How to set landscape detection on iOS

[menghengmen]

No description provided.

[pderrenger]

@menghengmen for iOS orientation handling, you'll need to implement landscape detection in your application's image preprocessing pipeline before passing frames to YOLO. Ultralytics YOLO models process input images as provided without inherent orientation awareness.

Ensure your iOS application:

1. Uses native iOS image handling (UIImage/Vision frameworks) to detect device orientation
2. Rotates input images to your desired orientation using Core Graphics transforms
3. Maintains aspect ratio and proper scaling for model input

For implementation details, see Apple's UIImage orientation documentation and the Vision framework image handling guide.

If you're using YOLO commercially on iOS, please ensure compliance with our Ultralytics Licensing requirements.

[john-rocky]

@menghengmen Thank you for your feedback!
Currently the package only works in portrait orientation, landscape is in the works.

[UltralyticsAssistant]

A potential fix for this issue has been merged in PR #233, which introduces several key improvements to the Ultralytics YOLO Flutter app. This update adds full landscape mode support, enhances detection accuracy with more robust bounding box calculations, and improves camera controls for a smoother, more flexible user experience. UI elements now adapt better to both portrait and landscape orientations, and overall app performance and stability have been refined. The PR also brings updated documentation for easier integration and development.

Special thanks to the author and contributors for their valuable work on this enhancement!

To test if these changes resolve the issue, please update your local version of the app or install directly from the main branch using:

pip install git+https://github.com/ultralytics/yolo-flutter-app.git@main

Alternatively, you may wait for the next official release if you prefer.

Feedback on whether this resolves the issue would be greatly appreciated. Thanks again for reporting the issue, and feel free to share any further feedback if you encounter any related problems!