In order to improve the accuracy of facial expression recognition, a facial expression recognition based on data enhancement strategy is proposed. The training data set used in the experiment is enhanced by adding different weight allocation strategies. The weights of each training are randomly generated to ensure the diversity of the training data, and which weight distribution strategy is suitable for the enhancement of facial expression recognition data by comparing the experimental results. At the same time, it solves the problem of facial expression recognition due to the lack of diversity and low accuracy of facial expression recognition, and improves the accuracy and robustness of facial expression recognition. In addition, it also uses VGG19 feature extraction network to achieve high-precision facial expression recognition by learning robustness and distinguishing features from the data. The experimental results show that the model trained by the enhanced data in this way can improve the recognition rate of seven kinds of facial expressions on Fer2013 and extended Cohn-Kanade (CK+) database compared with the original data set.