Research map / Vision-language and multimodal / CLIP and zero-shot
DLVP-CLIP: Enhancing Fine-Grained Zero-Shot Anomaly Detection via Dynamic Local Visual Prompting
Gaowei Zhang, Lihe 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 SLFS and MLoP (DLVP) together (ablation) | DLVP-CLIP · MVTec-AD | I-AUROC: 89.5 -> 94.2 | Improved |
| add SLFS only (no MLoP) (ablation) | DLVP-CLIP · VisA | I-AUROC: 84.2 -> 86.3 | Improved |
| add MLoP only (no SLFS) (ablation) | DLVP-CLIP · MVTec-AD | I-AUROC: 89.5 -> 90.4 | Improved |
| add HFD wavelet decomposition (ablation) | DLVP-CLIP · MVTec-AD | I-AUROC: 92.8 -> 94.2 | Improved |
| add HFD wavelet decomposition (ablation) | DLVP-CLIP · VisA | I-AUROC: 85.8 -> 87.9 | Improved |
| use SLFS local patch prompts instead of a global token (ablation) | DLVP-CLIP · VisA | I-AUROC: 85.6 -> 87.9 | Improved |
| use wavelet HFD instead of Sobel filter (ablation) | DLVP-CLIP · VisA | I-AUROC: 86.7 -> 87.9 | Improved |
| increase number of local prompts Top-K from 1 to 4 (ablation) | SLFS · VisA | I-AUROC: 84.5 -> 87.9 | Improved |
| increase Top-K from 4 to 5 (ablation) | SLFS · VisA | I-AUROC: 87.9 -> 85.7 | Hurt |
| DLVP-CLIP versus Bayes-PFL on MVTec-AD pixel-level segmentation | DLVP-CLIP · MVTec-AD | P-AUROC: 90.4 vs 91.8 | Hurt |
Closest papers by meaning
- Aligning and Prompting Anything for Zero-Shot Generalized Anomaly Detection (AAAI 2025 · similarity 0.90)
- AA-CLIP: Enhancing Zero-Shot Anomaly Detection via Anomaly-Aware CLIP (CVPR 2025 · similarity 0.90)
- AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detection (ICLR 2024 · similarity 0.88)
- Bayesian Prompt Flow Learning for Zero-Shot Anomaly Detection (CVPR 2025 · similarity 0.88)
- Towards Zero-Shot Anomaly Detection and Reasoning with Multimodal Large Language Models (CVPR 2025 · similarity 0.88)
- PromptAD: Learning Prompts with only Normal Samples for Few-Shot Anomaly Detection (CVPR 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.
