A method and a device for optimizing a human face picture quality evaluation model. The method comprises: establishing a human face picture test set; recognizing a similarity between a human face picture to be tested and a human face sample picture in a preset human face database, and obtaining a recognition result of each human face picture to be tested according to the similarity and picture identity information; determining a quality score of each human face picture to be tested according to the recognition resu and taking the human face pictures to be tested and the corresponding quality scores thereof as training data to perform neural network training to obtain an optimized human face picture quality evaluation model and parameters. By use of the optimized human face picture quality evaluation model and the parameters, the evaluation of the human face picture quality is unaffected by artificial subjective factors.
展开▼
