Skip to content
Source code for my blog post series "On-device training with Core ML"
Jupyter Notebook Swift Python
Branch: master
Clone or download
Latest commit affcf57 Aug 11, 2019
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
Dataset Code for part 2 Aug 10, 2019
Models Code for part 2 Aug 10, 2019
Scripts Code for part 2 Aug 10, 2019
iOS App Code for part 2 Aug 10, 2019
.gitignore
LICENSE Initial commit Jul 19, 2019
README.markdown Code for part 2 Aug 10, 2019

README.markdown

Training with Core ML 3

This is the sample code for my blog post series On-device training with Core ML.

Included are:

  • Dataset: a small dataset of 30 training images and 15 test images

  • iOS App: the source code of the demo app described in the blog post

  • Models: the empty and pre-trained models used by the app

    • TuriOriginal.mlmodel: the SqueezeNet classifier trained by Turi Create
    • HandsTuri.mlmodel: the TuriOriginal model but made updatable
    • HandsEmpty.mlmodel: like HandsTuri but with a classifier layer that has random weights
    • HandskNN.mlmodel: like TuriOriginal but with an untrained k-Nearest Neighbors classifier
  • Scripts:

    • make_nn.py: converts TuriOriginal.mlmodel to HandsTuri and HandsEmpty.mlmodel
    • make_knn.py: creates the k-Nearest Neighbor model, HandskNN.mlmodel
    • TuriCreate.ipynb: the notebook used to train TuriOriginal.mlmodel

Credits:

The source code is copyright 2019 M.I. Hollemans and licensed under the terms of the MIT license.

You can’t perform that action at this time.