Export from TensorFlow

Models built with TensorFlow should be converted to the TensorflowLite format before importing to Lens Studio. Here is the link to the official TensorFlow Lite Converter guide. The TensorFlow Lite converter takes a TensorFlow model and generates a TensorFlow Lite model (an optimized FlatBuffer format identified by the .tflite file extension).

In order to run on mobile devices, TensorFlow models will be converted to an internal format that can leverage limited computational powers. Our TensorFlow converter aims to provide support to popular neural network architectures and common neural operations.

Quantization

SnapML supports TensorFlow Lite 8-bit quantization. Learn more about Quantization from the Tensorflow website.

Check out the Multi Class Classification Template and corresponding training notebook to learn how to generate and quantize multi-class classification models.