StructCore: Structure-Aware Image-Level Scoring for Training-Free Unsupervised Anomaly Detection

TL;DR

StructCore introduces a training-free, structure-aware scoring method for unsupervised anomaly detection that captures distributional and spatial information, outperforming max pooling in image-level AUROC scores.

Contribution

It proposes a novel, training-free approach that leverages structural descriptors and Mahalanobis calibration for improved anomaly detection.

Findings

Achieves 99.6% AUROC on MVTec AD

Achieves 98.4% AUROC on VisA

Outperforms max pooling by capturing structural information

Abstract

Max pooling is the de facto standard for converting anomaly score maps into image-level decisions in memory-bank-based unsupervised anomaly detection (UAD). However, because it relies on a single extreme response, it discards most information about how anomaly evidence is distributed and structured across the image, often causing normal and anomalous scores to overlap. We propose StructCore, a training-free, structure-aware image-level scoring method that goes beyond max pooling. Given an anomaly score map, StructCore computes a low-dimensional structural descriptor phi(S) that captures distributional and spatial characteristics, and refines image-level scoring via a diagonal Mahalanobis calibration estimated from train-good samples, without modifying pixel-level localization. StructCore achieves image-level AUROC scores of 99.6% on MVTec AD and 98.4% on VisA, demonstrating robust image-level anomaly detection by exploiting structural signatures missed by max pooling.