Research map / Vision-language and multimodal / CLIP and zero-shot
Anomagic: Crossmodal Prompt-driven Zero-shot Anomaly Generation
Yuxin Jiang, Wei Luo, Hui Zhang, Qiyu Chen, Haiming Yao, Weiming Shen, Yunkang Cao
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 CPE and LoRA fine-tuning over the DRAEM-style baseline (ablation) | Anomagic · VisA (INP-Former++) | P-F1: 51.94 -> 54.00 | Improved |
| add CPE and LoRA over baseline (ablation) | Anomagic · VisA anomaly generation | IS: 1.85 -> 2.16 | Improved |
| add second module (CPE plus LoRA vs one module) (ablation) | Anomagic · VisA (INP-Former++) | P-F1: 53.87 -> 54.00 | Improved |
| add second module (CPE plus LoRA vs one module) (ablation) | Anomagic · VisA anomaly generation | IS: 2.16 -> 2.16 | No clear effect |
| use crossmodal prompts instead of unimodal prompts (ablation) | Anomagic · VisA | qualitative realism | Improved |
| use Anomagic anomalies instead of AnoGen (few-shot) for INP-Former++ training | INP-Former++ · VisA | P-F1: 52.61 -> 54.00 | Improved |
| use Anomagic instead of AnomalyAny for generation | anomaly generation · VisA | IS: 1.94 -> 2.16 | Improved |
| use Anomagic inpainting instead of AnomalyAny attention-based generation | generation runtime · VisA | time per sample: approximately 3 minutes -> 1.2 seconds | Improved |
| use Anomagic instead of AnoGen on detection I-ROC | INP-Former++ · VisA | I-ROC: 99.09 -> 99.08 | No clear effect |
Closest papers by meaning
- AnoStyler: Text-Driven Localized Anomaly Generation via Lightweight Style Transfer (AAAI 2026 · similarity 0.91)
- VarProtoAD: Variational Prototype-Conditioned Prompting for Zero-Shot Anomaly Detection (ECCV 2026 · similarity 0.90)
- Towards Zero-Shot Anomaly Detection and Reasoning with Multimodal Large Language Models (CVPR 2025 · similarity 0.89)
- Bidirectional Multimodal Prompt Learning with Scale-Aware Training for Few-Shot Multi-Class Anomaly Detection (CVPR 2026 · similarity 0.89)
- Unseen Visual Anomaly Generation (CVPR 2025 · similarity 0.88)
- One-to-Normal: Anomaly Personalization for Few-shot Anomaly Detection (NeurIPS 2024 · similarity 0.87)
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.
