The Studyof Wafer Defect Classification Method Based on Improved Dense Neural Network
A deep learning model based on convolutional neural networks and dense neural networks(DenseNet)is pro-posed to address the issues of low parallelism and inability to learn global information well in existing wafer defect classification mod-els,and wafer defects are classified.On the basis of convolutional neural networks,a convolutional attention module is introduced to consider the characteristics of channel and spatial dimensions,improve the convergence effect of the model,and construct an im-proved dense neural network to achieve the classification of wafer defects.We achieved an average accuracy of 98.9%and an F1 value of 92.7%on the MIR-WM811K dataset,with an average accuracy improvement of approximately 2%compared to DenseNet.
万方数据
维普
