Compile a CoreML model locally (coreml.compile_model)

Declaration

compiled_path, err = coreml.compile_model(model_path)

Parameters

• model_path
  String. Path of the model to compile. iOS 15+ supports .mlpackage models; iOS 13 and iOS 14 support legacy .mlmodel only.

  YOLO11 to CoreML documentation

  macOS command examples for converting YOLO11 formats

  # Assume you already have Python 3.10. If not, create one. (Only tested on Python 3.10)
  # Install ultralytics if not installed
  pip install ultralytics
  
  # For iOS 13 or iOS 14, export legacy .mlmodel by setting format=mlmodel
  yolo export format=mlmodel nms=True model=best.pt
  
  # For iOS 15+, export .mlpackage by setting format=coreml
  yolo export format=coreml nms=True model=best.pt
  
  # nms=True is required

Returns

• compiled_path
  String | nil. On success, the compiled model directory path; nil on failure.
• err
  String | nil. nil on success; error message on failure.

Notes

• Available since app version 1.3.8+.

• Not supported on iOS versions below 13.

• Compiles a .mlpackage or .mlmodel into a device-loadable .mlmodelc bundle.

• .mlmodelc compiled on different devices or iOS versions may not be interchangeable.

  YOLO11 training commands on an NVIDIA (CUDA) PC

  # Install Visual Studio C++ Desktop Development workload (if not already)
  # https://visualstudio.microsoft.com/
  # Install CUDA 11.8 Toolkit (if not already)
  # https://developer.nvidia.com/cuda-11-8-0-download-archive
  
  # Assume Python 3.10 is available
  pip install ultralytics
  
  # Reinstall torch libraries compatible with CUDA 11.8
  pip uninstall torch torchvision torchaudio
  pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu118
  
  yolo train model=yolo11m.pt data=COCO.yml epochs=300 imgsz=640 device=0

  YOLO11 training commands on macOS

  # Assume Python 3.10 is available
  pip install ultralytics
  
  # CPU/Metal torch
  pip install torch torchvision torchaudio
  
  yolo train model=yolo11m.pt data=COCO.yml epochs=300 imgsz=640 device=mps

Example

compiled_model_path = XXT_HOME_PATH..'/models/yolo11m.mlmodelc' -- Where to store the compiled model

local noexecute = require('no_os_execute')
noexecute.rm_rf(compiled_model_path) -- During testing, always recompile; comment out for stable models

if not file.exists(compiled_model_path) then -- Compile if not compiled yet
    local tmp_path, err = coreml.compile_model(XXT_HOME_PATH..'/models/yolo11m.mlpackage')
    if not tmp_path then
        error(err)
    end
    local noexecute = require('no_os_execute')
    noexecute.cp_r(tmp_path, compiled_model_path)
    noexecute.rm_rf(tmp_path)
end