基于小样本度量迁移学习的表面缺陷检测

The metric learning method in few-shot learning is introduced into the field of defect detection, and a few-shot metric transfer learning method is proposed to solve the problem of requiring a large number of learning samples in deep learning methods. The method is mainly divided into two stages: in the first stage, a deep network is pre-trained using public or easily accessible large datasets; in the second stage, the relevant knowledge learned by the network is transferred to the field of surface defect detection through a metric learning module. Experiments show the feasibility of few-shot learning in the field of defect detection.

将小样本学习中的度量学习方法引入缺陷检测领域,提出小样本度量迁移学习方法,用于解决深度学习方法中需要大量学习样本的问题.方法主要分为两个阶段:第一阶段使用公开或便于获得的大型数据集预训练深度网络;第二阶段将网络学习到的相关知识通过度量学习模块迁移到表面缺陷检测领域.实验表明,小样本学习在缺陷检测领域的可行性.