Realtime image recognition in ReactNative with CoreML

Today we’re open-sourcing a new ReactNative component that lets you quickly and easily integrate realtime image recognition into your apps using Apple’s CoreML framework (no servers required, all the code runs on your users’ devices).

jigsawxyz/react-native-coreml-image

What can you do with this component?

• Integrate realtime image recognition into your app with your own custom machine learning models
• Use the iPhone camera to get live classification confidences against your trained models
• Build your own machine learning models with TuriCreate and use these in ReactNative (see below for a tutorial on how to do this)
• 🌭 or not 🌭

Getting started

We’ve broken this article up into 2 key stages:

1. Creating and training a custom machine learning image recognition model with TuriCreate
2. Integrating the model you’ve created into a ReactNative app and using our component

If you don’t want to train your own custom model, and just want to jump straight into using the component, take a look at the open source models Apple has made available here, and then go straight to the second part of the article.

The Example

If you follow this example through, we’ll build a Hot Dog or Not Hot Dog model and then integrate it into a ReactNative app.

Full source code for the example is available here:

jigsawxyz/react-native-coreml-image-example