Converting PyTorch & TensorFlow Models Into Apple Core ML Using CoreMLTools

CoreMLTools is a framework created by Apple that allows you to convert models from third-party libraries to the Core ML format.

Core ML is an Apple framework that allows developers to integrate machine learning/deep learning models into their applications. However, it does not support model creation and training, i.e., you first need to create the model in a framework like TensorFlow or PyTorch, then you can convert and use it. There are two ways you can convert your machine learning model from the framework of your choice to the Core ML model format: through an intermediary model format like ONNX or by using Apple’s own CoreMLTools Python library. 

Although ONNX works just fine for the conversion, CoreMLTools offers other useful functionalities like model optimization. Also, you’ll need to use CoreMLTools for the final conversion from ONNX format to Core ML format anyway. Currently, it supports the conversion of models created using the following libraries:

  • PyTorch
  • TensorFlow 1.x & 2.x
  • TensorFlow’s Keras APIs
  • scikit-learn
  • XGBoost
  • LibSVM


Install CoreMLTools from PyPI.

AIM Daily XO

Join our editors every weekday evening as they steer you through the most significant news of the day, introduce you to fresh perspectives, and provide unexpected moments of joy
Your newsletter subscriptions are subject to AIM Privacy Policy and Terms and Conditions.

The package is only available for macOS (10.13+) and Linux.

pip install coremltools

Download our Mobile App

You can learn more about the other methods of the installation here.

Converting models into Core ML model using CoreMLTools

From PyTorch

For conversion from PyTorch, you can either use the TorchScript object or TorchScript object saved as a .pt file. 

  1. Create and train or load a pre-trained model and set it to evaluation mode.
 import torch
 import torchvision
 import coremltools as ct
 # load a pre-trained MobileNetV2 model
 torch_model = torchvision.models.mobilenet_v2(pretrained=True)

 # set it to evaluation mode
  1. Create TorchScript object using the torch.jit.trace and random data
 random_input = torch.rand(1, 3, 224, 224) 
 traced_model = torch.jit.trace(torch_model, random_input) 
  1. Convert the TorchScript object to Core ML using the CoreMLTools convert() method and save it.
 # Convert to Core ML using the Unified Conversion API
 model = ct.convert(
     inputs=[ct.ImageType(name="input_1", shape=random_input.shape)]
  1. Make predictions on the converted model using the predict() method. 

Note: This is only supported on macOS version 10.13 or later.

 from PIL import Image
 example_image ="test.jpg").resize((224, 224))
 out_dict = model.predict({"input_1": example_image}) 

From TensorFlow

For conversion from TensorFlow 2.x you can use tf.keras.Model object, HDF5 .H5 file, SavedModel path or concrete functions. For TensorFlow 1.x, CoreMLTools also supports frozen grpah (tf.Graph) objects and .pb file path.

  1. Create and train or load a pre-trained model and set it to evaluation mode. 
 import tensorflow as tf
 import coremltools as ct
 model = tf.keras.Sequential(
         tf.keras.layers.Flatten(input_shape=(28, 28)),
         tf.keras.layers.Dense(128, activation=tf.nn.relu),
         tf.keras.layers.Dense(10, activation=tf.nn.softmax),
 # get the data
 (X_train, y_train), (x_test, y_test) = tf.keras.datasets.mnist.load_data()

 model.compile(optimizer='adam', loss='binary_crossentropy', metrics=['accuracy'])
 # fit the model, y_train, batch_size=32) 
  1. Convert the TensorFlow model object using the convert() method.
mlmodel = ct.convert(model)
  1. Make a prediction using the predict() method. 

Note: This is only supported on macOS version 10.13 or later.

  import numpy as np
  from IPython.display import display
  from PIL import Image
  img ='two.png').resize((28, 28)).convert('L')
  output =  mlmodel.predict({"input_1": example_img})

Colab Notebook of the above implementation.


For a more comprehensive understanding of CoreMLTools, refer to the following resources:

Sign up for The Deep Learning Podcast

by Vijayalakshmi Anandan

The Deep Learning Curve is a technology-based podcast hosted by Vijayalakshmi Anandan - Video Presenter and Podcaster at Analytics India Magazine. This podcast is the narrator's journey of curiosity and discovery in the world of technology.

Aditya Singh
A machine learning enthusiast with a knack for finding patterns. In my free time, I like to delve into the world of non-fiction books and video essays.

Our Upcoming Events

24th Mar, 2023 | Webinar
Women-in-Tech: Are you ready for the Techade

27-28th Apr, 2023 I Bangalore
Data Engineering Summit (DES) 2023

23 Jun, 2023 | Bangalore
MachineCon India 2023 [AI100 Awards]

21 Jul, 2023 | New York
MachineCon USA 2023 [AI100 Awards]

3 Ways to Join our Community

Telegram group

Discover special offers, top stories, upcoming events, and more.

Discord Server

Stay Connected with a larger ecosystem of data science and ML Professionals

Subscribe to our Daily newsletter

Get our daily awesome stories & videos in your inbox

Council Post: Evolution of Data Science: Skillset, Toolset, and Mindset

In my opinion, there will be considerable disorder and disarray in the near future concerning the emerging fields of data and analytics. The proliferation of platforms such as ChatGPT or Bard has generated a lot of buzz. While some users are enthusiastic about the potential benefits of generative AI and its extensive use in business and daily life, others have raised concerns regarding the accuracy, ethics, and related issues.