3DFace@Home: A pilot study for robust and highly accurate facial 3D reconstruction from mobile devices for facial growth monitoring at home

3DFace@Home的目标是开发和测试一种系统，以便在真实世界的条件下，从终端用户的移动设备(如家中的智能手机)中稳健地获得高精度的面部3D重建。它是由一个移动应用程序组成的，用户可以通过引导和有效性检查自己捕捉自己的面部图像。这些图像被发送到计算服务器，在那里将使用嵌入多视点几何约束的基于深度学习的面部3D重建来获得面部的3D形状。该项目的目标是在影响面部生长的疾病的背景下，达到测量解剖结构所需的亚毫米精度。将获得一个专门构建的数据集，包括面部变形的示例，并反映系统需要稳健地应对的真实世界条件，如照明不良、图像模糊或部分遮挡。未来，这项工作将为影响面部发育的疾病的基因研究大规模收集数据提供便利，并使临床翻译在各种医学领域的探索成为可能。

The goal of 3Dface@Home is to develop and test a system to robustly obtain highly accurate facial 3D reconstruction from end users' mobile devices in real world conditions, such as one's smartphone at home. It is made of a mobile application allowing the user to capture images of their face by themselves with guidance and validity check. Those images are sent to a computational server where a deep learning based facial 3D reconstruction embedding multi-view geometry constraints will be used to obtain the 3D shape of the face. The project will aim to reach sub-millimetre accuracy required for measuring anatomical structures in the context of disease affecting facial growth. A purpose-built dataset will be acquired, including examples of facial dysmorphology, and reflecting real world conditions the system needs to be robust to such as bad illumination, blurry images or partial occlusion. In the future, this work will facilitate mass collection of data for genetic studies on disease affecting facial growth and enable the exploration of clinical translation in a variety of medical fields.