PatchCore
A memory-bank-based industrial anomaly detection method that stores features of normal samples and uses nearest-neighbor retrieval for anomaly scoring. It is highly discriminative but incurs substantial storage overhead.
Timeline (last 90 days)
Memory bank methods (such as PatchCore and PaDiM) store a set of normal sample features during training and determine anomalies at inference via nearest-neighbor retrieval, with storage cost growing linearly with the training set size.
PaDiM and PatchCore are anomaly detection algorithms trained solely on normal samples and do not require defect samples to function.
PatchCore outputs anomaly heatmaps, and post-processing rules such as minimum area and connected-component constraints can be applied to filter sporadic noise, retaining only true defect signals with spatial clustering.
All Facts (3)
Memory bank methods (such as PatchCore and PaDiM) store a set of normal sample features during training and determine anomalies at inference via nearest-neighbor retrieval, with storage cost growing linearly with the training set size.
50%UnverifiedPaDiM and PatchCore are anomaly detection algorithms trained solely on normal samples and do not require defect samples to function.
50%UnverifiedPatchCore outputs anomaly heatmaps, and post-processing rules such as minimum area and connected-component constraints can be applied to filter sporadic noise, retaining only true defect signals with spatial clustering.
50%