Christopher Zach

Post-doctoral Researcher
Computer Vision and Geometry Group
ETH Zürich
Universitätstrasse 6
CH-8092 Zürich

Office: CNB G 102.2
+41 44 632 44 76 (phone)

chzach -at- inf.ethz.ch

News

Dec. 2011: I am going to join the Machine Learning and Perception group at MSR Cambridge in January 2012.
Dec. 2011: Updated SfM code to version 1.1. Most relevant changes:
- Added more general non-linear least squares optimizer enabling more general instances of bundle adjustment (see V3D/Math/v3d_nonlinlsq.h).
  This one relaxes the assumption of the bipartite structure of unknowns as in standard BA (e.g. cameras and 3D points).
  The Levenberg-Marquardt implementation plus the corresponding bundle adjuster is also available as separate zip package (requiring only SuiteSparse as dependency).
- VLFeat and OpenCV can now be used for SIFT feature extraction (in addition to SiftGPU).
- Changed license to LGPL.
Oct. 2010: full structure-and-motion code available.
Since Sep. 2009 I am working in the Computer Vision and Geometry Group at ETH, Zürich
GPU-based KLT tracker updated to version 1.2

Research interests

My main research interests include

Structure from motion techniques
Dense 3D model generation from images
Computer vision methods accelerated by graphics processing units

Fore more detailed information on recent work see my research page.

Available software

The following open source software packages are available:

SfM — Structure-and-motion code
SSBA — Simple Sparse Bundle Adjustment
GPU-KLT+FLOW — Efficient tracking of features in exposure varying videos using the GPU, dense TV-L1 optical flow.

More information is available on the open source page.