Animl comprises a variety of machine learning tools for analyzing ecological data. This Python package includes a set of functions to classify subjects within camera trap field data and can handle both images and videos.
MIT License
6
stars
3
forks
source link
Missing dependencies in package setup (and setup documentation) #16
When installing the animl package, dependencies of the package are not currently being automatically installed. For example, pandas (which is a dependency of the package) is not installed. Typically, we'd make sure to gather all the dependencies of the package into the setup.cfg file (or equivalent) so they are installed when a user does pip install animl. In the case of this project, for certain kinds of installs, it might be that we'd only want some of the dependencies to be installed automatically with perhaps some of the tensorflow bits and pieces installed manually, particularly in the case of a GPU install. We should do some thinking about that. However we resolve that matter, though, the process of installing the package and its dependencies, particularly any manual installs, should be documented in the README so new users know what they need to do.
I know that in a lot of cases we are using animl-py in a conda environment but I think we can also make it work for straight pip installs from PyPI (and with pip install -e for local dev installs), for at least CPU execution, without much work and that that would help with development and testing.
When installing the
animl
package, dependencies of the package are not currently being automatically installed. For example,pandas
(which is a dependency of the package) is not installed. Typically, we'd make sure to gather all the dependencies of the package into thesetup.cfg
file (or equivalent) so they are installed when a user doespip install animl
. In the case of this project, for certain kinds of installs, it might be that we'd only want some of the dependencies to be installed automatically with perhaps some of the tensorflow bits and pieces installed manually, particularly in the case of a GPU install. We should do some thinking about that. However we resolve that matter, though, the process of installing the package and its dependencies, particularly any manual installs, should be documented in the README so new users know what they need to do.I know that in a lot of cases we are using
animl-py
in a conda environment but I think we can also make it work for straight pip installs from PyPI (and withpip install -e
for local dev installs), for at least CPU execution, without much work and that that would help with development and testing.