMet-NF | DifferNet · VIADUCT | precision and recall (%): 79.1 -> 77.1 precision, 70.3 -> 90.8 recall | Mixed |
| replace DifferNet with CoMet-NF | DifferNet · KSDD2 | precision and recall (%): 90.9 -> 87.4 precision, 87.5 -> 94.3 recall | Mixed |
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.
- SimpleNet — SimpleNet: A Simple Network for Image Anomaly Detection and Localization (CVPR 2023) · better on 6 of 6 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
- Unsupervised Anomaly Detection Algorithms on Real-world Data: How Many Do We Need? (NeurIPS 2024 · similarity 0.90)
- UniNet: A Contrastive Learning-guided Unified Framework with Feature Selection for Anomaly Detection (CVPR 2025 · similarity 0.83)
- Dinomaly: The Less Is More Philosophy in Multi-Class Unsupervised Anomaly Detection (CVPR 2025 · similarity 0.81)
- Generating and Reweighting Dense Contrastive Patterns for Unsupervised Anomaly Detection (AAAI 2024 · similarity 0.81)
- ResAD: A Simple Framework for Class Generalizable Anomaly Detection (NeurIPS 2024 · similarity 0.81)
- Unlocking the Potential of Reverse Distillation for Anomaly Detection (AAAI 2025 · similarity 0.81)
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.
