Ultralytics

🌟 Summary
The v8.2.81 release of Ultralytics YOLO introduces significant documentation improvements and a refined model upload process to enhance usability and reliability for machine learning practitioners.

📊 Key Changes
- **Documentation Enhancements**: Updated how code examples and citations are presented across various dataset guides to improve readability and user accessibility.
- **Model Upload Process**: Improved handling of model uploads with added safeguards and informative logging, particularly in transient environments like Google Colab.
- **Technical Adjustment**: Corrected bounding box conversion details in the results module, changing from top-left coordinates to center-point coordinates.

🎯 Purpose & Impact
- **Easier Understanding and Use**: The new documentation format with examples in both Python and CLI makes it more straightforward for users to implement various datasets in their projects.
- **Robust Model Management**: Enhanced model upload functionality ensures continuity and mitigates issues in cloud-based environments, promoting more reliable training sessions.
- **Accurate Representation**: The bounding box conversion adjustment aligns technical descriptions with actual functionalities, ensuring users correctly apply transformations.

What's Changed
* Add https://youtu.be/WwXnljc7ZUM to docs by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/15751
* Improve Docs dataset layout issues by jk4e in https://github.com/ultralytics/ultralytics/pull/15696
* Fix `xywh` docstring by Laughing-q in https://github.com/ultralytics/ultralytics/pull/15765
* Ensure matplotlib backend gets reset with plt_settings by rschroll in https://github.com/ultralytics/ultralytics/pull/15759
* `ultralytics 8.2.81` fix HUB missing 'best.pt' resumed checkpoint upload by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/15754

New Contributors
* rschroll made their first contribution in https://github.com/ultralytics/ultralytics/pull/15759

**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.2.80...v8.2.81

🌟 Summary

The v8.2.80 release of the Ultralytics YOLO software brings a mix of behind-the-scenes improvements, documentation updates, and usability enhancements, particularly around data handling and workflow automation. 📦

📊 Key Changes

- **GitHub Workflows Adjustment**: Modifications made to GitHub Actions workflows to streamline pull request and publication processes by refining the way PR details are extracted.
- **Enhanced Dataset Management**: Added a new section in documentation about downloading datasets directly from Ultralytics HUB.
- **Documentation Language Links Update**: Some corrections and improvements in language link references in both the main README and tutorials.
- **Model Interface Improvement**: Expanded support for input types, including PIL (Python Imaging Library) images, for more versatile image processing.
- **Metrics Code Update**: Tweaked how class names are handled in plotting functions, shifting from list to dictionary format for better identification and flexibility.

🎯 Purpose & Impact

- **Streamlined Processes**: The workflow updates aim to improve developer efficiency and reduce errors in managing code changes, enhancing the overall continuous integration experience. 🚀
- **User-Friendly Dataset Access**: The added documentation helps users easily download datasets, making data preparation more straightforward and less time-consuming. 📚
- **Better Compatibility**: By accepting PIL images, the model becomes more versatile, especially beneficial for users integrating with various image processing libraries. 🖼️
- **Improved Accuracy**: Using dictionaries for class names in metrics promotes better clarity and precision in data interpretation and result visualization, aiding in more insightful analysis. 📈

What's Changed
* Remove Hindi and Dutch languages by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/15703
* Remove Hindi and Dutch languages by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/15704
* Update publish.yml to use `gh` by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/15686
* Update HUB Docs for new dataset download feature by sergiuwaxmann in https://github.com/ultralytics/ultralytics/pull/15728
* Fix model `names` docstring type to `dict` by yuefanhao in https://github.com/ultralytics/ultralytics/pull/15726
* `ultralytics 8.2.80` add PIL type hints to `model()` by JoshJarabek7 in https://github.com/ultralytics/ultralytics/pull/15719

New Contributors
* yuefanhao made their first contribution in https://github.com/ultralytics/ultralytics/pull/15726
* JoshJarabek7 made their first contribution in https://github.com/ultralytics/ultralytics/pull/15719

**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.2.79...v8.2.80

🌟 Summary
The v8.2.79 update for Ultralytics YOLO brings improvements in documentation workflows, a refined deployment process, and better compatibility for inference tasks.

📊 Key Changes
- **Documentation Workflow Overhaul**: The documentation workflow (`docs.yml`) has been established as a separate process to streamline the documentation updates for user clarity.
- **Publishing Process Updates**: Adjustments to the publishing workflow ensure cleaner and more efficient version checking and deployment, focusing now solely on PyPI without doc deployments.
- **Inference Resolution Change**: The inference example now uses a resolution of 640x640 instead of 640x480 for improved detection precision.
- **Enhanced Compatibility**: Updates in the model's post-processing provide better compatibility with Apple's MPS and CoreML.

🎯 Purpose & Impact
- **Documentation Clarity and Efficiency**: Separating doc updates into a dedicated workflow allows for more organized and error-free documentation updates, ensuring users have access to the latest insights and instructions.
- **Streamlined Release Management**: Fine-tuning the release process focuses the development efforts on core functionalities and ensures versions are correctly managed and published.
- **Improved Detection Performance**: The updated resolution in inference examples enhances detection accuracy, making the solution better suited for high-resolution object detection tasks.
- **Broadened Compatibility**: The changes ensure that Ultralytics software runs smoothly on macOS, expanding the capability across different hardware setups and enhancing user experience.

These enhancements strive to make Ultralytics' tools more robust, user-friendly, and functionally versatile for all users, from developers to dataset trainers. 🚀

What's Changed
* Split Docs action into separate workflow by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/15656
* Add https://youtu.be/28JV4rbzklM to docs by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/15663
* Fix YOLOv8 C++ Example model input size by AD-lite24 in https://github.com/ultralytics/ultralytics/pull/15665
* Skip Docs push if no changes by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/15675
* Simplify publish action by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/15657
* Bump contributor-assistant/github-action from 2.4.0 to 2.5.1 in /.github/workflows by dependabot in https://github.com/ultralytics/ultralytics/pull/15681
* `ultralytics 8.2.79` YOLOv10 CoreML and MPS training "gather" op error fix by Oil3 in https://github.com/ultralytics/ultralytics/pull/15672

New Contributors
* AD-lite24 made their first contribution in https://github.com/ultralytics/ultralytics/pull/15665
* Oil3 made their first contribution in https://github.com/ultralytics/ultralytics/pull/15672

**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.2.78...v8.2.79

🌟 Summary
Ultralytics 8.2.78 introduces several improvements to the code formatting process, updates links, and makes minor bug fixes to further enhance the project.

📊 Key Changes
- **GitHub Actions Update:** Modify the formatting workflow to handle additional file types and include a 'review_requested' trigger.
- **README and Documentation Updates:** Update URLs for YOLO Vision events across multiple README files and documentation.
- **Code Refactor:** Improved naming consistency and added docstrings in example scripts.
- **Notebook Enhancements:** Update introductory text and examples in several notebooks.

🎯 Purpose & Impact
- **Enhanced Code Formatting:** The updated workflow now formats YAML, JSON, Markdown, and CSS files and includes a new trigger for 'review_requested' events, ensuring comprehensive code quality.
- *Impact:* Streamlined contributions with consistent formatting across various file types.
- **Updated Links:** Pointing to the latest events and resources ensures that users access the most current information.
- *Impact:* Easier navigation to relevant events and resources.
- **Refactored Examples and Notebooks:** Improved clarity and functionality, including better documentation and consistent naming.
- *Impact:* Easier for users to understand and follow example scripts, enhancing the learning experience.

By integrating these updates, Ultralytics takes a step towards more refined and user-friendly development practices. 🛠️📈

🌟 Summary
Ultralytics' v8.2.77 introduces a variety of enhancements, including technical infrastructure improvements and updated documentation for improved user and developer experience.

📊 Key Changes
- **Cleanup Tool Cache**: Added a step to free up space on the GitHub Actions runner.
- **Removal of `.pre-commit-config.yaml`**: Deleted the configuration file used for pre-commit hooks.
- **Documentation Updates**: Improved contributing guide to provide clearer instructions and visuals.
- **New Parameter in YOLOv8 Plot Function**: Added `color_mode` parameter to `plot` method.
- **Inference Modifications**: Modified device check conditions in DDP training.

🎯 Purpose & Impact
- **Enhanced CI/CD Efficiency**: The cleanup step aims to free up disk space before installing packages, potentially preventing failures due to lack of space.
- **Streamlined Codebase**: Removing the `.pre-commit-config.yaml` simplifies the repository by eliminating unnecessary configuration, making it lighter and easier to manage.
- **Contributor Friendliness**: Improved documentation provides a more welcoming and guiding environment for new contributors, encouraging community engagement.
- **Visualization Flexibility**: The addition of `color_mode` to the `plot` method allows for instance-based or class-based color settings, providing more customization options in visual outputs.
- **Training and Inference Optimization**: Adjustments to device handling enable more robust handling of non-GPU environments, such as CPU or MPS.

These updates collectively enhance both the developer and user experience, making the project more efficient, accessible, and customizable. 🚀

What's Changed
* Update Contributing guidelines by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/15373
* Fixed multiscale preprocess_batch by ambitious-octopus in https://github.com/ultralytics/ultralytics/pull/15392
* Improve trainer DDP device handling by alanZee in https://github.com/ultralytics/ultralytics/pull/15383
* Update Conda CI by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/15443
* Update Tracker docstrings by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/15469
* `ultralytics 8.2.77` new `color_mode=instance` plot arg by Laughing-q in https://github.com/ultralytics/ultralytics/pull/15034

New Contributors
* alanZee made their first contribution in https://github.com/ultralytics/ultralytics/pull/15383

**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.2.76...v8.2.77

🌟 Summary
Ultralytics 'v8.2.76' release includes updates focused on improving documentation, adding new plugins, and enhancing code functionality.

📊 Key Changes
- **Documentation Updates**:
- Introduced `mkdocs-macros-plugin` for better content duplication across docs.
- Added video support in documentation for enhanced visualization.
- Corrected bibliography formatting and references.
- **Code Enhancements**:
- Improved error handling and handling of large inputs in the SAHI integration example.
- Added a workaround for YouTube test skips in GitHub Actions to avoid unauthorized errors.
- Adjusted `convert_segment_masks_to_yolo_seg` function for better user guidance.
- **Dependencies**:
- Included `mkdocs-macros-plugin` in the list of development dependencies for documentation builds.

🎯 Purpose & Impact
- **Better Documentation Management**:
- Usage of `mkdocs-macros-plugin` allows for content reuse, making docs maintenance easier and more consistent.
- Enhanced documentation aesthetics and correctness help users better understand and utilize the tools.
- **Improved Code Base**:
- The SAHI example now provides more robust handling, which may prevent runtime errors when processing videos.
- Setting up better error skips in testing ensures smoother Continuous Integration (CI) workflows.
- **User Guidance**:
- The `convert_segment_masks_to_yolo_seg` function now includes clearer instructions, aiding users in effectively preparing their datasets.

These improvements collectively enhance the user experience by providing clearer documentation, more robust code examples, and smoother testing and deployment workflows. 🌟

What's Changed
* Update Docs CSS by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/15062
* Use macros for Docs tables de-duplication by ambitious-octopus in https://github.com/ultralytics/ultralytics/pull/14990
* Delete `/macros` dir from Docs site by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/15068
* Delete macros from sitemap.xml by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/15105
* Add https://youtu.be/EeDd5P4eS6A to docs by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/15107
* Optimized SAHI video inference by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/15183
* Update `convert_segment_masks_to_yolo_seg` to support custom datasets by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/15176
* `ultralytics 8.2.76` Autobackend TensorRT/Triton Segment metadata fix by Y-T-G in https://github.com/ultralytics/ultralytics/pull/15185


**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.2.75...v8.2.76