摘要
总结了人脸跨域合成技术的起源、任务类型与难点、技术发展与挑战、潜在应用与问题等,从自监督与弱监督跨域合成、基于预训练大模型跨域合成、基于跨域合成隐私保护3个方面探讨了人脸跨域合成技术未来发展趋势与挑战。
Advances in deep learning technology and the development of the digital economy have promoted the development of artificial intelligence-generated content(AIGC)technologies such as virtual humans.Cross-domain face synthesis is one of the key technologies in virtual human production,and it has a wide range of applications in social media,film and television production and other fields.This paper summarizes the origin of cross-domain face synthesis technology,and its typical task types and difficulties,technological development and challenges,potential applications,and issues,and discusses its future development trend and challenges from the aspects of self-supervised and weakly supervised cross-domain synthesis,utilization of pre-trained large models,and privacy protection.
作者
刘琦
吴昊展
谢添鑫
韩琥
LIU Qi;WU Haozhan;XIE Tianxin;HAN Hu(Department of Network Security,Henan Police College,Zhengzhou 450000,China;University of Chinese Academy of Sciences,Beijing 100049,China;Institute of Computing Technology,Chinese Academy of Sciences,Beijing 100190,China)
出处
《科技导报》
CAS
CSCD
北大核心
2023年第16期113-123,共11页
Science & Technology Review
基金
河南省科技厅科技攻关项目(222102210089)。
关键词
人脸跨域合成
虚拟人
生成对抗网络
神经渲染
扩散模型
cross-domain face synthesis
virtual human
generative adversarial network
neural rendering
diffusion model
