Attention-relation network for mobile phone screen defect classification via a few samples

作     者：Jiao Mao Guoliang Xu Lijun He Jiangtao Luo Jiao Mao;Guoliang Xu;Lijun He;Jiangtao Luo

作者机构：College of Communication and Information EngineeringChongqing University of Posts and TelecommunicationsChongqing400065China Institute of Electronic Information and Networking EngineeringChongqing University of Posts and TelecommunicationsChongqing400065China 

出 版 物：《Digital Communications and Networks》 (数字通信与网络（英文版）)

年 卷 期：2024年第10卷第4期

页      面：1113-1120页

核心收录：

学科分类：0809[工学-电子科学与技术（可授工学、理学学位）] 08[工学] 

主　　题：Mobile phone screen defects A few samples Relation network Attention mechanism Dilated convolution 

摘      要：How to use a few defect samples to complete the defect classification is a key challenge in the production of mobile phone *** attention-relation network for the mobile phone screen defect classification is proposed in this *** architecture of the attention-relation network contains two modules:a feature extract module and a feature metric *** from other few-shot models,an attention mechanism is applied to metric learning in our model to measure the distance between features,so as to pay attention to the correlation between features and suppress unwanted ***,we combine dilated convolution and skip connection to extract more feature information for follow-up *** validate attention-relation network on the mobile phone screen defect *** experimental results show that the classification accuracy of the attentionrelation network is 0.9486 under the 5-way 1-shot training strategy and 0.9039 under the 5-way 5-shot *** achieves the excellent effect of classification for mobile phone screen defects and outperforms with dominant advantages.