Research map / Label-efficient and robust learning / OOD and anomaly detection
CutPaste: Self-Supervised Learning for Anomaly Detection and Localization
Check 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 |
|---|---|---|---|
| Use CutPaste augmentation rather than Cutout (ablation) | Self-supervised defect detector · MVTec AD | image-level AUC: 71.3 -> 90.9 | Helped |
| Add CutPaste-Scar as a separate prediction class alongside CutPaste (ablation) | CutPaste defect detector · MVTec AD | image-level AUC: 90.9 -> 95.2 | Helped |
| Model CutPaste and CutPaste-Scar as separate classes rather than one union class (ablation) | CutPaste defect detector · MVTec AD | image-level AUC: 94.3 -> 95.2 | Helped |
| Replace standard Cutout fill with mean pixel values (ablation) | Self-supervised defect detector · MVTec AD | image-level AUC: 71.3 -> 73.9 | Helped |
| Replace standard Cutout fill with random colors (ablation) | Self-supervised defect detector · MVTec AD | image-level AUC: 71.3 -> 76.1 | Helped |
| Use Confetti noise instead of Cutout (ablation) | Self-supervised defect detector · MVTec AD | image-level AUC: 71.3 -> 77.8 | Helped |
| Fine-tune ImageNet-pretrained EfficientNet-B4 using CutPaste (3-way) (ablation) | EfficientNet-B4 anomaly detector · MVTec AD | image-level AUC, pool feature: 94.5 -> 96.6 | Helped |
| Fine-tune ImageNet-pretrained EfficientNet-B4 using CutPaste (3-way) (ablation) | EfficientNet-B4 anomaly detector · MVTec AD | image-level AUC, level-7 feature: 96.8 -> 97.1 | Helped |
| Use patch-level representations for localization instead of image-level GradCAM (ablation) | CutPaste localization model · MVTec AD | pixel-level localization AUC: 88.3 -> 96.0 | Helped |
| Use CutPaste (3-way) rather than Cutout for synthetic anomalies (ablation) | Anomaly detector · Synthetic anomaly detection | AUC: 81.5 -> 98.3 | Helped |
| Use CutPaste rather than Cutout (ablation) | Semantic anomaly detector · CIFAR-10 | AUC: 60.2 -> 69.4 | Helped |
| Use rotation prediction rather than CutPaste | Semantic anomaly detector · CIFAR-10 | AUC: 69.4 -> 91.3 | Helped |
Closest papers by meaning
- Boosting Fine-Grained Visual Anomaly Detection with Coarse-Knowledge-Aware Adversarial Learning (AAAI 2025 · similarity 0.84)
- AnomalyNCD: Towards Novel Anomaly Class Discovery in Industrial Scenarios (CVPR 2025 · similarity 0.84)
- A Diffusion-Based Framework for Multi-Class Anomaly Detection (AAAI 2024 · similarity 0.84)
- UniVAD: A Training-free Unified Model for Few-shot Visual Anomaly Detection (CVPR 2025 · similarity 0.84)
- A Comprehensive Augmentation Framework for Anomaly Detection (AAAI 2024 · similarity 0.84)
- Long-Tailed Anomaly Detection with Learnable Class Names (CVPR 2024 · similarity 0.84)
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.