In case you end up using our source code in research, please add a reference to one of our relevant publications:
Binary UMF detection:
I. Szentandrasi, M. Zacharias, J. Havel, A. Herout, M. Dubska, and R. Kajan. Uniform Marker Fields: Camera localization by orientable De Bruijn tori. In ISMAR 2012, 2012
Grayscale UMF detection:
Adam Herout, Istvan Szentandrasi, Michal Zacharias, Marketa Dubska, and Rudolf Kajan. Five shades of grey for fast and reliable camera pose estimation. In Proceedings of CVPR, pages 1384–1390. IEEE Computer Society, 2013
Unity plugin or Chromakeying UMF:
I. Szentandrasi, M. Dubska, M. Zacharias, and A. Herout. Poor man’s virtual camera: Real-time simultaneous matting and camera pose estimation. IEEE Computer Graphics and Applications, 2016
Dependencies
- argtable2
- opencv [optional | recommended]
- libdc1394-2 [optional] - firewire support
- Halide [optional] - optimized code
Included
- Eigen3
- glm
$ touch .depend
$ make depend
$ make all
To generate halide
$ cd ext/halide
$ export HALIDE_PATH=/your/path/to/halide
$ make all
This generates an object archive for linux (you should add the USE_NATIVE in defines.h, and add the ext/lib/native_x.a to Makefile) and assembly files for android build.
Dependencies
- Visual Studio 2013
- opencv
- Halide
- argtable2
- environment variables (see vsproj/README.txt)
Open solution: vsproj/UMF/UMF.sln This should generate the native libraries with Halide and a .dll