Research map / Label-efficient and robust learning / OOD and anomaly detection
A Comprehensive Analysis about Unsupervised Outlier Detection for Images
Zhonghang Liu, Siyuan Chen, Jingwen Yu, Changshuo Wang, Kunyang Li, Jiangbo Lu
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 VUOD feature enhancement (pseudo labels + contrastive learning) before OCSVM (ablation) | OCSVM · CIFAR-10 | AUC: 0.824 -> 0.972 | Improved |
| add VUOD feature enhancement before OCSVM (ablation) | OCSVM · MPDD | AUC: 0.575 -> 0.916 | Improved |
| add VUOD feature enhancement before OCSVM (ablation) | OCSVM · MVTec-AD | AUC: 0.754 -> 0.921 | Improved |
| add VUOD feature enhancement before OCSVM (ablation) | OCSVM · Caltech-101 | AUC: 0.955 -> 0.994 | Improved |
| add VUOD feature enhancement before OCSVM (ablation) | OCSVM · BrainMRI | AUC: 0.605 -> 0.797 | Improved |
| use Affinity Propagation instead of k-Means with the same cluster count (ablation) | clustering stage · 10,000 ResNet-50 features | seconds: 207.3 -> 84.60 | Improved |
| use refined distance (Eq. 6) instead of mean distance baseline (ablation) | global separability · STL-10 across outlier ratios | AUC | Improved |
| shift thresholds (eps+, eps-) from default (0.9, 0.5) (ablation) | VUOD · CIFAR-10 | AUC: 0.972 -> 0.963 | No clear effect |
| filter training set with VUOD | ResNet-18 · Imagenette | top-1 accuracy (%): 90.81% -> 91.45% | Improved |
| filter training set with VUOD | ResNet-18 · Imagenette | top-5 accuracy (%): 99.11% -> 99.39% | Improved |
| use VUOD instead of RobustVGGT for outlier image filtering | VGGT · 3D reconstruction input filtering | ms per image: 118 ms -> 27 ms | Improved |
| use VUOD instead of EfficientAD for defect scoring | industrial inspection · MVTec-AD bottle | seconds: 42.10 -> 17.18 | Improved |
Closest papers by meaning
- E3 Outlier: a Self-Supervised Framework for Unsupervised Deep Outlier Detection (TPAMI 2022 · similarity 0.90)
- RODEO: Robust Outlier Detection via Exposing Adaptive Out-of-Distribution Samples (ICML 2024 · similarity 0.87)
- Quantile-Based Maximum Likelihood Training for Outlier Detection (AAAI 2024 · similarity 0.87)
- FlexUOD: The Answer to Real-world Unsupervised Image Outlier Detection (CVPR 2025 · similarity 0.87)
- ODIM: Outlier Detection via Likelihood of Under-Fitted Generative Models (ICML 2024 · similarity 0.86)
- Diffusion-based Semantic Outlier Generation via Nuisance Awareness for Out-of-Distribution Detection (AAAI 2025 · 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.
