Abstract
Anomaly defect detection has become an indispensable part of industrial production process. In previous study, a large part of the traditional anomaly detection algorithms belong to the category of supervised learning, while the unsupervised situation is more common for most practical application scenarios. Hence gradually unsupervised anomaly detection has been the subject of much research over the last few years. In this survey, we provide a comprehensive introduction to newly proposed approaches for visual anomaly detection. We hope that it can help the research community as well as the industry field to build a broader and cross-domain perspective.
Abstract (translated)
URL
https://arxiv.org/abs/2204.11161