DiffBody: Human Body Restoration by Imagining with Generative Diffusion Prior

arXiv:2404.03642v1 Announce Type: new
Abstract: Human body restoration plays a vital role in various applications related to the human body. Despite recent advances in general image restoration using generative models, their performance in human body restoration remains mediocre, often resulting in foreground and background blending, over-smoothing surface textures, missing accessories, and distorted limbs. Addressing these challenges, we propose a novel approach by constructing a human body-aware diffusion model that leverages domain-specific knowledge to enhance performance. Specifically, we employ a pretrained …

abstract advances applications arxiv cs.cv diffusion general generative generative models human image image restoration mediocre performance prior role surface type vital