Face Recognition Based on Global and Local Feature Fusion
You Zhou, Yiyue Liu, Guijin Han, Zichao Zhang · 2019
In the training process of face images, the traditional convolutional neural network does not fuse the information of high and low convolutional layers. In order to make full use of the feature information of each layer of image, a convolutional neural network model based on global and local feature fusion of MobileNet network is proposed. The algorithm fuses the local features of the first layer of the network with the global features after the principal component analysis. Thus, the expression of shallow features is increased, the extraction effect of deep features is strengthened, and the information extracted by the improved MobileNet network is more complete.