XUE, Tianfan


XUE, Tianfan
MIT Computer Vision
Computer Science and Artificial Intelligence Laboratory (CSAIL)
Department of Electrical Engineering and Computer Science (EECS)
Massachusetts Institute of Technology
Supervisor: Prof. William T. Freeman

Email: loading...
32-D460, The Stata Center, Building 32 32 Vassar Street
Cambridge, MA 02139


I am currently the third year Ph.D. student in MIT Computer Science and Artificial Intelligence Laboratory. Before that, I received my M.Phil also in the Information Engineering Department, The Chinese University of Hong Kong in 2011, under the supervised of Prof. Xiaoou Tang and Prof. Jianzhuang Liu. I received my B.E. degree in the Computer Science and Technology Department from Tsinghua University, China, in 2009. My current supervisor is Prof. William T. Freeman, and I also worked closely with Prof. Frédo Durand and Dr. Michael Rubinstein. Here is my Curriculum Vitae [PDF].

My research interest includes computer vision, computer graphics, pattern recognition and multimedia processing.



Selected Projects

RefractionWiggles for Measuring Fluid Depth and Velocity from Video


We present principled algorithms for measuring the velocity and 3D location of refractive fluids, such as hot air or gas, from natural videos with textured backgrounds. Our main observation is that intensity variations related to movements of refractive fluid elements, as observed by one or more video cameras, are consistent over small space-time volumes. We call these intensity variations “refraction wiggles”, and use them as features for tracking and stereo fusion to recover the fluid motion and depth from video sequences. Based wiggle features, we give algorithms for 1) measuring the (2D, projected) motion of refractive fluids in monocular videos, and 2) recovering the 3D position of points on the fluid from stereo cameras.

This work will appear in ECCV 2014 (Oral) [PDF] [Slides] [Data] [Website] [Presentation].

Example-Based 3D Object Reconstruction from Line Drawings


We propose a novel approach, called example-based 3D object reconstruction from line drawings. This method is based on the observation that a natural or man-made complex 3D object normally consists of a set of basic 3D objects.

This work appeared in CVPR 2012 [PDF] [Poster].

Symmetric Piecewise Planar Object Reconstruction from a Single Image


We proposed an automatic 3D reconstruction algorithm from a single view of symmetric object taking advantage of symmetric information.

This work appeared in CVPR 2011 [PDF].

Object Cut: Complex 3D object reconstruction through line drawing separation


We propose a new 3D reconstruction method to recover 3D geometry from line drawing. It decomposes a complex line drawing into simple ones using the geometry property of manifold. The propose methods can greatly reduce the searching space in the reconstruction.

This work appeared in CVPR 2010 [PDF].

Image Reranking by Example: A Semi-supervised Learning Formulation


We proposed a new semi-supervised learning methods for example image based reranking. We adopts a graph-based formulation to solve this problem. We further deduct a closed form solution and an iterative solution for this problem. Detailed description is in [PDF].

Human Tracking using Online-Offline detector


We designed a human tracking framework combining an online detector using color histogram, and an offline detector using HOG, in which occlusion cases are dealt with by a novel EM algorithm. The project won Outstanding undergraduate thesis of Computer Science and Engineering Department, Tsinghua Univ. [PPT].