MAEDAY: MAE for few- and zero-shot AnomalY-Detection

We propose using Masked Auto-Encoder (MAE), a transformer model self-supervisedly trained on image inpainting, for anomaly detection (AD). Assuming anomalous regions are harder to reconstruct compared with normal regions. MAEDAY is the first image-reconstruction-based anomaly detection method that u...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Computer vision and image understanding Jg. 241; S. 103958
Hauptverfasser: Schwartz, Eli, Arbelle, Assaf, Karlinsky, Leonid, Harary, Sivan, Scheidegger, Florian, Doveh, Sivan, Giryes, Raja
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Elsevier Inc 01.04.2024
Schlagworte:
Anomaly-detection
Foreign object detection
Masked autoencoder
Masked autoencoder
41A10
65D05
65D17
Anomaly-detection
Foreign object detection
41A05
ISSN:1077-3142, 1090-235X
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:We propose using Masked Auto-Encoder (MAE), a transformer model self-supervisedly trained on image inpainting, for anomaly detection (AD). Assuming anomalous regions are harder to reconstruct compared with normal regions. MAEDAY is the first image-reconstruction-based anomaly detection method that utilizes a pre-trained model, enabling its use for Few-Shot Anomaly Detection (FSAD). We also show the same method works surprisingly well for the novel tasks of Zero-Shot AD (ZSAD) and Zero-Shot Foreign Object Detection (ZSFOD), where no normal samples are available. •Pre-training MAE on an arbitrary set of images and using it for Anomaly-Detection.•Suggesting the new task of Zero-Shot Anomaly-Detection and demonstrating strong results.•Demonstrating strong results for the new task of Zero-Shot Foreign Object Detection.•Releasing a new Foreign Object Detection dataset.
ISSN:1077-3142
1090-235X
DOI:10.1016/j.cviu.2024.103958