Appearance and Pose-guided Human Generation: A Survey
Fangjian Liao, Xingxing Zou, Wai Keung Wong
Abstract
Appearance and pose-guided human generation is a burgeoning field that has captured significant attention. This subject’s primary objective is to transfer pose information from a target source to a reference image, enabling the generation of high-resolution images or videos that seamlessly link the virtual and real worlds, leading to novel trends and applications. This survey thoroughly illustrates the task of appearance and pose-guided human generation and comprehensively reviews mainstream methods. Specifically, it systematically discusses prior information, pose-based transformation modules, and generators, offering a comprehensive understanding and discussion of each mainstream pose transformation and generation process. Furthermore, the survey explores current applications and future challenges in the domain. Its ultimate goal is to serve as quick guidelines, providing practical assistance in human generation and its diverse applications.