Single-task Temporal Face Synthesis Model Based on Migration Confrontation

Quality of face images generated by existing methods is not high and there is a lack of research on Asian face datasets. For synthesizing face images in a specific age domain, single-task temporal face synthesis model based on migration confrontation is proposed here. Transfer learning is used to redesign the network structure of the generated confrontation network and network structure of the discriminant network. And we also improve the loss function, so that model can better obtain feature information in a small amount of data set in a single-task scenario. Through experimental analysis, the model proposed in this paper performs better under objective evaluation indicators than existing models, and the model is more scalable and diverse.