Good news! there’s a module that will help with Tensorflow ML stuff made for React Native apps: @tensorflow/tfjs-react-native - npm
From the introduction it looks like it will support most of what you need for Sign2Word - and works well with Expo managed apps too
These points seemed relevant:
has full support for training and fine tuning models that TensorFlow.js supplies. You can customize models based on user data while keeping that data on the client device.
provide two new IOHandlers that allow loading models which are bundled with the app bundle itself (and thus do not require a remote network call). This also saves customized models to local storage
Image & Video Handling
See (TensorFlow.js for React Native is here! — The TensorFlow Blog for more
Further reading about the loading of models and bundling with local app assets - there are ways to do this, but may mean ejecting from the Expo managed app (not a big deal - needs native compiling with Xcode or Android studio)
See: bundleResourceIO TensorFlow.js React Native API
Other links worth checking “Assorted Guides in the Expo Guides: Guides to get things done - Expo Documentation Assorted Guides
Also the API Reference - Expo Documentation for more technical descriptions
Some specifics relevant to the questions we are discussing:
Offline Support - Expo Documentation Expo can bundle assets into your standalone binary during the build process so that they will be available immediately, even if the user has never run your app before.
Persisting data in React Native
FileSystem - Expo Documentation
AsyncStorage - Expo Documentation AsyncStorage is an unencrypted, asynchronous, persistent, key-value storage system that is global to the app. It should be used instead of LocalStorage.
Layout with Flexbox - Expo Documentation This is well described with diagrams in the docs
Routing & Navigation - Expo Documentation I’ll have a look for a good (simple) example !!