Use Python and OpenCV to track pigeons (and other moving objects) for automated analysis of animal behavior experiments.
The easiest way to get started is by simpling using the provided docker-compose file, which will automatically build the corresponding Docker image:
docker-compose up
Please note, that this will also mount the project directory into the container, which runs on UID 0. This will result in the root user owning files created by the container on Linux.
If you want to use rendered GTK windows, use docker_run.sh. Note that this needs an active X11 session on the host
(or multiple quirks on Wayland, like setting xhost +local:root).
Try to import the conda environment:
conda create --name pigeon-tracker --file spec-file.txt
conda activate pigeon-tracker
Because of Python foobar, it's necessary to install pango from asmeurer repo when running on Fedora 28.
Not recommended, but might give a hint when debugging non working environemnts.
For getting mp4 playback to work, you can setup the conda environment with
conda install -c loopbio -c conda-forge -c pkgw-forge ffmpeg gtk2 opencv
On Windows you can ommit the gtk2 package when using Anaconda 5.3 Windows Installer
The hard part was getting a working opencv version that's compiled with gtk2 and ffmpeg support, so results may vary.
Example videos are managed by git-lfs. They can be found under example-videos/ and can
be retrieved using git-lfs as well (which should happen automatically if correctly installed on the machine).