NF.π Niccolò Ferrari

Industrial computer vision and anomaly detection

MH-PatchCore

Covariance-aware and streaming-compatible PatchCore research for industrial anomaly detection.

PreprintPublic code
arXiv Repository

Overview

MH-PatchCore is a research and implementation line around PatchCore-style industrial anomaly detection. The arXiv preprint covers covariance-aware scoring, streaming behavior, bounded memory, and deployment constraints that become important when the method has to run outside a controlled benchmark.