Abstract
While the task of face swapping has recently gained attention in the researchcommunity, a related problem of head swapping remains largely unexplored. Inaddition to skin color transfer, head swap poses extra challenges, such as theneed to preserve structural information of the whole head during synthesis andinpaint gaps between swapped head and background. In this paper, we addressthese concerns with GHOST 2.0, which consists of two problem-specific modules.First, we introduce enhanced Aligner model for head reenactment, whichpreserves identity information at multiple scales and is robust to extreme posevariations. Secondly, we use a Blender module that seamlessly integrates thereenacted head into the target background by transferring skin color andinpainting mismatched regions. Both modules outperform the baselines on thecorresponding tasks, allowing to achieve state of the art results in headswapping. We also tackle complex cases, such as large difference in hair stylesof source and target. Code is available athttps://github.com/ai-forever/ghost-2.0