For every-time use, add the following lines (with paths adjusted) to your ~/.bashrc
LABELFUSION_SOURCE_DIR=/path/to/LabelFusion DIRECTOR_INSTALL_DIR=/path/to/Director/install source $LABELFUSION_SOURCE_DIR/setup_environment.sh
Create a symlink from your data folder to LabelFusion/data. For example. (Make sure that ./data doesn't exist before calling this command, or the linking will be nested):
ln -s /path/to/data ./data
If the data is hosted on another computer, you can use sshfs to mount the data from the remote computer:
mkdir -p ./data # create mount point sshfs user@hostname:path/to/data ./data
Source the setup_environment.sh file. This file appends LabelFusion/modules to the PYTHONPATH so that Python scripts can import the labelfusion module. The labelfusion module contains reusable functions and utilities that are used by the scripts.
Checkout the branch pf/corl-master
from the repository github.com/peteflorence/director.git
. This will serve as our internal director branch. The idea is that these changes will make their way back to director/master eventually, but that process shouldn't slow down our workflow.
Install CUDA (we have verified that ElasticFusion works on CUDA 7.5 and 8.0)
Install ElasticFusion:
git clone [email protected]:peteflorence/ElasticFusion.git cd ElasticFusion ./build.sh
The build script is great, but if it doesn't work, follow the build instructions in the README.md.
Install SegNet scripts:
cd / git clone [email protected]:peteflorence/SegNet-Tutorial.git mv SegNet-Tutorial SegNet && cd SegNet git checkout pf-moving-camera
Install caffe-segnet inside of SegNet and test, and install Python bindings:
git clone [email protected]:alexgkendall/caffe-segnet.git cp Makefile.config.example Makefile.config # Adjust Makefile.config (for example, if using Anaconda Python, or if cuDNN is desired) make all make test make runtest make pycaffe
These are the key commands, but if you need full instructions, see http://mi.eng.cam.ac.uk/projects/segnet/tutorial.html and http://caffe.berkeleyvision.org/installation.html. (Note that caffe-segnet is particular distribution of segnet, you do not need to separately install caffe.)