After working with Apple yesterday, Google's TensorFlow team announced that TensorFlow Lite officially supports Core ML. In the future, iOS developers will be able to leverage the strengths of Core ML to deploy the TensorFlow model, and TensorFlow Lite will continue to support cross-platform formats (.tflite) including iOS.
Earlier, Google announced TensorFlow Lite at the Google I / O conference in May this year, followed by Google on November 14 with a preview of the TensorFlow Lite mobile device developer preview for the machine learning framework. And said that TensorFlow Lite already supports a number of well-trained and optimized mobile models, including visual models, photo recognition models, and device dialog models, all running on both Android and iOS, and significantly improving mobile device loading Time and so on.
This time TensorFlow Lite supports Core ML by using the TensorFlow model and the tool to convert it to the Core ML model format (.mlmodel). More specific information is available at: