Start the database
To run the database:
docker run --name tabletop --rm --publish 5985:5984 shadowrobot/tabletop-detection
We're remapping the standard 5984 port to 5985 on the host so that it doesn't clash with a couchdb running locally.
Start the object recognition pipeline
If not already installed, install it:
sudo apt-get install ros-indigo-object-recognition-*
Start your kinect, don't forget to set the depth registration from the driver's dynamic reconfigure.
Download the detection config.
Then run the recognition pipeline pointing to that config file:
rosrun object_recognition_core detection -c ~/Downloads/detection.object.ros.ork
Building the docker image
sr-build-tools/docker/tabletop_detection, create a folder objects and copy the meshes you want to add to the database with a sensible name for each file (it'll be used as the name of the object in the db) in that folder.
docker build -t shadowrobot/tabletop-detection . . Don't forget to push the new image if you want:
docker push shadowrobot/tabletop-detection.