# New Release: Ultralytics v8.4.191

**URL:** <https://community.ultralytics.com/t/new-release-ultralytics-v8-4-191/2286>\
**Category:** Discussion\
**Tags:** ultralytics-official, releases, announcements\
**Created:** [October 28, 2026, 1:06pm UTC](https://community.ultralytics.com/t/new-release-ultralytics-v8-4-191/2286 "2026-10-28T13:06:44Z")\
**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:** [October 28, 2026, 1:06pm UTC](https://community.ultralytics.com/t/new-release-ultralytics-v8-4-191/2286/1 "2026-10-28T13:06:44Z")

</div>

# 🚀 Ultralytics v8.4.191: Broader RKNN INT8 Support and Smarter Tracking

**Ultralytics YOLO v8.4.191 is here!** This release expands RKNN INT8 exports across vision tasks, reduces repeated work in segmentation training and parking management, improves tracking associations, and preserves prediction-source ordering.

Explore the [v8.4.191 release notes](https://github.com/ultralytics/ultralytics/releases/tag/v8.4.191) for the full release details.

## 🌟 Main Update: RKNN INT8 Export

[PR #26904](https://github.com/ultralytics/ultralytics/pull/26904) by [@Bovey0809](https://github.com/Bovey0809) enables INT8 export for **detection, segmentation, pose, and oriented bounding box (OBB) models**.

The updated export splits inference work to address accuracy loss from INT8 box decoding:

- **NPU:** Runs the head’s output convolutions.
- **CPU:** Runs box decoding in floating point using a new `tail.onnx` file.
- **FP16-capable targets:** Use RKNN’s `auto_hybrid` option to keep high-error layers in floating point.

### ⚠ RKNN Deployment Notes

- **Re-export older INT8 models** to avoid incorrect coordinate scaling.
- The decode tail requires **ONNX Runtime** , which Ultralytics installs automatically when needed.
- Custom runtimes must run `tail.onnx` or implement equivalent decoding.
- INT8 accuracy has been tested in the **RKNN simulator** , but **has not yet been validated on Rockchip hardware**.

## ⚡ Performance Improvements

- **Faster segmentation training:** [PR #26914](https://github.com/ultralytics/ultralytics/pull/26914) by [@raimbekovm](https://github.com/raimbekovm) finds foreground anchors once across the batch instead of repeating the work per image. Losses and gradients remain unchanged, with modest training-step improvements reported on a Tesla T4.
- **More efficient parking management:** [PR #26910](https://github.com/ultralytics/ultralytics/pull/26910) by [@raimbekovm](https://github.com/raimbekovm) computes each tracked car’s center once per frame rather than once per parking-space comparison.

## 🛠 Tracking and Prediction Fixes

- **More robust OC-SORT matching:** [PR #26912](https://github.com/ultralytics/ultralytics/pull/26912) by [@mottopanikeiku](https://github.com/mottopanikeiku) measures detection direction over the track’s recent `delta_t`-frame window, making associations less sensitive to bounding-box jitter.
- **Prediction manifests keep your order:** [PR #26903](https://github.com/ultralytics/ultralytics/pull/26903) by [@MohammadHijjawi97](https://github.com/MohammadHijjawi97) preserves the listed order of images and videos in TXT and CSV sources instead of sorting them alphabetically—especially useful for frame sequences.

## 📚 Platform and Guide Updates

- **Annotation provider documentation:** [PR #26902](https://github.com/ultralytics/ultralytics/pull/26902) by [@glenn-jocher](https://github.com/glenn-jocher) adds MiniMax to the paid annotation providers documented for Ultralytics Platform.
- **Clearer import and label guidance:** [PR #26906](https://github.com/ultralytics/ultralytics/pull/26906) by [@raimbekovm](https://github.com/raimbekovm) explains saved Roboflow and W&B import keys, along with OBB labels in detection datasets.
- **Updated training content:** [PR #26908](https://github.com/ultralytics/ultralytics/pull/26908) by [@RizwanMunawar](https://github.com/RizwanMunawar) refreshes the Model Training Tips guide with an updated YOLO26 batch-processing video.

## 🚀 Try It and Share Your Feedback

Upgrade to this release with:

```bash
pip install --upgrade ultralytics==8.4.191

```

Browse the [full changelog](https://github.com/ultralytics/ultralytics/compare/v8.4.190...v8.4.191) to review all changes.

**How does v8.4.191 work with your setup?** Share your results, questions, or issues below—especially if you’re testing RKNN exports on Rockchip hardware. Thanks to all the contributors who helped make this release possible! 🙌