Create export.py (or something similar) for exporting models from PyTorch -> CoreML

[mrdbourke]

Could have multiple export types as well:

• PyTorch -> CoreML (for iOS)
• PyTorch -> ONNX Web Runtime (for web)

More:

• Should be able to run python export.py for a model of multiple kinds to be created somewhere
• See here for exporting models — https://coremltools.readme.io/docs/convert-a-torchvision-model-from-pytorch
• See here for reference — https://github.com/ultralytics/yolov5/blob/master/export.py
• See also here — TFLite, ONNX, CoreML, TensorRT Export ultralytics/yolov5#251
• See here for converting to CoreML and preparing the image for PyTorch — https://coremltools.readme.io/docs/image-inputs#preprocessing-for-torch
• Write tests for comparing the CoreML model to the original PyTorch model and see how they perform compared to each other
• Read this on compressing ML Program Weights (make the model smaller for better loading) — https://coremltools.readme.io/docs/compressing-ml-program-weights

[mrdbourke]

See this notebook for an example conversion: https://github.com/mrdbourke/nutrify/blob/main/foodvision/notebooks/06_export_model.ipynb

Note: Prediction/evaluation can only take place on macOS with CoreML, should setup some code to make a prediction over ~100 images with CoreML & PyTorch for a sanity check on how the model is performing.