Summary

In this chapter, we covered two cutting-edge tools of using machine learning and deep learning models on mobile and embedded devices: TensorFlow Lite and Core ML. While TensorFlow Lite is still in developer preview, with limited support for TensorFlow operations, its future releases will support more and more TensorFlow features, while keeping the latency low and app size small. We offered step-by-step tutorials on how to develop TensorFlow Lite iOS and Android apps to classify an image from scratch. Core ML is Apple's framework for mobile developers to integrate machine learning in iOS apps, and it has great support for converting and using classical machine learning models built with Scikit Learn, as well as good support for Keras-based models. We also showed how to convert Scikit Learn and Keras models to Core ML models and use them in Objective-C and Swift apps...