Mahalanobis PatchCore: Covariance-Aware and Streaming-Compatible Industrial Anomaly Detection 事件

Mahalanobis PatchCore: Covariance-Aware and Streaming-Compatible Industrial Anomaly Detection arXiv:2605.27748v1 Announce Type: new Abstract: Industrial visual anomaly detection is usually one-class: normal images are abundant, while defects are rare, heterogeneous, and often unavailable during system design. PatchCore-style retrieval suits this setting because it scores test images from a memory bank of normal patch features, but the standard Euclidean geometry ignores feature correlations and