# New Release: Ultralytics v8.3.157

**URL:** <https://community.ultralytics.com/t/new-release-ultralytics-v8-3-157/1166>\
**Category:** Discussion\
**Tags:** releases, announcements, ultralytics-official\
**Created:** [June 20, 2025, 1:13pm UTC](https://community.ultralytics.com/t/new-release-ultralytics-v8-3-157/1166 "2025-06-20T13:13:27Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![glenn-jocher](https://sea1.discourse-cdn.com/flex001/user_avatar/community.ultralytics.com/glenn-jocher/32/83_2.png) [@glenn-jocher](https://community.ultralytics.com/u/glenn-jocher)\
**Post date:** [June 20, 2025, 1:13pm UTC](https://community.ultralytics.com/t/new-release-ultralytics-v8-3-157/1166/1 "2025-06-20T13:13:28Z")

</div>

# 🚀 Ultralytics v8.3.157 Release: Lightning-Fast Evaluation & Enhanced Usability!

## 🌟 Summary

We’re excited to announce Ultralytics **v8.3.157** —a major update focused on making your model development experience faster, easier, and more flexible! This release brings **dramatically improved evaluation speeds** , smarter segmentation & pose handling, greater dataset compatibility, and even better documentation to accelerate your workflows. Whether you’re a newcomer or a seasoned YOLO developer, you’ll notice the difference right away. 🚀

* * *

## 🏆 New Features & Major Improvements

### Major Evaluation Upgrade

- **Faster COCO/LVIS Evaluation:** We’ve integrated the [`faster-coco-eval`](https://github.com/ultralytics/ultralytics/pull/17020) library by [@MiXaiLL76](https://github.com/MiXaiLL76) as a drop-in replacement for `pycocotools`. Enjoy up to **4.5x faster validation** on COCO and LVIS datasets! Benchmarking and model iteration just got a huge boost.

### Segmentation & Pose Enhancements

- **Mask Post-Processing:** No more custom mask headaches! Changes by [@Laughing-q](https://github.com/Laughing-q) [fix mask shape mismatches](https://github.com/ultralytics/ultralytics/pull/21083), making YOLO segmentation more robust—especially with custom pipeline modifications.
- **Reliable Keypoints:** Pose models now treat keypoint labels and predictions in a consistent format, reducing errors and making training and inference more dependable.

### Dataset Flexibility

- **Grounding Dataset Improvements:** Thanks to [@mohiuddin-khan-shiam](https://github.com/mohiuddin-khan-shiam), label verification is now smarter: known datasets are strictly validated, but custom datasets load more gracefully. No more loading interruptions!

* * *

## 📚 Documentation & UX Improvements

- **Video Tutorial for Beginners:** A new [dog pose estimation tutorial](https://youtu.be/J-RH22rwx1A) added by [@RizwanMunawar](https://github.com/RizwanMunawar) is now featured in the docs—perfect for quick onboarding and visual learners!
- **Clarity in Dataset Guides:** [@glenn-jocher](https://github.com/glenn-jocher) [updated GQA dataset links](https://github.com/ultralytics/ultralytics/pull/21087) for YOLOE and YOLO-World, ensuring you always have the latest references.
- **Easier Data Augmentation:** Solutions docs now include a direct button linking to the [YOLO augmentation guide](https://github.com/ultralytics/ultralytics/pull/21061) (thanks [@RizwanMunawar](https://github.com/RizwanMunawar)).

* * *

## 💡 Dependency Updates & Fixes

- **Security & Compatibility:**

- **Better Error Handling:**

* * *

## 👏 New Contributors

Big thanks to our newest contributors for their valuable additions:

- [@WillieMaddox](https://github.com/WillieMaddox)
- [@MiXaiLL76](https://github.com/MiXaiLL76)
- [@mohiuddin-khan-shiam](https://github.com/mohiuddin-khan-shiam)

You’re fueling innovation for the whole YOLO community!

* * *

## 🚦 How to Get Started

Update now to the latest version (`pip install -U ultralytics`) and experience faster evaluation, improved segmentation and pose workflows, and slicker documentation.  
We encourage all users to [review the full changelog](https://github.com/ultralytics/ultralytics/compare/v8.3.156...v8.3.157), try out the new features, and let us know about your experience or suggestions right here in the forum.

* * *

## 📢 Feedback Welcome

Your input drives our roadmap! Please share your thoughts, issues, or success stories—every bit helps us build better tools for you and the whole computer vision community.

**Check out the full release announcement:** [Ultralytics v8.3.157 Release Notes](https://github.com/ultralytics/ultralytics/releases/tag/v8.3.157)

Happy training! 🚀  
— The Ultralytics Team & YOLO Community