Read the imgsz parameter in metadata

[x2031]

Dynamically adjust the imgsz in the prediction process by reading the imgsz parameter in the metadata file.

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Enhancements in model input size flexibility and label loading for the YOLO Flutter App.

📊 Key Changes

• Introduced variability in INPUT_SIZE to accommodate different model input sizes.
• Added support for loading custom label sizes from the model metadata.

🎯 Purpose & Impact

• Adaptable Model Support: The app can now support YOLO models with different input sizes, making it more flexible for various use cases. 🔄
• Enhanced Label Handling: By loading label sizes directly from the model's metadata, the app ensures that labels are accurately matched with the detected objects, improving detection accuracy and user experience. 🏷️

These changes aim to make the YOLO Flutter app more versatile and user-friendly, catering to a wider range of models and detection scenarios.

[pderrenger]

Hi there! 👋

Thank you for reaching out! If you're looking to dynamically adjust imgsz based on the metadata, you might consider doing this right before your prediction call. Assuming your metadata is in a readable format like JSON, you could do something like this:

import json

# Assuming 'metadata.json' contains {'imgsz': 640}
with open('metadata.json') as f:
    metadata = json.load(f)

# Update imgsz
imgsz = metadata['imgsz']

# Now, proceed to use imgsz in your prediction settings

This snippet reads the imgsz from a metadata.json file, allowing you to then utilize this size in your prediction workflow. Keep the operation outside critical performance paths to ensure efficiency.

For more info on managing prediction configurations, please visit our documentation at https://docs.ultralytics.com.

Hope that helps! Let the community know if you have further questions. 😊

[github-actions[bot]]

CLA Assistant Lite bot All Contributors have signed the CLA. ✅

[x2031]

I have read the CLA Document and I sign the CLA