A face sketch by a Japanese artist
A human sketch by a Chinese artist Cao Wei.

Despite the tremendous efforts devoted to the study of human figure and face, the face and human models in computer vision are still far from being satisfactory. This is largely due to the enormous subtle details in human face, hair, dress, expression, and gesture etc. The models are often simplified for certain specific vision tasks. What is the right mathematical model? We look into how artists represent human face, hair and figures and learn from their sketches. For example, see the two sketches below. Can computers draw such sketch from images automatically? We proposed a number of generative models for face, hair and human body/clothes in the following projects

1. Multi-resolution generative model of human faces, Z.J. Xu, H.Chen, and S.C. Zhu, 2005.

2. Generative Model of human Hair, H. Chen and S.C. Zhu, 2005

3. Generative model of human body/clothes, H.Chen, Z.J. Xu, Z.Q.Liu, and S.C. Zhu, 2005

Some pioneer projects were done by Liu Ce, and Chen Hong at MSR in collaboration with Harry Shum and Yingqing Xu.

4. Example-based facial sketch H. Chen, Y. Q. Xu, H. Y. Shum, S. C. Zhu, and N. N. Zheng 2001.

5. Learning inhomogeneous Gibbs models of faces, C. Liu, S. C. Zhu, and H. Y. Shum, 2001.


Representing Humans: Face, Hair, Clothes Modeling and Sketch