Research map / Label-efficient and robust learning / OOD and anomaly detection
ResAD++: Towards Class Agnostic Anomaly Detection via Residual Feature Learning
Xincheng Yao, Chao Shi, Muming Zhao, Guangtao Zhai, C L 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 |
|---|---|---|---|
| use residual features instead of initial features (ablation) | ResAD++ · MVTecAD | kurtosis: 0.261 -> 1.679 | Improved |
| use residual features instead of initial features (ablation) | ResAD++ · VisA | kurtosis: 0.757 -> 1.313 | Improved |
| add feature hypersphere constraining to residual features (ablation) | ResAD++ · MVTecAD | feature scale std across classes: 0.147 -> 0.005 | Improved |
| add feature hypersphere constraining to residual features (ablation) | ResAD++ · VisA | feature scale std across classes: 0.098 -> 0.004 | Improved |
| use ImageBind instead of WideResNet50 as feature extractor (ablation) | ResAD++ · Average over 8 AD datasets, 2-shot | image-level AUROC: 78.1 -> 82.1 | Improved |
| use ImageBind instead of WideResNet50 as feature extractor (ablation) | ResAD++ · Kvasir-Seg, 8-shot | image-level AUROC: 90.5 -> 83.4 | Hurt |
| increase reference shots from 2 to 8 | ResAD++ (W50) · Average over 8 AD datasets | image-level AUROC: 78.1 -> 83.7 | Improved |
| increase reference shots from 2 to 8 | ResAD++ (ImageBind) · Average over 8 AD datasets | pixel-level AUROC: 91.2 -> 92.1 | Improved |
| add log-barrier contraction, VQ matching and new loss over ResAD (ImageBind) | ResAD++ · Average over 8 AD datasets, 2-shot | image-level AUROC: 79.6 -> 82.1 | Improved |
| add log-barrier contraction, VQ matching and new loss over ResAD (ImageBind) | ResAD++ · Average over 8 AD datasets, 8-shot | PRO: 80.0 -> 80.9 | Improved |
Closest papers by meaning
- ResAD: A Simple Framework for Class Generalizable Anomaly Detection (NeurIPS 2024 · similarity 0.97)
- Long-Tailed Anomaly Detection with Learnable Class Names (CVPR 2024 · similarity 0.90)
- AnomalyNCD: Towards Novel Anomaly Class Discovery in Industrial Scenarios (CVPR 2025 · similarity 0.88)
- ADPretrain: Advancing Industrial Anomaly Detection via Anomaly Representation Pretraining (NeurIPS 2025 · similarity 0.88)
- Anomaly Heterogeneity Learning for Open-set Supervised Anomaly Detection (CVPR 2024 · similarity 0.88)
- Normal-Abnormal Guided Generalist Anomaly Detection (NeurIPS 2025 · 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.
