Skip to content

Conversation

@andrewginns
Copy link

@andrewginns andrewginns commented Feb 11, 2024

Disclaimer: I've never written a line of Swift or followed a tutorial on it in my life. This code was developed with the help of gpt-4-turbo-0125 and compiles and runs on XCode 15.2.

Changes

  1. Added support for .mlpackage alongside the already supported .mlmodel file format.
    • Added file format information to info.plist
    • Added UTType for the .mlpackage to Constants.swift
  2. Fixed a compilation error Initializer for conditional binding must have Optional type, not CGRect in DetectionView.swift on GeometryReader
    • Directly use the result of getVideoRect without conditional binding

Testing

Tested with a yolov8x.mlpackage generated using yolo export model=yolov8x.pt format=coreml nms=true through the latest ultralytics package

Built Application

My unofficial version 1.0.3

…izations, and UX clarity (#1)

* Add test target and Core ML pipeline coverage

* Add test hooks for video detection timing

* Fix test imports and Vision observation setup

* Fix Vision pipeline test invocation

* Link test target to app product

* Add bundle loader to test target

* Handle compiled model resource in tests

* Document expected precisionRecallCurves warning

* tests: Coverage before framework improvements

* feat: Modernise CoreML loading & inference for stateful models, safer optimizations, and UX clarity

* refact: Split test into behaviour related suites

* chore: Bump macos and xcode versions

* chore: Change CI to only run on PR
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant