中文说明：基于CNN的卷积神经网络应用，近年来深度学习算法的应用越来越广泛， 针对深度学习，重点研究了其中较为主流的两类模型结构， CNN 与 DBNs ，并在MATLAB 中构建这两种深度学习模型结构并分别应用于表面缺陷识别中，利用 CNN 学习输入数据中的特征信息，进行分类识别；利用DBNs 进行重构得到的模板图片，进而检测表面缺陷。
English Description：
Convolutional neural network based on CNN has been widely used in recent years. Aiming at deep learning, this paper focuses on two kinds of popular model structures, CNN and DBNs, and constructs these two kinds of deep learning model structures in MATLAB and applies them to surface defect recognition The feature information in the input data is learned for classification and recognition, and the template image reconstructed by DBNs is used to detect surface defects.