The homework for Lesson 2 of fastai was to deploy our deep learning models. I followed the tutorial on the course site, Deploying on Google App Engine, to setup the Google App, and the GCP tutorial to setup a Jupyter Notebook environment to create a web app to classify three types of dogs: Bichon Frise, Maltese and Poodle. My dog is a Bichon Frise/Maltese mix, and he was my inspiration for this project.
I modified a couple lines in the google-engine-app starter app given in the course, with my particular model's Google Drive raw download link and a list of classes.
Here is the link to the repo for my web app, which I had to un-deploy (?) because the VM instance was eating too many of my $300 GCP credits. Here is a screen recording where I upload one image each of a Bichon Frise, Maltese and Poodle and test my model (which correctly classifies all three!):