期刊
IEEE ACCESS
卷 10, 期 -, 页码 78446-78454出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/ACCESS.2022.3193699
关键词
Feature extraction; Location awareness; Transfer learning; Anomaly detection; Heating systems; Measurement; Licenses; Anomaly localization; deep neural network; transfer learning
资金
- Institute of Information and Communications Technology Planning and Evaluation (IITP) [RS-2022-00155915]
- Arti-cial Intelligence Convergence Research Center, Inha University
- National Research Foundation of Korea (NRF) - Korean Government [Ministry of Science and ICT (MSIT)] [2022R1A2C2010095, 2022R1A4A1033549]
- National Research Foundation of Korea [2022R1A2C2010095] Funding Source: Korea Institute of Science & Technology Information (KISTI), National Science & Technology Information Service (NTIS)
Previous studies on anomaly localization lacked adaptation to the target dataset. This paper proposes a Coupled-hypersphere-based Feature Adaptation method that accurately discriminates between normal and abnormal features, outperforming previous methods in both quantitative and qualitative evaluations. The paper also highlights the issue of biased features in pre-trained CNNs.
For a long time, anomaly localization has been widely used in industries. Previous studies focused on approximating the distribution of normal features without adaptation to a target dataset. However, since anomaly localization should precisely discriminate between normal and abnormal features, the absence of adaptation may make the normality of abnormal features overestimated. Thus, we propose Coupled-hypersphere-based Feature Adaptation (CFA) which accomplishes sophisticated anomaly localization using features adapted to the target dataset. CFA consists of (1) a learnable patch descriptor that learns and embeds target-oriented features and (2) a scalable memory bank independent of the size of the target dataset. And, CFA adopts transfer learning to increase the normal feature density so that abnormal features can be clearly distinguished by applying patch descriptor and memory bank to a pre-trained CNN. The proposed method outperforms the previous methods quantitatively and qualitatively. For example, it provides an AUROC score of 99.5% in anomaly detection and 98.5% in anomaly localization of MVTec AD benchmark. In addition, this paper points out the negative effects of biased features of pre-trained CNNs and emphasizes the importance of adaptation to the target dataset.
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