Hey,
I’ve been working lately on a project which involved deploying a model in a more advanced way than what the tutorial shows (how to use it as an API and not just a web-app, how to build a mobile app using this model, how to make an image-to-image API and not just image-to-class, …).
The result is a repo that I hope can be useful for other people who wish to deploy fastai2 models. In particular, it features:
- A web API using FastAPI that handles both image-to-class and image-to-image models
- A jupyter notebook that shows how to call the back-end APIs
- A mobile app written in Flutter based on this repo.
- A Web App similar to all the previous fastai webapps
Btw, it’s also a good opportunity for me to thank here @muellerzr, since he helped me on the forum several times for this project, and I used his boilerplate as a basis for mine 
.
Here are some screenshots of the app:
Please let me know if you have any comment or feedback!
