BEAUTY PREDICTION METHOD AND DEVICE BASED ON MIGRATION AND WEAK SUPERVISION, AND STORAGE MEDIUM

摘要

A facial beauty prediction method and device based on migration and weak supervision, and a storage medium. The method comprises: preprocessing an input face image; training a source domain network by using the preprocessed image, and migrating parameters of the source domain network to a target domain network; inputting a noise image marked with a noise tag and a true-value image marked with a true-value tag into the target domain network to obtain image features; and inputting the image features into a classification network to obtain a final facial beauty prediction result. The problems of model data volume requirements being too large, model overfitting, weak model generalization ability and long model training times are effectively solved, and model stability and robustness are improved; and the problem of database tag unreliability is solved.