New Release: Ultralytics v8.3.132

:rocket: Ultralytics v8.3.132 Release: Smarter Weight Loading, New Indoor Dataset, and More!

:glowing_star: Summary

Ultralytics v8.3.132 is here! This release brings smarter pretrained model handling for multi-channel images, a brand new HomeObjects-3K indoor dataset, enhanced object counting (now with rotated boxes), and numerous improvements to segmentation and documentation. Whether you’re developing in smart home, robotics, or cutting-edge computer vision, this update makes Ultralytics YOLO even more robust and user-friendly!


:clipboard: New Features

Smarter Model Weight Loading

HomeObjects-3K Indoor Dataset

  • Introducing a new high-quality dataset with 12 common household objects—such as beds, sofas, and TVs—ideal for indoor detection in smart home, AR, and robotics projects.
  • New HomeObjects-3K dataset by @RizwanMunawar

Object Counting for Rotated Boxes (OBB)

Enhanced Segmentation Mask Workflows


:hammer_and_wrench: Improvements & Fixes


:bullseye: Purpose & Impact

  • Adapt Models to Any Dataset: Pretrained weights now integrate more smoothly with custom datasets, no matter the image channel configuration.
  • Expand to Indoor Applications: HomeObjects-3K dataset opens new possibilities for smart, AR, and robotics applications in households and indoor environments.
  • Precision with Rotated Objects: Reliable tracking and counting in industrial, aerial, and off-axis scenes with OBB support.
  • Better User Experience: Improved error messages, onboarding videos, and detailed logs smooth the path for both new and advanced users.
  • Professional and Consistent Branding: “Ultralytics YOLO” branding and cleanup across the board enhance project clarity.

:technologist: Try it Out!

You can install or upgrade to v8.3.132 today:

pip install -U ultralytics

Try out the new features, explore the HomeObjects-3K dataset, or adapt a pretrained model to your own multi-channel data. We hope these changes make your workflow smoother and your models even stronger!


:speech_balloon: Feedback Wanted!

Your feedback drives Ultralytics forward! Please share your experiences, feature requests, and questions right here in Discourse. Reporting issues or successes helps the community and the Ultralytics team improve with every release.

For a deep dive, see the full changelog and detailed GitHub release notes.

Thanks for being part of this journey—the progress is a collective achievement from the entire YOLO community and Ultralytics team!

Stay curious and keep building! :rocket: