Research map / Vision-language and multimodal / CLIP and zero-shot
FE-CLIP: Frequency Enhanced CLIP Model for Zero-Shot Anomaly Detection and Segmentation
Tao Gong, Qi Chu, Bin Liu, Wei Zhou, Nenghai Yu
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 FFE adapter instead of a plain conv adapter (ablation) | CLIP · MVTec AD image-level | AUROC: 87.8 -> 90.0 | Improved |
| add LFS adapter instead of a plain conv adapter (ablation) | CLIP · MVTec AD image-level | AUROC: 87.8 -> 90.6 | Improved |
| add LFS adapter instead of a plain conv adapter (ablation) | CLIP · VisA image-level | AUROC: 82.3 -> 83.9 | Improved |
| add FFE adapter instead of a plain conv adapter (ablation) | CLIP · VisA pixel-level | AUROC: 94.8 -> 95.1 | Improved |
| combine FFE and LFS adapters (add LFS to FFE) (ablation) | FE-CLIP · MVTec AD image-level | AUROC: 90.0 -> 91.9 | Improved |
| combine FFE and LFS adapters (add FFE to LFS) (ablation) | FE-CLIP · MVTec AD image-level | AUROC: 90.6 -> 91.9 | Improved |
| combine FFE and LFS adapters (ablation) | FE-CLIP · VisA pixel-level | AUROC: 95.5 -> 95.9 | Improved |
| combine FFE and LFS adapters (ablation) | FE-CLIP · MVTec AD pixel-level | AUROC: 91.5 -> 92.6 | Improved |
| use DCT instead of FFT for frequency transform (ablation) | FE-CLIP · MVTec AD image-level | AUROC: 92.0 -> 91.9 | No clear effect |
| use DCT instead of FFT for frequency transform (ablation) | FE-CLIP · VisA image-level | AUROC: 84.1 -> 84.6 | No clear effect |
| use DCT instead of FFT for frequency transform (ablation) | FE-CLIP · MVTec AD pixel-level | AUROC: 92.8 -> 92.6 | No clear effect |
| add more normal shots (1 to 4) | FE-CLIP · MVTec AD image-level few-shot | AUROC: 95.6 -> 96.6 | Improved |
Later papers that compared with this method
1 later accepted paper reports FE-CLIP as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- From Attraction to Equilibrium: Physics-Inspired Semantic Gravitons for Zero-Shot Anomaly Detection (CVPR 2026 · better on 5 of 6 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detection (ICLR 2024 · similarity 0.90)
- AA-CLIP: Enhancing Zero-Shot Anomaly Detection via Anomaly-Aware CLIP (CVPR 2025 · similarity 0.89)
- Aligning and Prompting Anything for Zero-Shot Generalized Anomaly Detection (AAAI 2025 · similarity 0.87)
- DZAD: Diffusion-based Zero-shot Anomaly Detection (AAAI 2025 · similarity 0.87)
- MuSc: Zero-Shot Industrial Anomaly Classification and Segmentation with Mutual Scoring of the Unlabeled Images (ICLR 2024 · similarity 0.85)
- Towards Zero-Shot Anomaly Detection and Reasoning with Multimodal Large Language Models (CVPR 2025 · 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.
