cuttlefish-code
Extracting chromatophore sizes using from high-resolution imaging of cuttlefish.
Getting started
You need a hpc cluster for computations, login must be password-less:
$ ssh-keygen
$ ssh-copy-id hpc
The programs shipped here must be installed locally and on the hpc server. Most of the programs rely purely on python libs, but following libraries have to be installed manually in the system
- OpenCV
- PyQt (only needed on the local machine)
The code must be checkout out locally and on the hpc system. After checkout, the install.sh must be run:
$ git checkout https://github.molgen.mpg.de/MPIBR-huelsdunkp/cuttlefish-code
$ ./install.sh
$ ssh hpc
$ git checkout https://github.molgen.mpg.de/MPIBR-huelsdunkp/cuttlefish-code
$ ./install.sh
Make sure ~/.local/bin is in the $PATH enviroment variable, both locally and on the hpc
$ echo $PATH
$ ssh hpc 'echo $PATH'
If not present, add export PATH=$HOME/.local/bin:$PATH into the ~/.bashrc of the local or hpc machine, respectively.
Locally create a ~/.cuttleline.cfg file with the contents of where to find the data and where to write the data:
$ cat ~/.cuttleline.cfg
sepia_dir=[Data directory]
If the output files should be written to some other directory add the following line into the configuration file:
outdir=[Output directory]
On the hpc system a configuration file has to be added in order to tell any program how to queue its scripts into the SLURM queue. Therefore, add a ~/.cuttleline.cfg:
$ ssh hpc
$ cat ~/.cuttleline.cfg
working_dir=[temporary directory on the hpc system]
code_dir=[location of the cloned git directory]
partition=[default SLURM partition to use]
mpicmd=[should be mpiexec or srun]
Processing a video
Training a classifier
Images are annotated and classifiers trained using the chromas package.