In this chapter, we will discuss ML Kit, which was announced by Firebase at the Google I/O 2018. This SDK packages Google's mobile machine learning offerings under a single umbrella.

Mobile application developers may want to implement features in their mobile apps that require machine learning capabilities. However, they may not have knowledge of machine learning concepts and which algorithms to use for which scenarios, how to build the model, train the model, and so on.

ML Kit tries to address this problem by identifying all the potential use cases for machine learning in the context of mobile devices, and providing ready-made APIs. If the correct inputs are passed to these, the required output is received, with no further coding required.

Additionally, this kit enables the inputs to be passed either to on-device APIs that work offline, or to online APIs that are hosted in the cloud.

To top it all, ML Kit also provides options for developers with expertise in machine learning, allowing them to build their own models using TensorFlow/TensorFlow Lite, and them import them into the application and invoke them using ML Kit APIs.

ML Kit also offers further useful features, such as model upgrade and monitoring capabilities (if hosted with Firebase).

We will cover the following topics in this chapter:

• ML Kit and its features
• Creating an image-labeling sample using ML Kit on-device APIs
• Creating the same sample using ML Kit cloud APIs
• Creating Face Detection application