3D modeling and reconstruction from sparse input using deep neural networks

Speaker:        Xiaoguang Han
                Department of Computer Science
                University of Hong Kong

Title:          "3D modeling and reconstruction from sparse input using
                 deep neural networks"

Date:           Monday, 8 May 2017

Time:           4:00pm - 5:00pm

Venue:          Lecture Theater F (near lifts 25/26), HKUST

Abstract:

In this talk, I will majorly introduce our work about deepsketch2face,
which is a deep learning based sketching system for 3D face and caricature
modeling. The system has labor-efficient sketching interface, that allows
the user to draw freehand imprecise yet expressive 2D lines representing
the contours of facial features. A novel CNN based deep regression network
is designed for inferring 3D face models from 2D sketches. Our system also
supports gesture based interactions for users to further manipulate
initial face models. Both user studies and numerical results indicate that
our sketching system can help users create face models quickly and
effectively. Given time, I will also wish to introduce our new work about
3D point cloud completion using a novel deep neural network architecture.


***********************
Biography:

Xiaoguang Han is currently a final-year Ph.D. student with the Department
of Computer Science at the University of Hong Kong since 2013. He received
his M.Sc. in Applied Mathematics (2011) from Zhejiang University, and his
B.S. in Information and Computer Science (2009) from Nanjing University of
Aeronautics and Astronautics, China. He was also a Research Associate of
School of Creative Media at City University of Hong Kong during 2011 to
2013. His research interests include Computer Graphics, Computer Vision
and Computational Geometry, especially on image/video editing, 3D
reconstruction, discrete geodesic computing. His current research focuses
on high-quality 3D modeling and reconstruction using deep neural networks.