catdog

a suite for training models that know cats from dogs

Setup

Using virtual environments

pyenv is a great tool for creating self-contained python environments. Rather than sully your system python's dependency namespace, use a virtual environment. Install pyenv. Once installed, wiew available python versions to install:

pyenv install -l

and then install one (preferably a recent version--i.e. >3.6):

pyenv install 3.10.7

Now to create the virtual environment. The following will create a virtual environment called "catdog":

pyenv virtualenv 3.10.7 catdog

The created environment can be activated with:

pyenv activate catdog

Now that you've got an active virtual environment, you can install this project and it's console scripts:

pip install .

This will package the current project's current working state and install the catdog project along with its dependencies. If you're wanting to avoid the need to re-install after changing project files, please do install using pip's "editable mode".

Usage

catdog comes which a couple of console scripts:

• predict
• train

Future work will improve the ergonomics of these scripts. The get the packaged model (clerverly named "model.h5") to classify your sample image, place an image of a cat or dog in the "data/samples/" directory and run something like this:

predict model.h5 <sample-file-name>