基于卷积收缩自编码器的半监督缺陷识别方法

Semi-supervised defect recognition method based on contractive convolutional autoencoder

提出了一种基于半监督卷积收缩自编码器的缺陷识别方法.从未标记数据中获取有效缺陷信息,结合少量标记样本,实现较高的缺陷识别效果,解决了传统的基于卷积神经网络的缺陷识别方法依赖大量标记样本的问题.实验结果表明:本文方法具有较高的识别精度,在少量标记样本下即可获得不错的识别效果,相较于其他方法,准确率提升4.93%~62.96%,可有效降低样本标记成本,加快模型部署速度,确保质量检测和生产计划顺利进行.

A defect recognition method based on semi-supervised convolutional contractive autoencoder is proposed.The effective defect information obtained from unlabeled data was combined with a small number of labeled samples to achieve a higher defect recognition effect,which effectively solves the problem that the traditional defect recognition method based on convolutional neural network relies on a large number of labeled samples.The experimental results show that the proposed method has a high recognition accuracy and can achieve a good recognition effect under a small number of labeled samples.Compared with other methods,the accuracy rate is increased by 4.93%~62.96%,which can effectively reduce the cost of sample labeling,speed up the model deployment,and ensure the smooth progress of quality testing and production planning.