You can cite by

@misc{jacky_liu_2018_1419222,
  author    = {Tse-An (Jacky) Liu},
  title     = {png to klg}
  version   = {1.0},
  publisher = {Zenodo},
  month     = Aug,
  year      = 2018,
  doi       = {10.5281/zenodo.1419222},
  url       = {https://github.com/HTLife/png_to_klg}
}

docker image is provided here https://github.com/HTLife/png_to_klg-docker

converted klg could be download from here

Convert TUM RGB-D png dataset to .klg format for Kintinuous and ElasticFusion

Also work with ICL-NUIM dataset (TUM RGB-D Compatible PNGs)

• Ubuntu 14.04, 15.04 or 16.04 (Though many other linux distros will work fine)
• CMake
• Boost
• zlib
• libjpeg
• OpenCV

sudo apt-get install g++ cmake cmake-gui libboost-all-dev build-essential

wget http://sourceforge.net/projects/opencvlibrary/files/opencv-unix/2.4.9/opencv-2.4.9.zip
unzip opencv-2.4.9.zip
cd opencv-2.4.9.zip
mkdir build
cd build
cmake -D BUILD_NEW_PYTHON_SUPPORT=OFF -D WITH_OPENCL=OFF -D WITH_OPENMP=ON -D INSTALL_C_EXAMPLES=OFF -D BUILD_DOCS=OFF -D BUILD_EXAMPLES=OFF -D WITH_QT=OFF -D WITH_OPENGL=OFF -D WITH_VTK=OFF -D BUILD_PERF_TESTS=OFF -D BUILD_TESTS=OFF -D WITH_CUDA=OFF -D BUILD_opencv_gpu=OFF ..
make -j8
sudo make install

• numpy

sudo apt-get install pip
pip install numpy

cd ./pngtoklg
mkdir build
cd build
cmake ..
make

All parameters are required.

• -w working directory
• -o output file name (the output file will be place under working directory)
• -r associations.txt is in reverse order (rgb)(depth)
• -t TUM format / defualt format is ICL-NUIM
• -s Scale factor in floating point. default=5000

For more scale factor detail, please reference: http://vision.in.tum.de/data/datasets/rgbd-dataset/file_formats#intrinsic_camera_calibration_of_the_kinect

Should place associations.txt under working directory. About how to generate associations.txt please read "Related files" section.

Download the file provided by ICL-NUIM (Living Room 'lr kt0') => (TUM RGB-D Compatible PNGs)

TUM RGB-D png dataset

./pngtoklg -w '/TUM/rgbd_dataset_freiburg2_desk/' -o '/TUM/rgbd_dataset_freiburg2_desk/fr2desk.klg' -t

After execute the command above, "/TUM/rgbd_dataset_freiburg2_desk" folder should contain fr2desk.klg file (about 4.4GB).

Run with ElasticFusion

./ElasticFusion -l (path to fr2desk.klg) -d 12 -c 3 -f

remove -t option which is stand for tum

./pngtoklg -w '/iclnuim/living_room_traj0_frei_png/' -o '/iclnuim/living_room_traj0_frei_png/liv.klg'

http://www.doc.ic.ac.uk/~ahanda/living_room_traj0_frei_png.tar.gz

One row contain two informations. First is time sequence. Actually the time is not important. We only need increasing number sequence.

timeSequence filePath

Sample file content

0.033333 ./rgb/scene_00_0000_rs.png
0.066666 ./rgb/scene_00_0001_rs.png
0.099999 ./rgb/scene_00_0002_rs.png
0.133332 ./rgb/scene_00_0003_rs.png
0.166665 ./rgb/scene_00_0004_rs.png
0.199998 ./rgb/scene_00_0005_rs.png
...

This code is developed by TUM, which use to associate rgb.txt and depth.txt

Type the following command

(Be aware you should remain the order of parameters, feed the depth file first and then rgb file.)

python associate.py PATH_TO_SEQUENCE/depth.txt PATH_TO_SEQUENCE/rgb.txt > associations.txt

Sample file content (TUM RGB-D dataset format) If you are using ICL-NUIM, the timestamp will be integer number

0.033333 ./depth/scene_00_0000_rs.png 0.033333 ./rgb/scene_00_0000_rs.png
0.066666 ./depth/scene_00_0001_rs.png 0.066666 ./rgb/scene_00_0001_rs.png
0.099999 ./depth/scene_00_0002_rs.png 0.099999 ./rgb/scene_00_0002_rs.png
0.133332 ./depth/scene_00_0003_rs.png 0.133332 ./rgb/scene_00_0003_rs.png
0.166665 ./depth/scene_00_0004_rs.png 0.166665 ./rgb/scene_00_0004_rs.png
0.199998 ./depth/scene_00_0005_rs.png 0.199998 ./rgb/scene_00_0005_rs.png