Skip to content
Implementation of Newcombe et al. CVPR 2015 DynamicFusion paper
Branch: master
Clone or download
Latest commit e3999ff Dec 27, 2017
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
apps Improved README and asserting that arguments are being passed to the … Dec 18, 2017
cmake Code is cleaner with utils moved to hpp instead of hpp + cpp Jun 27, 2017
deps Fixes in dependencies and build script Dec 27, 2017
kfusion Hopefully fixes #28 Dec 6, 2017
tests Adding Opt as submodule, improvements in documentation Dec 18, 2017
.gitignore CMake update to remove hardcoded paths Dec 27, 2017
.gitmodules Removing terra as submodule Dec 27, 2017
CMakeLists.txt More path fixes Dec 27, 2017
LICENSE.md Added kinectfusion code May 24, 2017
README.md Update README.md Dec 27, 2017
Report.md
build.sh More build fixes Dec 27, 2017
download_data.sh Trying to port to cuda, too slow to work with CPU code. Added data do… Aug 21, 2017

README.md

DynamicFusion

Implementation of Newcombe et al. 2015 DynamicFusion paper.

This project is still in active development and does not yet reproduce the results of the paper accurately.

The code is based on this KinectFusion implemenation

Building instructions:

Ubuntu 16.04

Clone dynamicfusion and dependencies.

git clone https://github.com/mihaibujanca/dynamicfusion --recursive

Install NVIDIA drivers.

  • Enable NVidia drivers (Search / Additional Drivers) selecting: "Using NVIDIA binary driver - version 375.66 from nvidia-375 (proprietary, tested)" "Using processor microcode firmware for Intel CPUs from intel-microcode (proprietary)"
  • Restart pc to complete installation

Alternatively a good tutorial with some common issues covered can be found here.

For fresh installs (this assumes you cloned your project in your home directory!):

chmod +x build.sh
./build.sh

If you are not on a fresh install, check build.sh for building instructions and dependencies.

If you want to build the tests as well, set -DBUILD_TESTS=ON.
To save frames showing the reconstruction progress, pass -DSAVE_RECONSTRUCTION_FRAMES=ON. The frames will be saved in <project_root>/output

To build documentation, go to the project root directory and execute

doxygen -g
doxygen Doxyfile

Running

./download_data 
./build/bin/dynamicfusion data/umbrella

Windows

Dependencies:

  • CUDA 5.0 or higher
  • OpenCV 2.4.8 or higher (modules opencv_core, opencv_highgui, opencv_calib3d, opencv_imgproc, opencv_viz). Make sure that WITH_VTK flag is enabled in CMake during OpenCV configuration.
  • Boost (libraries system, filesystem and program options. Only used in the demo. Tested with 1.64.0)
  • Ceres solver (Tested with version 1.13.0)

Implicit dependency (needed by opencv_viz):

  • VTK 5.8.0 or higher
  • SuiteSparse, BLAS and LAPACK for ceres Optional dependencies:
  • GTest for testing
  • Doxygen for documentation
  • OpenNI v1.5.4 for getting input straight from a kinect device.

Install NVIDIA drivers and CUDA

Optionals:

Download the dataset.
Create a data folder inside the project root directory.
Unzip the archive into data and remove any files that are not .png.
Inside data, create directories color and depth, and move color and depth frames to their corresponding folders.

To use with .oni captures or straight from a kinect device, use ./build/bin/dynamicfusion_kinect <path-to-oni> or ./build/bin/dynamicfusion_kinect <device_id>


Note: currently, the frame rate is too low (10s / frame) to be able to cope with live inputs, so it is advisable that you capture your input first.

References

DynamicFusion project page

@InProceedings{Newcombe_2015_CVPR,
author = {Newcombe, Richard A. and Fox, Dieter and Seitz, Steven M.},
title = {DynamicFusion: Reconstruction and Tracking of Non-Rigid Scenes in Real-Time},
booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2015}
}

The example dataset is taken from the VolumeDeform project.

@inbook{innmann2016volume,
author = "Innmann, Matthias and Zollh{\"o}fer, Michael and Nie{\ss}ner, Matthias and Theobalt, Christian 
         and Stamminger, Marc",
editor = "Leibe, Bastian and Matas, Jiri and Sebe, Nicu and Welling, Max",
title = "VolumeDeform: Real-Time Volumetric Non-rigid Reconstruction",
bookTitle = "Computer Vision -- ECCV 2016: 14th European Conference, Amsterdam, The Netherlands,
            October 11-14, 2016, Proceedings, Part VIII",
year = "2016",
publisher = "Springer International Publishing",
address = "Cham",
pages = "362--379",
isbn = "978-3-319-46484-8",
doi = "10.1007/978-3-319-46484-8_22",
url = "http://dx.doi.org/10.1007/978-3-319-46484-8_22"
}
You can’t perform that action at this time.