期刊
PATTERN RECOGNITION
卷 112, 期 -, 页码 -出版社
ELSEVIER SCI LTD
DOI: 10.1016/j.patcog.2020.107706
关键词
Anomaly detection; Video anomaly detection; Inpainting; CNN
资金
- Slovenian Research Agency (ARRS) [J2-9433, J2-8175, P2-0214]
The article discusses the issue of visual anomaly detection and the shortcomings of the auto-encoder method in reconstructing anomalous regions. A new approach (RIAD) is proposed to address this problem by reconstructing images through inpainting, setting a new state-of-the-art in anomaly detection benchmarks.
Visual anomaly detection addresses the problem of classification or localization of regions in an image that deviate from their normal appearance. A popular approach trains an auto-encoder on anomaly-free images and performs anomaly detection by calculating the difference between the input and the reconstructed image. This approach assumes that the auto-encoder will be unable to accurately reconstruct anomalous regions. But in practice neural networks generalize well even to anomalies and reconstruct them sufficiently well, thus reducing the detection capabilities. Accurate reconstruction is far less likely if the anomaly pixels were not visible to the auto-encoder. We thus cast anomaly detection as a self supervised reconstruction-by-inpainting problem. Our approach (RIAD) randomly removes partial image regions and reconstructs the image from partial inpaintings, thus addressing the drawbacks of autoenocoding methods. RIAD is extensively evaluated on several benchmarks and sets a new state-of-the art on a recent highly challenging anomaly detection benchmark. (C) 2020 Elsevier Ltd. All rights reserved.Y
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据