arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.16229cs.CV
收藏

Head2HeadFS:基于视频的头部再现与少样本学习

Head2HeadFS: Video-based Head Reenactment with Few-shot Learning

Imperial College London(帝国理工学院);University of Exeter(埃克塞特大学)

Michail Christos Doukas, Mohammad Rami Koujan, Viktoriia Sharmanska, Stefanos Zafeiriou

摘要:

在过去的几年里,大量的工作已经完成了面部重现的问题,解决方案主要来自图形社区。头部再现是一项更具挑战性的任务,其目的不仅是将面部表情,而且是将整个头部姿势从源人物转移到目标人物。目前的方法要么是训练特定于个人的系统,要么是使用面部标志来模拟人类头部,这种表示可能会将不需要的身份属性从源转移到目标。我们提出了head2headFS,一种新的易于适应的管道头重演。我们的条件合成的目标人物密集的3D人脸形状信息的来源,这使得高质量的表达和姿态转移。我们的基于视频的渲染网络在少样本学习策略下进行了微调,仅使用少量样本。这允许在多人数据集上训练的通用生成器快速适应特定于个人的生成器。

英文摘要:

Over the past years, a substantial amount of work has been done on the problem of facial reenactment, with the solutions coming mainly from the graphics community. Head reenactment is an even more challenging task, which aims at transferring not only the facial expression, but also the entire head pose from a source person to a target. Current approaches either train person-specific systems, or use facial landmarks to model human heads, a representation that might transfer unwanted identity attributes from the source to the target. We propose head2headFS, a novel easily adaptable pipeline for head reenactment. We condition synthesis of the target person on dense 3D face shape information from the source, which enables high quality expression and pose transfer. Our video-based rendering network is fine-tuned under a few-shot learning strategy, using only a few samples. This allows for fast adaptation of a generic generator trained on a multiple-person dataset, into a person-specific one.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑