Research map / Label-efficient and robust learning / OOD and anomaly detection
Focus the Discrepancy: Intra- and Inter-Correlation Learning for Image Anomaly Detection
Xincheng Yao, Ruoqi Li, Zefeng Qian, Yan Luo, Chongyang Zhang
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 entropy constraint to the intra-correlation branch (ablation) | FOD intra-correlation · MVTecAD, BTAD, MVTec3D-RGB average | image-level AUROC: 0.740 -> 0.817 | Improved |
| combine reconstruction and KL divergence scoring instead of divergence alone (intra branch) (ablation) | FOD intra-correlation · average over three datasets | image-level AUROC: 0.817 -> 0.915 | Improved |
| use mean features instead of coreset features as external reference (ablation) | FOD inter-correlation · average over three datasets | image-level AUROC: 0.836 -> 0.923 | Improved |
| add intra- and inter-correlation to patch-wise reconstruction Transformer (ablation) | FOD · average over three datasets | image-level AUROC: 0.905 -> 0.945 | Improved |
| add intra-correlation view to patch-wise reconstruction on MVTecAD (ablation) | FOD · MVTecAD | image-level AUROC: 0.972 -> 0.974 | No clear effect |
| add reconstruction to correlation-only scoring (Intra+Inter) (ablation) | FOD · average over three datasets | image-level AUROC: 0.877 -> 0.945 | Improved |
| FOD versus PatchCore using the same backbone | FOD · MVTecAD | image-level AUROC: 0.983 -> 0.992 | Improved |
| FOD versus best competitor CFLOW on harder MVTec3D-RGB | FOD · MVTec3D-RGB | image-level AUROC: 0.851 -> 0.884 | Improved |
| FOD versus CFLOW and PatchCore on anomaly localization | FOD · BTAD pixel-level | pixel-level AUROC: 0.978 -> 0.975 | No clear effect |
Closest papers by meaning
- Beyond Normal References: Discriminative Few-Shot Anomaly Detection (NeurIPS 2026 · similarity 0.87)
- ADPretrain: Advancing Industrial Anomaly Detection via Anomaly Representation Pretraining (NeurIPS 2025 · similarity 0.86)
- CNC: Cross-modal Normality Constraint for Unsupervised Multi-class Anomaly Detection (AAAI 2025 · similarity 0.86)
- Boosting Fine-Grained Visual Anomaly Detection with Coarse-Knowledge-Aware Adversarial Learning (AAAI 2025 · similarity 0.86)
- UniVAD: A Training-free Unified Model for Few-shot Visual Anomaly Detection (CVPR 2025 · similarity 0.86)
- Long-Tailed Anomaly Detection with Learnable Class Names (CVPR 2024 · similarity 0.85)
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.
