Department of Computer Science and Engineering
University of Washington


3D Head Reconstruction


Project Summary:

Acquiring high-detailed 3D head shape is challenging due to the highly non-rigid nature of human's face. However, current high-detailed reconstruction methods usually require the subject to participate in the capture session with a complex setup.

The goal of this work is to provide easy 3D head reconstruction methods to generate a personalized and high-detailed 3D shape of a person's head. We explored the input using a single RGBD frame, 2D photos from the Internet or in-the-wild videos. 3D hairstyle reconstruction is also explored. 



Ira Kemelmacher-Shlizerman

Linda Shapiro 

Shu Liang




  • S. Liang, Xiufeng Huang, Xianyu Meng, Kunyao Chen, L. G. Shapiro, I. Kemelmacher-Shlizerman, ``Video to Fully Automatic 3D Hair Model'', Siggraph Asia, 2018

  • S. Liang, L. G. Shapiro, I. Kemelmacher-Shlizerman, ``Head Reconstruction from Internet Photos'', European Conference on Computer Vision, 2016

  • S. Liang, I. Kemelmacher-Shlizerman, L. G. Shapiro, ``3D Face Hallucination from a Single Kinect Frame,'' International Conference on 3D Vision, 2014.