This repo contains the demo app from MobOS 2020, "On-device machine learning – Core ML vs Create ML".
Conference site: https://romobos.com/
With Core ML 3, on-device learning has been fully upgraded. Create ML got its own identity, now being a standalone application and is providing support for types of models such as image classifiers, object detection, sound or text classification.
You can create and train your models very easy. We will check how Create ML and Core ML were intended to work together. After this talk, you will have an overview of models creation and training, how to integrate them in your iOS app and enhance them further with on-device training.
Why is it a good talk?
I want to provide a demo of an iOS app that has a model created with Create ML via the Image Classifier and make it recognise 2-3 objects on stage.
Showing people that Core ML 3 has been given a big boost this year, making use of on-device training, for sure will turn into a great subject for this talk. And with Create ML, it has never been easy to create and train your models.
- Practical examples for On-device Training - https://machinethink.net/blog/coreml-training-part1/
- An in-depth look at Core ML 3 - https://machinethink.net/blog/new-in-coreml3/
- How to create updatable models using Core ML 3 - https://medium.com/better-programming/how-to-create-updatable-models-using-core-ml-3-cc7decd517d5
- Create ML - https://www.appcoda.com/create-ml/
- How to train your Core ML model on device - https://www.iowncode.com/how-to-train-your-core-ml-model-on-device/
- Core ML 3 Framework - https://developer.apple.com/videos/play/wwdc2019/704/
- Introducing the Create ML App - https://developer.apple.com/videos/play/wwdc2019/430/
- What's New in Machine Learning - https://developer.apple.com/videos/play/wwdc2019/209/
- Training Object Detection Models in Create ML - https://developer.apple.com/videos/play/wwdc2019/424/
- Training Recommendation Models in Create ML - https://developer.apple.com/videos/play/wwdc2019/427/
- Training Sound Classification Models in Create ML - https://developer.apple.com/videos/play/wwdc2019/425/
- Creating Great Apps Using Core ML and ARKit - https://developer.apple.com/videos/play/wwdc2019/228/
- Training with Core ML 3 - https://github.com/hollance/coreml-training
- Collection of CoreML Models - https://github.com/likedan/Awesome-CoreML-Models
- coremltools - https://github.com/apple/coremltools
- Demo for Core ML Car Recognition - https://github.com/likedan/Core-ML-Car-Recognition
- Dices - https://developer.apple.com/documentation/coreml/understanding_a_dice_roll_with_vision_and_object_detection
- Classifying images with Vision and Core ML- https://developer.apple.com/documentation/vision/classifying_images_with_vision_and_core_ml
- Personalizing a Model with On-Device Updates - https://developer.apple.com/documentation/coreml/core_ml_api/personalizing_a_model_with_on-device_updates
- Finding Answers to Questions in a Text Document - https://developer.apple.com/documentation/coreml/finding_answers_to_questions_in_a_text_document
- Recognizing Objects in Live Capture - https://developer.apple.com/documentation/vision/recognizing_objects_in_live_capture