In this demo, we build off the idea of the emoji scavenger hunt. The goal of the scavenger hunt is to find a real-world object that our image model recognizes as (corresponding to) a particular emoji. However, the model we're using isn't perfect, so in case it fails to recognize an object, you can click "I'm looking at it," and the model will update itself using our federated learning library.
First, install dependencies in both client and server.
Install server deps:
cd demo/emoji_hunt/server
yarn
Install client deps:
cd demo/emoji_hunt/client
yarn
To start the server:
cd demo/emoji_hunt/server
yarn dev
To start the client:
cd demo/emoji_hunt/client
yarn watch