Carleton 3D Scene Reconstruction Comps

Pulling the repo

We use git-lfs to track large files in the repo. Install that on OSX like this:

brew install git-lfs

or take a look at their docs.

Then run the ol' clone command:

git clone [email protected]:CarletonScenes/Scene-Reconstruction.git

Installing OpenCV for feature detection

On Mac OSX:

brew tap homebrew/science
brew install opencv3 --c++11 --with-cuda --with-contrib
brew ln opencv3 --overwrite --force

To get the environment set up

cd python
sudo easy_install pip
sudo pip install virtualenv matplotlib
virtualenv env
source ./env/bin/activate
pip install -r requirements.txt

brew install opencv3 --c++11 --with-cuda --with-contrib
brew ln opencv3 --overwrite --force
cp /usr/local/lib/python2.7/site-packages/cv* env/lib/python2.7/site-packages

At this point, you have opencv installed on your computer, copied the shared object files from opencv to the virtualenv python sources folder, and installed matplotlib to your regular python installation.

To activate the environment, run (in bash): . ./env/bin/activate export PYTHONPATH=$PYTHONPATH:/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/

Or in fish: source ./venv_activate.fish # This will both start the virtual env and set the PYTHONPATH variable.

Moving forward

Once you have the virtualenv installed, and activated (activation is custom, check out the commands above, they set the $PYTHONPATH env variable for you), you can run the program from the project root:

✗ python python/do_comps.py
Welcome to do_comps.py!
        To run this program, you'll need to select one of the
        modes below and perhaps provide more input.

        modes:
            python do_comps.py detect -i img1.jpg [-i img2.jpg ...] [-f input_folder] [-o output.jpg]
            python do_comps.py match -i img1.jpg -i img2.jpg [-i img3.jpg ...] [-f input_folder] [-o output.jpg]
            python do_comps.py triangulate -i img1.jpg -i img2.jpg [-i img3.jpg ...] [-f input_folder]
                                --scene_output scene.ply [--projection_output projection.ply]

python python/do_comps.py triangulate -i python/photos/c1.jpg -i python/photos/c2.jpg --scene_output tri_out.ply
python python/do_comps.py triangulate -i python/photos/c1.jpg -i python/photos/c2.jpg --scene_output tri_out.ply --projection_output proj_out.ply

python python/do_comps.py manual_pts python/points/pdppoints.txt -i python/photos/pdp1.jpeg -i python/photos/pdp2.jpeg --scene_output pdp_out.ply --projection_output pdp_proj.ply

Pythonpath stuff (in fish)

set -x PYTHONPATH /Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/