Shape evolution and image segmentation by elastica regularization and graph cuts.
- gf-shape-evolution: Evolves a digital shape towards the shape of minimum Elastica energy.
- interactive-seg. Interactive segmentation with Elastica regularization.
Create a container from the image danoan/graphflow-vnc
docker run --rm -it danoan/graphflow:0.2.8.1vnc
To execute the applications, type
./graph-flow/install/bin/gf-shape-evolution <output_folder>
./graph-flow/interactive-seg.sh <output_folder>
It is recommended to create a vnc server in the container and access it via a vnc client to make use of the graphical interface of the applications.
Start the vnc server in the container
./graph-flow/scripts/start-vnc.sh
Next, check the container IP address with docker inspect
docker inspect <container_id>
With a VNC client, connect it to the container vnc server: <container_IP>:1
You should be able to access the container via a graphical interface now. Use
the flag -d
to display the shape evolution.
./graph-flow/install/bin/gf-shape-evolution -d <output_folder>
Before proceeding with the build and installation, make sure the dependencies below are installed
- libboost1.70.0-dev
- opencv-3.4
- DGtal1.1
- lemon1.3.1
- cmake>=3.10
- qt5-default
- zlib1g-dev
- libgtk2.0-dev
- graphviz
- g++
- gfortran
- libgraphicsmagick++1-dev
- wget
Next, open a console and type the commands below
cd [GRAPHFLOW_ROOT_FOLDER]
mkdir build
cd build
cmake ..
make install
If all dependencies are installed in their standard locations, that should be enough. Otherwise, if errors occur, it is likely that a manual configuration should be done. In this scenario, ccmake (cmake-curses-gui package) is quite handy.
For a detailed description, type: <application_executable> -?
We give output examples of the main applications: gf-shape-evolution and interactive-seg
Evolve a digital shape towards the shape of minimum elastica energy
./gf-shape-evolution -Striangle -r15 -h0.1
./gf-shape-evolution -Sflower -r15 -h0.1
Load a image, select foreground/background seeds and then execute the gf-segmentation algorithm.
./interactive-seg.sh -I input/images/coala.jpg output/coala
Check the influence of each parameter in this report.