Firebase Machine Learning ^BETA

Machine learning for mobile developers

Add machine learning capabilities to your app

Use Firebase ML to train and deploy custom models, or use a more turn-key solution with the Cloud Vision APIs.

Deploy custom models that run on-device

Whether you are starting with an existing TensorFlow Lite model or training your own, you can use Firebase ML model deployment to distribute models to your users over the air. This reduces initial app installation size since models are downloaded by the device only when needed. It also allows you to A/B test multiple models, evaluate their performance and update models regularly without having to republish your entire app. Just upload your model to the Firebase console, and we'll take care of hosting and serving it to your app. Or if you prefer, you can deploy models directly from your ML production pipeline or Colab notebook using the Firebase Admin SDK.

Solve for common use cases with turn-key APIs

Firebase ML also comes with a set of ready-to-use cloud-based APIs for common mobile use cases: recognizing text, labeling images, and recognizing landmarks. Unlike on-device APIs, these APIs leverage the power of Google Cloud's machine learning technology to give a high level of accuracy. You simply pass in data to the library, which seamlessly makes a request to models running on Google Cloud, and get back the information you need–all in a few lines of code.

eBay Motors uses Firebase ML to quickly categorize images, reduce costs and improve user experience

eBay Motors allows users to search and find cars for sale in their area. Learn how they used AutoML Vision Edge in Firebase ML to create their own model and improve the user experience.

Documentation

Learn how to get started with ML by reviewing our technical documentation.

View docs

Pricing

Understand ML pricing.

View pricing