NOTE: The speedups introduced in this branch of trackpy have been merged back into the main branch, available at https://github.com/soft-matter/trackpy. That version has many more features, is potentially faster, and does not have the memory bugs that are present in this version. I have switched my development work to the group effort there, and am no longer maintaining this branch. In summary: don't use the code here! --- Nathan Keim, 16 February 2014
The easiest route is to first install the free Anaconda Python distribution from Continuum Analytics.
You can install directly from this repository, no download necessary, with
pip install 'git+http://github.com/nkeim/trackpy/@numba#egg=trackpy'
Tip for novices: Be sure that the pip or python you are running in these directions belongs to the Python installation you'll use for tracking (e.g. anaconda).
This also works for upgrading to the latest version. If you want to muck about with the source code, add a -e after pip install, and a src directory will be created in your current directory.
Download the source and run
python setup.py install
in the source directory.
If you're not using Anaconda, make sure you have the following Python packages:
numbascipyandnumpy