Demystifying ML in iOS
summary
Shows how to build an on device image classifier for flowers using Create ML and export to Core ML in minutes via transfer learning the demo trains in minutes with a small dataset and reports 100% accuracy on its test folder the workflow is explicit collect images in labeled folders run up to 10 iterations evaluate precision recall and the confusion matrix then drag the .mlmodel into Xcode which generates a Swift class use Vision to wrap the request feed a CVPixelBuffer and note limits like no on device retraining iOS 11 plus and model size tradeoffs handled by quantization
AI-written summary · about
chapters (9)
- 0:00Start
- 3:43Availability
- 5:16Compatibility
- 5:52Recap
- 15:35Confusion Matrix
- 19:14Precision
- 22:18Data Augmentation
- 29:03Machine Learning with Vision
- 31:38Vision Manager
related talks
40mNavigating the Machine Learning Landscape for Apple Devices
22mGetting Started with Apple ML
01h 03mTraining dynamic ML models on iOS 15
40mDeep Learning on iOS
42mSmartly Smartest Apps with Cross-Platform Machine Learning (Using Flutter and ML Kit)
25mGetting Started with Machine Learning in your iOS apps
55mCore ML: Vision
27mMachine Learning and Mobile