New Release: Ultralytics v8.3.69

Discussion
,
1

:rocket: Announcing Ultralytics v8.3.69 Release: Improved Data Integration & Usability Enhancements! :tada:

:star2: Summary

We’re excited to share the latest Ultralytics v8.3.69 release, which introduces powerful new export capabilities, improved documentation, better benchmarking workflows, and essential bug fixes. This update is tailored to enhance your experience, making it even easier to work with YOLO models in diverse applications. Highlights include the game-changing to_sql() function for SQL database integration and expanded export options, as well as further refinements to stability and usability across the platform. :rocket:

:bar_chart: Key Changes in v8.3.69

New Features

  • SQL Export Capability:
    Effortlessly export YOLO model inference results to SQL databases using the new to_sql() method. Perfect for advanced analysis and data integration workflows. :file_cabinet:
    (Details in PR by @RizwanMunawar)

  • Extended Export Options:
    Save results in multiple formats—to_df, to_csv, to_xml, and to_json—for improved flexibility and user convenience.

Documentation Improvements

  • Interactive Charts:
    Added dynamic performance visualization charts to model documentation for easier and more engaging comparisons. :chart_with_upwards_trend:
    (Details in PR by @glenn-jocher)

  • Simplified YOLOv3 Tables:
    Improved layout and clarity for YOLOv3 documentation tables to support better readability. :books:
    (Details in PR by @glenn-jocher)

Benchmark Enhancements

  • Validation Checks:
    Added checks for square input sizes to improve benchmarking consistency for tasks like speed testing. :framed_picture:
    (Details in PR by @Y-T-G)

  • Improved Logging:
    Reduced verbosity during predictions and validations for a smoother user experience. :bulb:

Bug Fixes

  • Fixed AutoBatch compatibility with RT-DETR models to handle edge cases robustly. :white_check_mark:
    (Details in PR by @Laughing-q)

  • Ensured deep copy implementation for profiling, protecting models during GFLOP measurements. :lock:

CI Pipeline Adjustments

  • Temporarily disabled Windows CI and Raspberry Pi CI workflows for maintenance, ensuring overall platform stability. :hammer_and_wrench:
    (Details in PR by @lakshanthad)

:dart: Purpose & Impact

Purpose

  • Streamline export workflows with the new SQL functionality and a suite of format options.
  • Provide users with clear, engaging documentation for better research and development outcomes.
  • Refine tool usability and benchmarking workflows for an efficient experience.

Impact

  • For Developers: Save and manage results like a pro with SQL and other exports while enjoying simplified workflows.
  • For Researchers: Leverage interactive documentation for better performance insights and informed decisions.
  • For Everyone: Experience a more intuitive, polished YOLO platform for all your computer vision projects! :star2:

:hammer_and_wrench: What’s Changed?

Full Changelog: v8.3.69…v8.3.68
Release URL: v8.3.69 Release Notes

:star2: Try It Today!

Update to the latest version to explore these exciting improvements and share your feedback. Your insights help the entire YOLO community grow stronger! :orange_heart:
To upgrade, simply run:

pip install ultralytics --upgrade

We can’t wait to hear what you think. Thank you for being part of the Ultralytics journey! :raised_hands: