摘要
针对瓷砖表面异常检测中人工检测效率低、成本高和自动检测标记样本不足、漏检率高等问题,提出了一种自监督学习模型,无需大量缺陷样本,即可实现瓷砖表面常见异常的检测与定位。自监督学习通过样本扩充产生负样本,利用分布增强对比学习提高数据不规则性和扩展样本分布,进而降低对比表示的均匀性,使表示特征分布与分类目标保持一致。在自监督学习表示基础上,构建一类分类器实现了准确的异常检测与定位。实验结果表明,在异常检测标准评估度量(AUROC)准则下,该方法相比其他两种先进方法异常检测率分别提高了3.71%和2.74%;异常定位率分别提高了1.22%和4.01%,且具有更可靠的检测性能。
Aiming at the problems of low efficiency,high cost,insufficient automatic detection label samples and high missed detection rate in the surface defect detection of ceramic tiles,a self-supervised learning model is proposed,without a large number of defect samples,the detection and location of common defects on the surface of ceramic tiles can be realized.Self-supervised learning generates negative samples through sample expansion,and uses distribution-augmented contrastive learning to improve data irregularity and expand sample distribution,thereby reducing the consistency of comparative representation and making the representation feature distribution consistent with the classification target.Based on self-supervised learning representation,a class of classifiers is constructed to achieve accurate anomaly detection and localization.The experimental results show that compared with the other two advanced methods,under the standard evaluation criterion(AUROC)of anomaly detection,the anomaly detection rate is increased by 3.71%and 2.74%respectively;the abnormal location rate increased by 1.22%and 4.01%respectively,with more reliable detection performance.
作者
王飞州
程凡永
张明艳
张红
Wang Feizhou;Cheng Fanyong;Zhang Mingyan;Zhang Hong(School of Electrical Engineering,Anhui Polytechnic University,Wuhu 241000,China;Anhui Polytechnic University,Wuhu 241000,China)
出处
《电子测量技术》
北大核心
2023年第17期180-188,共9页
Electronic Measurement Technology
基金
国家自然科学基金(61976005)
安徽工程大学-鸠江区协同创新专项基金(2022cyxtb10)
芜湖市重点研发项目(2022yf42)
检测技术与节能装置安徽省重点实验室开放研究基金(JCKJ2021B06)
安徽省教育厅重点项目(2022AH050981)资助。
关键词
自监督
瓷砖
分布增强对比学习
异常检测
self-supervised
ceramic tile
distributed-augmented contrast learning
anomaly detection
