TFLite test matrix

[glenn-jocher]

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Enhanced TensorFlow TFLite export capabilities and resolved TensorFlow installation conflicts.

📊 Key Changes

• Implemented tests for exporting YOLO models to TFLite format, ensuring compatibility and addressing performance.
• Updated TensorFlow dependency requirements, removing specific version constraints, and added Keras as a separate requirement.
• Included conditional requirements to cater for different platforms and hardware configurations, improving installation success rates.
• Excluded test cases with incompatible settings (i.e., both int8 and half precision enabled) for better clarity and reliability.

🎯 Purpose & Impact

• Clarifies TensorFlow Installations: By specifying TensorFlow and Keras as separate requirements, users should experience fewer conflicts during installation, especially on Windows and macOS 🖥️🍏.
• Enhances Model Exporting: The new tests and refinements aim to make exporting models to TFLite format smoother, potentially leading to easier deployment on mobile and edge devices 📱🚀.
• Increases Compatibility: Adjusting dependencies based on platform and hardware ensures more users can successfully set up their environment without diving into dependency resolution issues 🧩.
• Improves Test Coverage and Reliability: By focusing on realistic test scenarios and preventing incompatible configurations, the quality and predictability of exports are enhanced ✅.

Overall, these changes aim to make the process of using TensorFlow with Ultralytics projects more intuitive and trouble-free for developers and enthusiasts alike, promoting broader use and application of machine learning models 🌐.

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 1 lines in your changes are missing coverage. Please review.

Project coverage is 69.43%. Comparing base (8caa1c9) to head (f8d4f8b).

Files	Patch %	Lines
ultralytics/engine/exporter.py	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #13083      +/-   ##
==========================================
- Coverage   69.67%   69.43%   -0.24%     
==========================================
  Files         124      124              
  Lines       15655    15655              
==========================================
- Hits        10907    10870      -37     
- Misses       4748     4785      +37     

Flag	Coverage Δ
Benchmarks	35.12% <0.00%> (-0.02%)	⬇️
GPU	37.27% <0.00%> (ø)
Tests	64.89% <0.00%> (-1.03%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[glenn-jocher]

Closed in favor of #13176