https://github.com/oil3/onnx-coreml-2025

Onnx to coreML converter . CoreML 8.1 - mlpackage
https://github.com/oil3/onnx-coreml-2025

coreml mlmodel mlpackage onnx onnxruntime

Last synced: 9 months ago
JSON representation

Onnx to coreML converter . CoreML 8.1 - mlpackage

• Host: GitHub
• URL: https://github.com/oil3/onnx-coreml-2025
• Owner: Oil3
• Created: 2023-12-15T04:49:35.000Z (over 2 years ago)
• Default Branch: current
• Last Pushed: 2025-03-07T09:52:04.000Z (over 1 year ago)
• Last Synced: 2025-03-24T18:47:19.726Z (over 1 year ago)
• Topics: coreml, mlmodel, mlpackage, onnx, onnxruntime
• Language: Python
• Homepage:
• Size: 34.1 MB
• Stars: 8
• Watchers: 2
• Forks: 0
• Open Issues: 3
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

          
Converting ONNX models to CoreML with ease.

A work in progress. 

`cd` into project directory, then `convert.py` to start.  

It will prompt if something can't be handled automatically.

In a [zipfile](https://github.com/Oil3/onnx-coreml-2025/blob/current/coremltoolsx/3signedLibraries.zip) I signed and notarized the libraries `libcoremlpython.so` `libmilstoragepython.so` `libmodelpackage.so`, you need to unpack them and put them back into the `coremltoolsx` folder from the `onnx-coreml-2025` folder.   

If there is an issue, you can copy your own libraries that you can find inside the oot of your existing `coremltools` python site-packages directory.     

I need to know what doesn't work so we can fix it, layer per layer.    

I mostly use Ultralytics 'Yolo' models, and there are many other ops I might even not know about. 

With coremltools 8.1, we can virtually convert everything.

The three main ideas here are: 

>make a model run -even if that means "resize bicubic" becomes "resize bilinear".  

>it has to be easy  

>every layer can be manually modified, changed or skipped