Ultralytics

📊 Key Changes
- **Memory Management**: Implemented a CPU fallback mechanism for task alignment calculations if a GPU `torch.OutOfMemoryError` occurs.
- **Method Refactoring**: Introduced a helper method `_forward` to elegantly manage memory overflow conditions.
- **Docker and Documentation Fixes**: Improved Docker image tagging and fixed a broken Jetson device documentation link.
- **Enhanced Features**: Simplified documentation examples and introduced a new `RegionCounter` module for easier region-based object counting.

🎯 Purpose & Impact
- **Stability and Reliability**: By ensuring task alignment processes can run on CPU under memory pressure, the update prevents application crashes and allows users with limited GPU resources to continue operations smoothly.
- **User Experience**: The changes make YOLO operations more flexible and robust, especially in environments with constrained computational resources, helping users to maintain performance without interruptions.
- **Documentation and Usability**: Improved documentation clarity makes it easier for both new and existing users to implement video analytics and other YOLO model features effectively. The `RegionCounter` addition simplifies integrating real-time object counting in specific video regions, broadening the tool's practical applications. 🔧

These updates and enhancements ensure that users have a smoother and more reliable experience with Ultralytics YOLO, particularly in resource-constrained settings.

What's Changed
* Fix Docker `jupyter` image naming by ambitious-octopus in https://github.com/ultralytics/ultralytics/pull/17479
* Include FPs for images with no labels in confusion matrix by Y-T-G in https://github.com/ultralytics/ultralytics/pull/17481
* Simplify Solutions Docs code examples by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/17493
* Fix broken Jetson doc URL by lakshanthad in https://github.com/ultralytics/ultralytics/pull/17519
* Update models.md by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/17525
* Add return check for 'yolo solutions help' by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/17518
* Update sony-imx500.md by ServiAmirPM in https://github.com/ultralytics/ultralytics/pull/17491
* Fix `file_name` in `save_crop` by M3nxudo in https://github.com/ultralytics/ultralytics/pull/17499
* Add region counter as ultralytics solution by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/17439
* `ultralytics 8.3.30` run TAL on CPU if `torch.OutOfMemoryError` by Laughing-q in https://github.com/ultralytics/ultralytics/pull/17515

New Contributors
* ServiAmirPM made their first contribution in https://github.com/ultralytics/ultralytics/pull/17491
* M3nxudo made their first contribution in https://github.com/ultralytics/ultralytics/pull/17499

**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.3.29...v8.3.30

🌟 Summary
The v8.3.29 release has introduced a new capability in the Ultralytics YOLO framework, enabling the export of YOLOv8 models to the Sony IMX500 format. This advancement supports AI deployment on devices like Raspberry Pi AI Cameras, enhancing their utility for smart applications.

📊 Key Changes
- **Sony IMX500 Export Support**: Added the option to export models in the Sony IMX500 format, crucial for devices with constrained resources.
- **New `FXModel` Class**: Implemented for increased compatibility with torch.fx, facilitating advanced model manipulations.
- **Updated `.gitignore`**: Now ignores `*_imx_model/` directories, which store exported model artifacts.
- **Documentation and Tests**: Comprehensive documentation and tests added to cover the new export functionality, ensuring smooth user experience and reliability.

🎯 Purpose & Impact
- **Enhanced Device Integration**: The ability to export to Sony's IMX500 format allows for efficient AI processing on Raspberry Pi AI Cameras, making edge computing more viable and accessible. 🛠️🎥
- **Improved User Guidance**: The updated documentation provides clear steps for users to leverage this new feature, enabling developers to quickly integrate it into their projects. 📚
- **Streamlined Development Process**: The addition of the `FXModel` class and the support for the IMX500 format simplifies the model deployment process, further reducing the barriers to implementation on edge devices. 🖥️💡

What's Changed
* Docs and CI updates by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/17386
* Fix `model.end2end` assert by Laughing-q in https://github.com/ultralytics/ultralytics/pull/17391
* Add environment to publish.yml by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/17395
* Fix PyPI downloads links by pderrenger in https://github.com/ultralytics/ultralytics/pull/17399
* Jupyter Docker Image, allow connection by ambitious-octopus in https://github.com/ultralytics/ultralytics/pull/17400
* Fix duplicated annotations by `yolo_bbox2segment` by Laughing-q in https://github.com/ultralytics/ultralytics/pull/17401
* Add device selection for `yolo_bbox2segment` by Laughing-q in https://github.com/ultralytics/ultralytics/pull/17409
* Update validation-args.md by Y-T-G in https://github.com/ultralytics/ultralytics/pull/17411
* Use new Ultralytics Retry action by pderrenger in https://github.com/ultralytics/ultralytics/pull/17419
* Changed model.names return type hint from list to dict[int, str] by KiSchnelle in https://github.com/ultralytics/ultralytics/pull/17402
* Match Docs caption to video thumbnail text by Burhan-Q in https://github.com/ultralytics/ultralytics/pull/17424
* Fix `conf` overwrite in results.py by keeper-jie in https://github.com/ultralytics/ultralytics/pull/17384
* Fix Ray Tune error by Y-T-G in https://github.com/ultralytics/ultralytics/pull/17416
* Create sony-imx500.md standalone Docs page by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/17452
* Update bug-report-template.yml to prompt for `collect_system_info` by Y-T-G in https://github.com/ultralytics/ultralytics/pull/17466
* `ultralytics 8.3.29` Sony IMX500 export by Laughing-q in https://github.com/ultralytics/ultralytics/pull/14878

New Contributors
* keeper-jie made their first contribution in https://github.com/ultralytics/ultralytics/pull/17384
* KiSchnelle made their first contribution in https://github.com/ultralytics/ultralytics/pull/17402

**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.3.28...v8.3.29

🌟 Summary
The release of version 8.3.28 introduces new command-line interface (CLI) commands for "Solutions," allowing users to easily execute various video analytics tasks.

📊 Key Changes
- **New Solutions CLI Commands**: Users can now use CLI commands to apply different video analytics solutions without needing to modify arguments manually.
- **Additional CLI Examples**: Includes CLI examples for tasks like object counting, heatmaps, queue management, workout monitoring, speed estimation, and more, complete with customizable parameters.
- **Enhanced Auto-Annotation**: Improved auto-annotation functionality with new parameters like `max_det` to limit detections and `classes` for class-specific filtering.
- **Updated Documentation and Badges**: Enhancements in documentation accuracy with updated contributor details and added visibility via new badges in README files.
- **Rust and TFLite Examples**: New and improved examples for Rust ONNX runtime and TFLite Python integration for YOLO models.
- **New Docker Support**: Added a JupyterLab Docker image for improved interactive development support.

🎯 Purpose & Impact
- **Ease of Use**: Simplifies using video analytics solutions directly from the command line, making it more accessible for users to implement complex video tasks with YOLO models.
- **Enhanced Control**: Users gain more precise control over dataset annotation outputs, aiding in task-specific preparation.
- **Improved Documentation**: Allows for better tracking of project metrics and user interactions with enhanced visibility.
- **Robust Cross-Platform Support**: New examples and JupyterLab Docker integration support diverse environments, improving user experience and accessibility.
- **Platform Precision**: Export compatibility checks ensure smoother model conversions across different hardware setups.

Overall, this release significantly enhances usability and equips users with flexible tools for effective computer vision tasks.

What's Changed
* Fix `Bboxes` numpy.reshape by Laughing-q in https://github.com/ultralytics/ultralytics/pull/17301
* Fix MNN Raspberry Pi benchmark attempt by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/17308
* Fix mkdocs_github_authors.yaml by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/17314
* Update mkdocs_github_authors.yaml by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/17320
* Refactor TFLite example. Support FP32, Fp16, INT8 models by Y-T-G in https://github.com/ultralytics/ultralytics/pull/17317
* [Example] YOLO-Series(v5-11) ONNXRuntime Rust by jamjamjon in https://github.com/ultralytics/ultralytics/pull/17311
* Fix Docker badges by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/17321
* Add ultralytics models publication notice in citations section by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/17318
* Optimize Auto-Annotation with all args by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/17315
* New JupyterLab Dockerfile by ambitious-octopus in https://github.com/ultralytics/ultralytics/pull/17071
* Update `overlap_mask` description. by Y-T-G in https://github.com/ultralytics/ultralytics/pull/17324
* Generalized M1/M2 references to "Apple silicon" in train.md for broader inclusion by JairajJangle in https://github.com/ultralytics/ultralytics/pull/17330
* Add Albumentations Integrations Docs Page by abirami-vina in https://github.com/ultralytics/ultralytics/pull/17297
* Fix error on TensorRT export with float `workspace` value by Y-T-G in https://github.com/ultralytics/ultralytics/pull/17352
* Added Error for TFLite int8 end2end model export by ambitious-octopus in https://github.com/ultralytics/ultralytics/pull/17360
* Update kfold-cross-validation.md by M-Amrollahi in https://github.com/ultralytics/ultralytics/pull/17332
* `ultralytics 8.3.28` new Solutions CLI commands by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/17233

New Contributors
* JairajJangle made their first contribution in https://github.com/ultralytics/ultralytics/pull/17330
* M-Amrollahi made their first contribution in https://github.com/ultralytics/ultralytics/pull/17332

**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.3.27...v8.3.28

🌟 Summary
The `v8.3.27` release for Ultralytics primarily includes improvements in repository management and default training settings, alongside several enhancements to benchmarking and export functionality.

📊 Key Changes
- **Default Training Epochs**: Set a fallback of 100 epochs in `trainer.py` if no specific value is provided.
- **Author Information Update**: Refreshed GitHub avatars and usernames for contributors in documentation.
- **Removal of Jupyter Checks**: Cleaned up unnecessary checks for Jupyter notebooks in `checks.py`.
- **Benchmark Visualization**: Introduced interactive benchmark graphs in `benchmark.md` allowing dynamic model comparison through selectable checkboxes.
- **Export Compatibility**: Added checks to skip MNN export tests on Raspberry Pi and NVIDIA Jetson due to known unsupported issues.

🎯 Purpose & Impact
- 🔧 **Enhanced Training Robustness**: By setting a default epoch, training processes are safeguarded against accidental misconfiguration, ensuring a more reliable setup for users.
- 🏷️ **Better Attribution**: Updating author profiles ensures that contributors receive the correct recognition, improving engagement and transparency in the community.
- 🛠️ **Clean Codebase**: Removing obsolete Jupyter checks streamlines the code, making the library more efficient and easier to maintain.
- 📊 **User-Friendly Benchmarking**: The addition of visual tools for model comparison makes it easier for users to evaluate performance metrics, aiding in informed decision-making.
- 🛡️ **Compatibility Safeguards**: Preventing exports to MNN format on unsupported devices avoids potential errors and enhances user experience by providing clear usage boundaries.

What's Changed
* Add model comparison graphs in `benchmark.md` by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/17212
* Skip MNN export for Raspberry Pi and NVIDIA Jetson by lakshanthad in https://github.com/ultralytics/ultralytics/pull/17292
* Benchmark graph fix by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/17296
* `ultralytics 8.3.27` HUB timed training fix by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/17298


**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.3.26...v8.3.27

🌟 Summary
The release of Ultralytics version 8.3.26 focuses on enhancing performance for pose estimation tasks, improving export functionality for various formats, and incorporating user-centric updates for ease of use.

📊 Key Changes
- **Pose Task Enhancements:** Improved scaling for pose coordinates to boost accuracy in pose estimation tasks.
- **Export Improvements:** Enhanced export support, including numerical stability for TFLite and EdgeTPU, and formatting fixes for NCNN.
- **Documentation Updates:** Revised default models in example files and documentation for clarity and accuracy.
- **Export Order Fix:** Adjusted test order for MNN and NCNN formats to avoid CI errors on Windows systems.
- **Case-insensitive Optimizers:** Made optimizer selection process case-insensitive.
- **Auto Annotation Customization:** Added new parameters (confidence, IoU, image size) for more flexible image auto-annotation.

🎯 Purpose & Impact
- **Enhanced Precision:** Improved pose estimation allows more accurate tracking of movements, crucial for applications in sports analysis, healthcare, and more.
- **Expanded Versatility:** With better export support, models can be used on more diverse hardware platforms seamlessly.
- **User Experience:** Documentation and usability updates help users gain quicker, more straightforward access to advanced features and models.
- **Platform Stability:** Ordering tests correctly means smoother application durability, especially for Windows users running CI processes.
- **Simplified Workflow:** Case-insensitive optimizer names and auto-annotation parameter flexibility reduce potential friction for users, promoting a more efficient development process.

What's Changed
* Update `sam.md` and `sam-2.md` by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/17286
* Update examples/README.md by dme-compunet in https://github.com/ultralytics/ultralytics/pull/17284
* Patch MNN test order bug by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/17290
* Case-insensitive optimizer name by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/17287
* Auto annotation new parameters for SAM models by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/17288
* `ultralytics 8.3.26` EdgeTPU Pose models fix by Laughing-q in https://github.com/ultralytics/ultralytics/pull/17281


**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.3.25...v8.3.26

🌟 Summary
The `v8.3.25` release of Ultralytics brings a major update with the introduction of support for the Alibaba MNN (Mobile Neural Network) format, enabling better deployment of YOLO models on mobile and embedded systems.

📊 Key Changes
- **Alibaba MNN Support**: Added the ability to export and predict with YOLO models in the MNN format.
- **Integration Updates**: Documentation and integration guides have been enhanced to include MNN details.
- **Improved ONNX Runtime**: The ONNX Runtime has been optimized for faster inference, reducing overheads and improving performance, allowing for efficient handling of dynamic and static shape inferences.
- **Tracking Improvements**: Lowered default confidence thresholds for trackers to align with detection predictions, facilitating better tracking performance.

🎯 Purpose & Impact
- **Enhanced Mobile Deployment**: With MNN support, users can deploy models with high efficiency on mobile and ARM devices, allowing for more versatile applications.
- **Performance Optimization**: Faster ONNX inference ensures reduced runtime and improved speed, which is particularly beneficial for applications requiring real-time processing.
- **User-Friendly**: Updated thresholds for tracking models now provide more intuitive operations, aligning the outputs of tracking systems with user expectations.
- **Streamlined Updates**: Improved configuration and integration steps reduce friction during the setup of modern AI workflows, enhancing overall usability and user satisfaction.

What's Changed
* Fix arbitrary imgsz for TFLite by ambitious-octopus in https://github.com/ultralytics/ultralytics/pull/17138
* Example ORT==2.0.0-rs.5 to support onnxruntime==1.19.x by yawnBright in https://github.com/ultralytics/ultralytics/pull/16962
* Update Triton Inference Server guide by Y-T-G in https://github.com/ultralytics/ultralytics/pull/17059
* Faster ONNX inference with bindings by Y-T-G in https://github.com/ultralytics/ultralytics/pull/17184
* Notify only on first CI run by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/17241
* Decrease default confidence threshold to start tracking new tracks by Y-T-G in https://github.com/ultralytics/ultralytics/pull/17172
* Update publish.yml by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/17251
* Pin `ray` `numpy<=2.0.0` test by Laughing-q in https://github.com/ultralytics/ultralytics/pull/17245
* Update notebooks by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/17260
* Fix missing argument by Laughing-q in https://github.com/ultralytics/ultralytics/pull/17253
* Update triton-inference-server.md by Y-T-G in https://github.com/ultralytics/ultralytics/pull/17252
* Disable Ray tests by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/17266
* `ultralytics 8.3.25` Alibaba MNN export and predict support by wangzhaode in https://github.com/ultralytics/ultralytics/pull/16802

New Contributors
* yawnBright made their first contribution in https://github.com/ultralytics/ultralytics/pull/16962
* wangzhaode made their first contribution in https://github.com/ultralytics/ultralytics/pull/16802

**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.3.24...v8.3.25