Skip to content

Releases: ultralytics/assets

v8.3.0 - New YOLO11 Models Release (#76)

29 Sep 17:59
0f20d14
Compare
Choose a tag to compare

🌟 Summary

Ultralytics YOLO11 is here! Building on the YOLOv8 foundation with R&D by @Laughing-q and @glenn-jocher in ultralytics/ultralytics#16539, YOLO11 offers cutting-edge improvements in accuracy, speed, and efficiency, redefining what's possible in real-time object detection and computer vision tasks.

YOLO11 Performance Plots

πŸ“Š Key Highlights

  • πŸš€ YOLO11 Model Unveiled: A significant upgrade over YOLOv8, YOLO11 is now the default model with enhanced architecture and optimized pipelines.
  • πŸ“š Revamped Documentation: Clearer, more detailed guides, examples, and resources to help users transition seamlessly to YOLO11.
  • πŸ› οΈ Streamlined CI & Dockerfiles: All continuous integration files and Docker environments are optimized for YOLO11, ensuring smooth workflows.
  • πŸ”„ Augmentation & Blocks Upgraded: New augmentations and block modules boost performance metrics across various tasks.
  • πŸ”§ YOLO11-Specific Configurations: Tailored model configuration files to get the most out of YOLO11's advanced features.

🎯 Purpose & Impact

  • Top-Tier Performance: YOLO11 delivers better accuracy with fewer parameters, enhancing real-time object detection and efficiency for your AI needs.
  • Versatility in Computer Vision Tasks: Supports a broader range of tasks, including object detection, instance segmentation, pose estimation, and oriented bounding box detection, adaptable across edge to cloud environments.
  • Easy Adoption: With updated resources, tutorials, and an intuitive model structure, developers can quickly adopt and maximize YOLO11's capabilities.

What's Changed

New Contributors

  • @Y-T-G made their first contribution in #65

Full Changelog: v8.2.0...v8.3.0

v8.2.0 - YOLOv8-World and YOLOv9-C/E Models

17 Apr 05:32
2f4d518
Compare
Choose a tag to compare

Ultralytics v8.2.0 Release Notes

Introduction

Ultralytics is excited to announce the v8.2.0 release of YOLOv8, comprising 277 merged Pull Requests by 32 contributors since our last v8.1.0 release in January 2024, marking another milestone in our journey to make state-of-the-art AI accessible and powerful. This release brings a host of new features, performance optimizations, and expanded integrations, reflecting our commitment to continuous improvement and innovation. πŸŒπŸš€

Ultralytics v8.2.0 Key Highlights

  • New Models: Introduced support for YOLOv8-World, YOLOv8-World-v2 (by @Laughing-q in PR #9268), YOLOv9-C, YOLOv9-E (by @Laughing-q in PR #8571), and YOLOv9 Segment models (by @Burhan-Q in PR #9296), expanding the versatility of the Ultralytics platform.
  • New Features: Added distance calculation in vision-eye, per-class object counting (by @RizwanMunawar in PR #9443), and queue management utilities (by @RizwanMunawar in PR #9494), enhancing the functionality and applicability of YOLOv8.
  • Performance Optimizations: Achieved 40% faster ultralytics imports (by @glenn-jocher in PR #9547), faster batch same_shapes, and immediate checkpoint serialization (by @glenn-jocher in PR #9437), further optimizing the efficiency of the framework.
  • Enhanced Export Capabilities: Improved export support, including OpenVINO 2023.3 updates (by @adrianboguszewski in PR #8417), TensorRT 10 support (by @Burhan-Q in PR #9516), and fixes for TFLite, ONNX, and OpenVINO exports.
  • Documentation Expansion: Significantly expanded the documentation with new guides, integration pages for TorchScript, TFLite, NCNN, PaddlePaddle, TF GraphDef, TF SavedModel, TF.js (by @abirami-vina in multiple PRs), and updates to existing pages, providing comprehensive resources for users.
  • Training Enhancements: Introduced YOLO-World training support (by @Laughing-q in PR #9268), fixed learning rate issues (by @Laughing-q in PR #9468), and improved robustness for stopping and resuming training (by @glenn-jocher in PR #9384).
  • Platform Support: Added support for NVIDIA Jetson (by @lakshanthad in PR #9484), Raspberry Pi (by @lakshanthad in PR #8828), and Apple M1 runners for tests and benchmarks (by @glenn-jocher in PR #8162), expanding the usability of YOLOv8 across various platforms.
  • CI/CD Improvements: Enhanced Ultralytics Actions using OpenAI GPT-4 for PR summaries (by @pderrenger in PR #7867) and introduced self-hosted Raspberry Pi 5 CI (by @lakshanthad in PR #8828), streamlining the development and testing processes.
  • Bug Fixes: Resolved various issues related to model loading, inference, plotting, and exports, ensuring a smoother user experience.
  • Community Contributions: Welcomed contributions from 31 new contributors, reflecting the growing engagement and collaborative spirit within the Ultralytics community.

Summary

Ultralytics v8.2.0 represents a significant leap forward, introducing new models, features, and optimizations while expanding platform support and integration capabilities. We extend our gratitude to our dedicated users and contributors for their invaluable support and contributions. As we continue to push the boundaries of AI and computer vision, we look forward to the exciting possibilities and advancements that lie ahead! πŸŒŸπŸš€πŸŽ‰

What's Changed

Read more

v8.1.0 - YOLOv8 Oriented Bounding Boxes (OBB)

10 Jan 03:09
6146984
Compare
Choose a tag to compare

Ultralytics v8.1.0 Release Notes

Introduction

Ultralytics proudly announces the v8.1.0 release of YOLOv8, celebrating a year of remarkable achievements and advancements. This version continues our commitment to making AI technology accessible and powerful, reflected in our latest breakthroughs and improvements.

2023 in Review

  • Record-Breaking Engagement: Over 20 million downloads of the Ultralytics package, with 4 million in December alone! πŸ“ˆ
  • Massive Model Training: An incredible 19 million YOLOv8 models were trained in 2023, showing the widespread adoption and versatility of our platform. 🌐
  • Diverse Model Usage: 64% of these models were for object detection, 20% for instance segmentation, 15% for pose estimation, and 1% for image classification. πŸ“Š
  • Expanding Global Reach: YOLOv8 reached 5 million users in 2023 and was run in 15 billion inference jobs across various industries, showcasing its real-world impact. 🌍
  • Documentation in Multiple Languages: Our docs are now available in 11 languages, catering to our diverse global community. πŸ“š

Ultralytics v8.1.0 Key Highlights

  • YOLOv8 OBB Models: The introduction of Oriented Bounding Box models in YOLOv8 marks a significant step in object detection, especially for angled or rotated objects, enhancing accuracy and reducing background noise in various applications such as aerial imagery and text detection.
  • Segmentation Support & Enhancements: Enhanced segmentation capabilities offer more precise image analysis, with improved classification augmentations integrated into Ultralytics training pipelines.
  • Performance Optimizations: Since our initial release last year we've focused on optimizing every aspect of the YOLOv8 framework, including training, validation, inference, and export, ensuring speed and efficiency without compromising performance.
  • Enhanced Model Architecture & Training Features: Incremental updates in model architecture, training features, and dataset support, including integration with Open Images V7 dataset and improved image classification models.
  • API and CLI Improvements: Enhanced user experience with refined API and CLI, including the Ultralytics Explorer tool for advanced dataset exploration and interaction.
  • PaddlePaddle, NCNN, PNNX, TensorRT & Other Integrations: Strengthened integration with multiple other platforms, offering users more deployment flexibility and compatibility for YOLOv8 users.
  • Diverse Contributions & Ultralytics HUB Evolution: The integration of over 1000 pull requests by 230 contributors and the growth of Ultralytics HUB, with it's own series of version updates, highlights the community's vital role in the development of YOLOv8.

Community Engagement and Support

  • Expanding Documentation: Our documentation now spans 11 languages, with over 200 pages, providing comprehensive guides for various real-world applications.
  • Custom-Trained YOLOv8 Models: With the ability to train models on custom data, 19 million YOLOv8 models were trained in 2023 alone, catering to diverse needs across object detection, segmentation, pose estimation, and image classification.
  • User Contributions: We encourage and appreciate user-contributed examples and stories, showcasing the versatility and real-world impact of YOLOv8.

Summary

Ultralytics v8.1.0 is a testament to a year of innovation, with the integration of Oriented Object Detection, enhanced classification models, and a strong focus on user experience and community engagement. We thank our users and contributors for their invaluable support and look forward to another year of groundbreaking advancements in the field of AI and computer vision in 2024! πŸŒŸπŸš€πŸŽ‰

What's Changed

New Contributors

Full Changelog: v0.0.0...v8.1.0

Initial Release

02 Jan 11:48
Compare
Choose a tag to compare

Ultralytics Assets v0.0.0 Release Notes

Introduction

Ultralytics is proud to announce the initial v0.0.0 release of our assets repository, establishing a centralized location for AI models, datasets, and other resources crucial to our computer vision and machine learning projects. This repository will serve as the backbone for managing and distributing the assets that power Ultralytics' cutting-edge AI solutions. πŸš€πŸ”¬

Ultralytics Assets v0.0.0 Key Highlights

  • Repository Structure: Implemented a well-organized directory structure for effortless navigation and asset management, including separate folders for models, datasets, and auxiliary resources.
  • Initial Model Collection: Uploaded a set of pre-trained YOLO models, including YOLOv5 and YOLOv8 variants, providing a strong foundation for object detection and image segmentation tasks.
  • Sample Datasets: Included a selection of curated datasets for testing, demonstration, and benchmarking purposes, showcasing the capabilities of Ultralytics' models.
  • Version Control: Established a robust versioning system for assets, ensuring users can access specific versions of models and datasets for reproducibility and consistency in their projects.
  • Documentation: Created initial README files and documentation to guide users on how to access and utilize the assets effectively within their Ultralytics-based projects.
  • Integration with Main Repository: Set up necessary links and references to seamlessly integrate this assets repository with the main Ultralytics project, facilitating easy access to the latest resources.
  • License Information: Clearly defined licensing terms for all included assets, promoting transparency and proper usage guidelines for the community.

Summary

Ultralytics Assets v0.0.0 marks the beginning of a structured approach to managing our AI resources. This repository will play a crucial role in supporting the Ultralytics ecosystem, enabling researchers, developers, and AI enthusiasts to access state-of-the-art models and datasets easily. As we continue to expand our collection of assets, we look forward to fostering innovation and advancing the field of computer vision and machine learning. We welcome contributions from the community and are excited to see how these assets will be utilized in various projects and applications! πŸŒŸπŸ€–πŸŽ‰

Full Changelog: https://github.com/ultralytics/assets/commits/v0.0.0