Using a custom TensorFlow Lite model in iOS

We have trained many custom TensorFlow models and frozen them for mobile use in the previous chapters. Unfortunately, if you try to use the bazel-bin/tensorflow/contrib/lite/toco/toco TensorFlow Lite converter tool built in the last section to convert the models from the TensorFlow format to the TensorFlow Lite format, they all will fail, except for the retrained models in Chapter 2, Classifying Images with Transfer Learning, which we covered in the previous subsection; most of the errors are of the "Converting unsupported operation" type. For example, the following command tries to convert the TensorFlow Object Detection model in Chapter 3, Detecting Objects and Their Locations, to the TensorFlow ...

Get Intelligent Mobile Projects with TensorFlow now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.