信息与控制2024，Vol.53Issue(5)：662-672,11.DOI:10.13976/j.cnki.xk.2024.3148
基于注意力机制和空洞金字塔池化的缺陷检测
Defect Detection Based on Attention Mechanism and Atrous Pyramid Pooling
摘要
Abstract
To solve the problem of low feature reconstruction accuracy in industrial product surface defect detection,which leads to high false positive rates at the image,pixel,and region levels,we pro-pose an unsupervised defect detection method with improved deep feature reconstruction(DFR).First,we introduce jump connections into the feature reconstruction process to improve the feature reconstruction accuracy and the ability of the model to reconstruct positive sample features.Sec-ond,we introduce an attention mechanism to improve the attention of the algorithm to defect re-gions and explore the effect of spatial attention on defect detection for different targets.Third,we introduce atrous pyramid pooling into the feature reconstruction module to capture context at multi-ple scales without increasing the number of parameters to improve the ability of the model to detect defects at different scales.Finally,we use the L2-SSIM loss function to constrain feature recon-struction to preserve the feature structure while maintaining pixel similarity.The proposed algo-rithm achieves 97.5％,97.2％,and 93.1％detection accuracy at the image,pixel,and region levels,respectively,outperforming the comparison algorithm.关键词
图像处理/缺陷检测/无监督学习/注意力机制/空洞金字塔池化Key words
image processing/defect detection/unsupervised learning/attention mechanism/atrous pyramid pooling分类
信息技术与安全科学引用本文复制引用
魏金洋,苑明哲,曹飞道,白海军,王文洪..基于注意力机制和空洞金字塔池化的缺陷检测[J].信息与控制,2024,53(5):662-672,11.基金项目
中国科学院科技服务网络计划(STS)—东莞专项(20211600200072) （STS）
