Prioritized Local Matching Network for Cross-Category Few-Shot Anomaly Detection

被引:0
|
作者
Deng H. [1 ]
Luo H. [1 ]
Zhai W. [1 ]
Cao Y. [1 ,2 ]
Kang Y. [1 ,2 ]
机构
[1] University of Science and Technology of China, Anhui
[2] School of Information Science and Technology, University of Science and Technology of China, Anhui
来源
关键词
Adaptation models; Anomaly detection; Anomaly Detection; Correlation; Cross-Category; Few-shot Learning; Semantics; Task analysis; Training; Visual Correspondence; Visualization;
D O I
10.1109/TAI.2024.3385743
中图分类号
学科分类号
摘要
In response to the rapid evolution of products in industrial inspection, this paper introduces the Cross-category Few-shot Anomaly Detection (C-FSAD) task, aimed at efficiently detecting anomalies in new object categories with minimal normal samples. However, the diversity of defects and significant visual distinctions among various objects hinder the identification of anomalous regions. To tackle this, we adopt a pairwise comparison between query and normal samples, establishing an intimate correlation through fine-grained correspondence. Specifically, we propose the Prioritized Local Matching Network (PLMNet), emphasizing local analysis of correlation, which includes three primary components: 1) Local Perception Network refines the initial matches through bidirectional local analysis; 2) Step Aggregation strategy employs multiple stages of local convolutional pooling to aggregate local insights; 3) Defect-sensitive Weight Learner adaptively enhances channels informative for defect structures, ensuring more discriminative representations of encoded context. Our PLMNet deepens the interpretation of correlations, from geometric cues to semantics, efficiently extracting discrepancies in feature space. Extensive experiments on two standard industrial anomaly detection benchmarks demonstrate our state-of-the-art performance in both detection and localization, with margins of 9.8% and 5.4% respectively. IEEE
引用
收藏
页码:1 / 12
页数:11
相关论文
共 50 条
  • [21] Network Pruning and Fine-tuning for Few-shot Industrial Image Anomaly Detection
    Zhang, Jie
    Suganuma, Masanori
    Okatani, Takayuki
    2023 IEEE 21ST INTERNATIONAL CONFERENCE ON INDUSTRIAL INFORMATICS, INDIN, 2023,
  • [22] Learning Hypersphere for Few-shot Anomaly Detection on Attributed Networks
    Guo, Qiuyu
    Zhao, Xiang
    Fang, Yang
    Yang, Shiyu
    Lin, Xuemin
    Ouyang, Dian
    PROCEEDINGS OF THE 31ST ACM INTERNATIONAL CONFERENCE ON INFORMATION AND KNOWLEDGE MANAGEMENT, CIKM 2022, 2022, : 635 - 645
  • [23] Insulator Anomaly Detection Method Based on Few-Shot Learning
    Wang, Zhaoyang
    Gao, Qiang
    Li, Dong
    Liu, Junjie
    Wang, Hongwei
    Yu, Xiao
    Wang, Yipin
    IEEE ACCESS, 2021, 9 : 94970 - 94980
  • [24] Few-shot weakly-supervised cybersecurity anomaly detection
    Kale, Rahul
    Thing, Vrizlynn L. L.
    COMPUTERS & SECURITY, 2023, 130
  • [25] Local spatial alignment network for few-shot learning
    Yu, Yunlong
    Zhang, Dingyi
    Wang, Sidi
    Ji, Zhong
    Zhang, Zhongfei
    NEUROCOMPUTING, 2022, 497 : 182 - 190
  • [26] Bidirectional matching and aggregation network for few-shot relation extraction
    Wei, Zhongcheng
    Guo, Wenjie
    Zhang, Yunping
    Zhang, Jieying
    Zhao, Jijun
    PEERJ COMPUTER SCIENCE, 2023, 9
  • [27] Bidirectional Matching Prototypical Network for Few-Shot Image Classification
    Fu, Wen
    Zhou, Li
    Chen, Jie
    IEEE SIGNAL PROCESSING LETTERS, 2022, 29 : 982 - 986
  • [28] Bidirectional matching and aggregation network for few-shot relation extraction
    Wei Z.
    Guo W.
    Zhang Y.
    Zhang J.
    Zhao J.
    PeerJ Computer Science, 2023, 9
  • [29] HMNet: Hybrid Matching Network for Few-Shot Link Prediction
    Xiao, Shan
    Duan, Lei
    Xie, Guicai
    Li, Renhao
    Chen, Zihao
    Deng, Geng
    Nummenmaa, Jyrki
    DATABASE SYSTEMS FOR ADVANCED APPLICATIONS (DASFAA 2021), PT I, 2021, 12681 : 307 - 322
  • [30] Category-Aware Siamese Learning Network for Few-Shot Segmentation
    Sun, Hui
    Zhang, Ziyan
    Huang, Lili
    Jiang, Bo
    Luo, Bin
    COGNITIVE COMPUTATION, 2024, 16 (03) : 924 - 935