# New Release: Ultralytics v8.4.172 **URL:** \ **Category:** Discussion\ **Tags:** releases, announcements, ultralytics-official\ **Created:** [October 3, 2026, 2:32pm UTC](https://community.ultralytics.com/t/new-release-ultralytics-v8-4-172/2249 "2026-10-03T14:32:37Z")\ **Posts on this page:** 1\ **Page:** 1
**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 3, 2026, 2:32pm UTC](https://community.ultralytics.com/t/new-release-ultralytics-v8-4-172/2249/1 "2026-10-03T14:32:37Z")
# ๐Ÿš€ Ultralytics v8.4.172: Rotation-Aligned OBB Crops and More Reliable Workflows **Ultralytics YOLO v8.4.172 brings upright oriented-box crops and reliability improvements across training, export, datasets, and deployment.** The [v8.4.172 release](https://github.com/ultralytics/ultralytics/releases/tag/v8.4.172) also expands Python support and refreshes Ultralytics Platform documentation. ## โœ‚ Highlight: Rotation-Aligned OBB Crops - [PR #26486](https://github.com/ultralytics/ultralytics/pull/26486) by [@Nikhi00718](https://github.com/Nikhi00718) adds rotation-aligned crops to `save_crop=True` and `ObjectCropper`. Crops are now upright and aligned to each oriented bounding box, with black padding where boxes extend beyond the image. ## ๐Ÿง  Model and Training Fixes - [PR #26483](https://github.com/ultralytics/ultralytics/pull/26483) by [@ArgoHA](https://github.com/ArgoHA) fixes AIFI positional embeddings for non-square feature maps. - [PR #26475](https://github.com/ultralytics/ultralytics/pull/26475) by [@Nikhi00718](https://github.com/Nikhi00718) computes RT-DETR Hungarian matching costs in float32, preventing AMP precision errors from affecting training assignments. - [PR #26487](https://github.com/ultralytics/ultralytics/pull/26487) by [@raimbekovm](https://github.com/raimbekovm) restores the `optimizer=auto` choice and early-stopping state when resuming training. - [PR #26477](https://github.com/ultralytics/ultralytics/pull/26477) by [@cainiao33](https://github.com/cainiao33) fixes corrupted dataset caches, recovered-epoch restarts after NaNs, YOLOE prediction state and batching, RT-DETR `save_txt`, and tracker, configuration, and solution issues. ## ๐Ÿ—‚ Dataset, Export, and Output Improvements - [PR #26497](https://github.com/ultralytics/ultralytics/pull/26497) by [@raimbekovm](https://github.com/raimbekovm) enables classification datasets to load every supported image format and refresh stale disk caches. - [PR #26507](https://github.com/ultralytics/ultralytics/pull/26507) by [@JaviChulvi](https://github.com/JaviChulvi) fixes OpenVINO INT8 calibration when the dataset is smaller than the requested batch. - [PR #26494](https://github.com/ultralytics/ultralytics/pull/26494) by [@raimbekovm](https://github.com/raimbekovm) replaces outputs from previous prediction and validation runs and gives each video its own output name. - [PR #26474](https://github.com/ultralytics/ultralytics/pull/26474) by [@Vaishnavi220506](https://github.com/Vaishnavi220506) fixes integer crop coordinates and JPEG output collisions. - [PR #26508](https://github.com/ultralytics/ultralytics/pull/26508) by [@wizzseen](https://github.com/wizzseen) fixes `yolo_bbox2segment` background-first detection and duplicate outputs on reruns, while adding support for JPEG and uppercase mask filenames. ## ๐Ÿ Deployment and Maintenance - [PR #26510](https://github.com/ultralytics/ultralytics/pull/26510) by [@glenn-jocher](https://github.com/glenn-jocher) expands Python 3.14 support across the package and several Docker and export environments. The Python export image remains on Python 3.13 for dependency compatibility. - [PR #26473](https://github.com/ultralytics/ultralytics/pull/26473) by [@glenn-jocher](https://github.com/glenn-jocher) fixes NaN convolution outputs caused by OpenBLAS 0.3.8 in the JetPack5 Docker image. - [PR #26471](https://github.com/ultralytics/ultralytics/pull/26471) by [@glenn-jocher](https://github.com/glenn-jocher) removes `cloudpickle` and `filelock` from base dependencies and adds safeguards against new base dependencies. - [PR #26481](https://github.com/ultralytics/ultralytics/pull/26481) by [@glenn-jocher](https://github.com/glenn-jocher) removes redundant Lost-state checks in ByteTrack and FastTracker second association. - [PR #26476](https://github.com/ultralytics/ultralytics/pull/26476) by [@JaviChulvi](https://github.com/JaviChulvi) uses pre-cached assets in COCO evaluation and CLI solutions tests. - [PR #26502](https://github.com/ultralytics/ultralytics/pull/26502) by [@glenn-jocher](https://github.com/glenn-jocher) restores the shared CI asset cache in Docker tests without including it in images. - [PR #26472](https://github.com/ultralytics/ultralytics/pull/26472) by [@glenn-jocher](https://github.com/glenn-jocher) updates Prettier from 3.8.5 to 3.9.9. - [PR #26501](https://github.com/ultralytics/ultralytics/pull/26501) by [@dependabot[bot]](https://github.com/dependabot) switches example requirements to `>=` version floors to avoid repeated dependency pin bumps. ## ๐Ÿ“š Documentation Updates - [PR #26467](https://github.com/ultralytics/ultralytics/pull/26467) by [@raimbekovm](https://github.com/raimbekovm) aligns validation, Edge TPU export, and Platform dataset, Agents API, and billing documentation with current behavior. - [PR #26485](https://github.com/ultralytics/ultralytics/pull/26485) by [@pderrenger](https://github.com/pderrenger) refreshes Ultralytics Platform documentation, including current workflows, Agents management, and dataset version comparisons. - [PR #26488](https://github.com/ultralytics/ultralytics/pull/26488) by [@raimbekovm](https://github.com/raimbekovm) names LabelMe and Labelbox in Platform import format lists and corrects Label Studio OBB and VOC notes. - [PR #26489](https://github.com/ultralytics/ultralytics/pull/26489) by [@lakshanthad](https://github.com/lakshanthad) moves AMD documentation into Deployment Integrations. ## ๐Ÿ™Œ Community Thanks Thank you to the YOLO community and the Ultralytics team for these improvements! A special welcome to **[@ArgoHA](https://github.com/ArgoHA)**, whose AIFI fix marks their first contribution. ## ๐Ÿ”ง Try It and Share Your Feedback Install this release with: ```bash pip install --upgrade ultralytics==8.4.172 ``` Explore the [full changelog](https://github.com/ultralytics/ultralytics/compare/v8.4.171...v8.4.172) for all changes, then try the update in your workflows. **How are the new OBB crops working for you?** Share your results, questions, or any issues belowโ€”your feedback helps improve YOLO for everyone.