AMI-Net adaptively masks anomalous regions in pretrained feature maps, inpaints them with a transformer, and uses reconstruction error for real-time industrial anomaly detection and localization.
Visual anomaly detection via partition memory bank module and error estimation,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CV 1years
2024 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
AMI-Net: Adaptive Mask Inpainting Network for Industrial Anomaly Detection and Localization
AMI-Net adaptively masks anomalous regions in pretrained feature maps, inpaints them with a transformer, and uses reconstruction error for real-time industrial anomaly detection and localization.