基于文本条件的语境化头像生成用于零样本定制化【英文版】

英文标题：Text-Conditional Contextualized Avatars For Zero-Shot Personalization中文摘要：本文提出了一个零射击的管道，该管道可以通过捕捉用户的身份以令人愉悦的方式进行头像生成，具有个性化的图像生成特征，此外，作者使用大规模图像数据集学习人类 3D 姿态参数，克服了动作捕捉数据集的局限性。英文摘要：Recent large-scale text-to-image generation models have made significantimprovements in the quality, realism, and diversity of the synthesized imagesand enable users to control the created content through language. However, thepersonalization aspect of these generative models is still challenging andunder-explored.