A Deep Face Recognition Method Based on Model Fine-tuning and Principal Component Analysis

In this paper, we propose a simple and effective deep face recognition method based on model fine-tuning and principal component analysis. At first, we use our own face dataset to fine tune the improved VGG-Face model. This can effectively solve the problem that the training dataset is too small and the data distribution is different. Through the part of the existing model parameters as the initial parameters of the new model, greatly accelerated the convergence rate of the model training. Then, for the facial features extracted by the deep learning method, we use principal component analysis to further remove redundant features, reduce the complexity of the features, and improve the face recognition rate. The experimental results prove that the proposed approach achieves a good face recognition accuracy on our test dataset.