This is the source code for the book Core ML Survival Guide -- now updated for iOS 14 and macOS 11.0.
If you don't have the book yet, here's my sales pitch: The Core ML Survival Guide is more than 80 chapters and almost 500 pages of Core ML tips and tricks.
Learn the best ways to convert your models to Core ML, how the mlmodel file format works, how to perform model surgery to fix issues with your mlmodel files, how use MLMultiArray
, and much more!
If you're serious about Core ML, you need this book. Seriously! :smile:
Included in this source code repo is the following:
CheckInputImage: Demo app for iOS that shows how to use a very basic image-to-image model to verify how Vision and the neural network preprocessing options modify the input image.
MobileNetV2+SSDLite: Conversion script and demo app for the SSDLite object detection model. (The original model is not included, you'll need to download this first. See the link in ssdlite.py.)
NeuralNetworkBuilder: Demo of how to write your own Core ML converter. Includes the trained Caffe model.
Scripts:
DOUBLE
to FLOAT32
.Also check out my other repo CoreMLHelpers for a collection of helper code that makes it a little easier to work with Core ML in Swift.