foreword
The work of style transfer is endless, from the transfer of pair pairs at the beginning, to the later unpair, and the style transfer of videos.
Among them, the cartoonization of characters is also a very popular technology.
Model
trends
Depth-Aware Generative Adversarial Network for Talking Head Video Generation
代码: https://github.com/harlanhong/CVPR2022-DaGAN
CCPL
Paper CCPL: Contrastive Coherence Preserving Loss for Versatile Style Transfer
Code: https://github.com/JarrentWu1031/CCPL
Blog post: ECCV 2022 Oral | CCPL: A Universal Relevance Preserving Loss Function for Universal Style Transfer