Display Val images per class

[sunmooncode]

Add a new parameter for users to choose from in the detected val script. The function of this parameter is to control the output method of images in the val result.

For more details please refer to here!

🛠️ PR Summary

_{Made with ❤️ by Ultralytics Actions}

🌟 Summary

Enhanced tracking of object detection, pose estimation, and segmentation metrics in YOLO models. 📈

📊 Key Changes

• CI Configuration: Removed a redundant line setting slow dependencies, ensuring dependencies are now conditionally installed based on the event that triggers the workflow. 🧹
• Detection, Pose, and Segmentation: Introduced tracking of unique images alongside class and prediction statistics to improve metric accuracy and reporting. 🖼️
• Metric Reporting: Enhanced metric reporting to include per-image statistics in addition to per-class statistics for deeper insights. This involves changes to logging and the calculation of nt_per_class and the newly added nt_per_image. 📊

🎯 Purpose & Impact

• Better Dependency Management: The update in CI configuration tidies up the dependency installation process, making the CI runs more efficient and less prone to errors. 🛠️
• Improved Accuracy & Insights: Adding nt_per_image allows developers and users to gain a clearer understanding of model performance on a per-image basis. This can lead to better-targeted improvements and more detailed performance analysis. 🔍
• Enhanced User Awareness: By including additional logging details, users can now receive a more granular breakdown of how the model is performing across different classes and images, enriching the user experience and aiding in debugging and model optimization. 🚀

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 70.16%. Comparing base (b95b583) to head (f235c0e).

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #12645      +/-   ##
==========================================
- Coverage   71.01%   70.16%   -0.86%     
==========================================
  Files         124      124              
  Lines       15658    15664       +6     
==========================================
- Hits        11119    10990     -129     
- Misses       4539     4674     +135     

Flag	Coverage Δ
Benchmarks	35.31% <90.90%> (-0.18%)	⬇️
GPU	37.26% <63.63%> (+0.01%)	⬆️
Tests	66.31% <100.00%> (-0.81%)	⬇️

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]

@sunmooncode hey great, this looks like a nice improvement. Once change is that can you please remove the argument so we use this change by default? Thanks!

[sunmooncode]

I deleted the optional parameter and used it by default.

[glenn-jocher]

That sounds great, @sunmooncode! Making the feature default simplifies the usage while providing enhanced insights consistently. Thanks for your contribution! 🚀 If anything else needs to be adjusted, feel free to let us know.

[glenn-jocher]

@sunmooncode can you please delete the detail_per_class: False # (bool) detailed per-class results argument entirely and we'll use this as the new default? Thank you!

[sunmooncode]

@glenn-jocher Of course. Please wait for me for a moment.

[glenn-jocher]

Sure thing, @sunmooncode! Take your time. 😊

[glenn-jocher]

@sunmooncode PR merged! Thank you for your contributions :)