Project for EECS 504 (Computer Vision) in Fall 2018.
Team Members: Alex Groh, Hojae Lee, Steven Liu
In our experience it took multiple hours for all of our code dependencies to compile (OpenCV, PCL, Python bindings to PCL). To save you some time, and in case the install process doesn't work for you, we included a video that shows the code running (sfm.mov).
We have tested this installation in Ubuntu 16.0.4 (LTS).
- Clone this directory:
git clone https://github.com/hojaeklee/SfM_3D_Reconstruction.git
- Install python dependencies:
pip install pyyaml cython==0.25.2
- Install c++ dependencies by following the links:
- Go to the src directory by
cd src
. - Run
python main.py -d ../data -f 3dreconstruct
. This will run our pipeline on the RGBD images stored in thedata
folder, and create the3dreconstruct.pcd
file. Note-d
specifies the path to images folder, and-f
specifies the filename to save the Point Cloud Data (.pcd
) file. The filename can be freely specified. - To visualize the 3D reconstruction (
.pcd
file), runpcl_viewer 3dreconstruct.pcd
.