This repository has a collection of Open Source machine learning models which work with Apples Core ML standard.
Apple has published some of their own models. They can be downloaded here. Those published models are: SqueezeNet, Places205-GoogLeNet, ResNet50, Inception v3, VGG16 and will not be republished in this repository.
If you want your model added simply create a pull request with your repository and model added. In order to keep the quality of this repository high you have to conform to this project structure (taken from @hollance).
├── Convert
│ ├── coreml.py
│ ├── mobilenet_deploy.prototxt
│ └── synset_words.txt
├── MobileNet.mlmodel
There has to be a Convert directory with the Python script and additional data to reproduce this model on your own. If your model requires a huge amount of data please include a script which downloads those files. The .mlmodel file is optional because you'll have to push it onto this repository anyways.
├── MobileNetCoreML
│ ├── *.swift
├── MobileNetCoreML.xcodeproj
│ ├── project.pbxproj
│ └── project.xcworkspace
│ └── contents.xcworkspacedata
├── README.markdown
You also have to have an Xcode project where the user can test the model (sample data included would be nice).
Model: MobileNet.mlmodel
Description: Object detection, finegrain classification, face attributes and large scale geo-localization
Author: Matthijs Hollemans
Reference: MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
Example: MobileNet-CoreML
Model: MNIST.mlmodel
Description: Handwritten digit classification
Author: Philipp Gabriel
Reference: MNIST handwritten digit database
Example: MNIST-CoreML
Model: Food101.mlmodel
Description: Food classification
Author: Philipp Gabriel
Reference: UPMC Food-101
Example: Food101-CoreML