This is an experiment that follows on from my blog post.
The machine learning model is stored in export.pkl and represents the machine learning model in a 'learned' state where it has layers built up of training data so it can predict the class of an image given an input image.
If you look at the GitHub Actions logs for this project, look at the passing jobs and you can see we run the model against a pair of aggressive skates and a pair of inline skates and it correcly predicts both.
This is an early experiment so won't be amazing! 👼