TY - JOUR
T1 - U-Flow
T2 - A U-Shaped Normalizing Flow for Anomaly Detection with Unsupervised Threshold
AU - Tailanian, Matías
AU - Pardo, Álvaro
AU - Musé, Pablo
N1 - Publisher Copyright:
© The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024.
PY - 2024
Y1 - 2024
N2 - In this work, we propose a one-class self-supervised method for anomaly segmentation in images that benefits from both a modern machine learning approach and a more classic statistical detection theory. The method consists of four phases. First, features are extracted using a multi-scale image transformer architecture. Then, these features are fed into a U-shaped normalizing flow (NF) that lays the theoretical foundations for the subsequent phases. The third phase computes a pixel-level anomaly map from the NF embedding, and the last phase performs a segmentation based on the a contrario framework. This multiple hypothesis testing strategy permits the derivation of robust unsupervised detection thresholds, which are crucial in real-world applications where an operational point is needed. The segmentation results are evaluated using the mean intersection over union metric, and for assessing the generated anomaly maps we report the area under the receiver operating characteristic curve (AUROC), as well as the area under the per-region-overlap curve (AUPRO). Extensive experimentation in various datasets shows that the proposed approach produces state-of-the-art results for all metrics and all datasets, ranking first in most MVTec-AD categories, with a mean pixel-level AUROC of 98.74%. Code and trained models are available at https://github.com/mtailanian/uflow.
AB - In this work, we propose a one-class self-supervised method for anomaly segmentation in images that benefits from both a modern machine learning approach and a more classic statistical detection theory. The method consists of four phases. First, features are extracted using a multi-scale image transformer architecture. Then, these features are fed into a U-shaped normalizing flow (NF) that lays the theoretical foundations for the subsequent phases. The third phase computes a pixel-level anomaly map from the NF embedding, and the last phase performs a segmentation based on the a contrario framework. This multiple hypothesis testing strategy permits the derivation of robust unsupervised detection thresholds, which are crucial in real-world applications where an operational point is needed. The segmentation results are evaluated using the mean intersection over union metric, and for assessing the generated anomaly maps we report the area under the receiver operating characteristic curve (AUROC), as well as the area under the per-region-overlap curve (AUPRO). Extensive experimentation in various datasets shows that the proposed approach produces state-of-the-art results for all metrics and all datasets, ranking first in most MVTec-AD categories, with a mean pixel-level AUROC of 98.74%. Code and trained models are available at https://github.com/mtailanian/uflow.
KW - A contrario
KW - Anomaly detection
KW - Anomaly localization
KW - NFA
KW - Normalizing flows
KW - Novelty detection
KW - Number of false alarms
KW - One-class classification
KW - Self-supervised learning
UR - http://www.scopus.com/inward/record.url?scp=85194820650&partnerID=8YFLogxK
U2 - 10.1007/s10851-024-01193-y
DO - 10.1007/s10851-024-01193-y
M3 - Artículo
AN - SCOPUS:85194820650
SN - 0924-9907
JO - Journal of Mathematical Imaging and Vision
JF - Journal of Mathematical Imaging and Vision
ER -