ONNX is an open format built to represent machine learning models. This notebook demonstrates the conversion process from an ONNX model (exported from MATLAB) to a tflite model (to be used within TensorFlow Lite, on an Android or iOS device.) In addition to conversion, this notebook contains cells for running inference using a set of test images to validate that predictions remain consistent across converted models.
Note: TensorFlow's API is constantly evolving. This notebook was written in November of 2019, during the transition period from TF 1.X to TF 2.X, so it is likely that relevant APIs will have updated since.