This post is a continuation of a series of earlier blogs on Mobile Machine Learning using Firebase’s ML Kit. In this series, I’ve explored a number of resources that could be used to build smart apps.
While the default APIs provided in ML Kit is powerful in covering the basic use cases, they won’t help if you’re making an app to fit your own custom needs.
This is where a custom TensorFlow Lite model steps into the picture. Since ML Kit also allows us to run custom machine learning models on a mobile device without requiring a high-end system with a monstrous GPU, we can upload a trained TFLite model onto Firebase and create an API of our own!