AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

AnomalyVFM -- Transforming Vision Foundation Models into Zero-Shot Anomaly Detectors

Matic Fučka, Vitjan Zavrtanik, Danijel Skočaj

CVPR 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Adapting VLMs for zero-shot classification · 6 citations (OpenAlex)

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 changedWhereMeasuredEffect
add synthetic dataset and feature adapters together to DINOv2DINOv2 · 9 industrial datasets avgimage-level AUROC: 83.0 -> 90.2Improved
add synthetic dataset and feature adapters to DINOv3DINOv3 · 9 industrial datasets avgpixel-level AUROC: 88.0 -> 95.3Improved
add synthetic dataset and feature adapters to RADIORADIO · 9 industrial datasets avgimage-level AUROC: 89.1 -> 94.1Improved
use AnomalyVFM (RADIO) over best prior zero-shot methodAnomalyVFM · 9 industrial datasets avgimage-level AUROC: 90.8 -> 94.1Improved
use AnomalyVFM over AdaCLIP on pixel-level F1-maxAnomalyVFM · 9 industrial datasets avgpixel-level F1-max: 51.0 -> 44.3Hurt
AnomalyVFM on medical image-level detection vs prior bestAnomalyVFM · medical datasets avgimage-level AUROC: 94.9 -> 94.0No clear effect
AnomalyVFM on medical pixel-level localisation vs AdaCLIPAnomalyVFM · medical datasets avgpixel-level AUROC: 88.8 -> 90.0Improved
add feature-based dataset filtering (ablation)AnomalyVFM · 9 industrial datasets avgpixel-level AUROC: -14.6 without filteringImproved
add foreground-guided anomaly location selection (ablation)AnomalyVFM · 9 industrial datasets avgpixel-level AUROC: -5.8 without foreground selectionImproved
add confidence-weighted loss (ablation)AnomalyVFM · 9 industrial datasets avgpixel-level AUROC: -2.0 without itImproved
replace FLUX with QWEN-Image or WAN generator (ablation)AnomalyVFM · 9 industrial datasets avgimage-level AUROC: -0.1 (QWEN), -0.4 (WAN)No clear effect
replace LoRA with VPT adapters (ablation)AnomalyVFM · 9 industrial datasets avgpixel-level F1-max: image AUROC -1.0, pixel F1 +4.0Mixed

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy