StructCore: Structure-Aware Image-Level Scoring for Training-Free Unsupervised Anomaly Detection
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910028724699136 |
|---|---|
| author | Chae, Joongwon Luo, Lihui Liu, Yang Wang, Runming Yu, Dongmei Liang, Zeming Yuan, Xi Zhang, Dayan Chen, Zhenglin Qin, Peiwu Chae, Ilmoon |
| author_facet | Chae, Joongwon Luo, Lihui Liu, Yang Wang, Runming Yu, Dongmei Liang, Zeming Yuan, Xi Zhang, Dayan Chen, Zhenglin Qin, Peiwu Chae, Ilmoon |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_17048 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | StructCore: Structure-Aware Image-Level Scoring for Training-Free Unsupervised Anomaly Detection Chae, Joongwon Luo, Lihui Liu, Yang Wang, Runming Yu, Dongmei Liang, Zeming Yuan, Xi Zhang, Dayan Chen, Zhenglin Qin, Peiwu Chae, Ilmoon Computer Vision and Pattern Recognition 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. |
| title | StructCore: Structure-Aware Image-Level Scoring for Training-Free Unsupervised Anomaly Detection |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2602.17048 |