AI Paper Atlas

Research map / Label-efficient and robust learning / OOD and anomaly detection

Formally Exploring Visual Anomaly Detection Evaluation Metrics

Nasar Iqbal, Dennis Wagner, Philipp Liznerski, Nabeel Hussain Syed, Sophie Fellenz, Niki Martinel, Marius Kloft

ICML 2026 · Label-efficient and robust learning › OOD and anomaly detection · approach: Unified multi-class anomaly detection

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 a redundant second true-positive component to the predictionSAAM-ALARM · MVTec-AD case study, redundancySAAM-ALARM score: 0.5002 -> 0.2500Hurt
move the false positive farther from the ground truthSAAM-ALARM · MVTec-AD case study, false-alarm distanceSAAM-ALARM score: 0.4526 -> 0.4492Hurt
split false positives into two disconnected componentsSAAM-ALARM · MVTec-AD case study, false-alarm countSAAM-ALARM score: 0.4517 -> 0.4017Hurt
add redundancy via a second true-positive componentpixel-level precision/recall/F1/IoU · MVTec-AD case study, redundancypixel-level metricsNo clear effect
choose SAAM-ALARM instead of recall or F1 to rank detectorsmodel ranking on MVTec AD · MVTec AD, 6 detectorsranking consistency: 0.96 -> -0.24Mixed

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