Published November 16, 2026
| Version v8.4.215
Software
Open
Ultralytics YOLO
Description
🌟 Summary
YOLO 8.4.215 improves segmentation memory efficiency and fixes several data, tracking, and reliability issues—without changing model architectures or default prediction settings.
📊 Key Changes
- 🧠 Lower-memory segmentation workflows (main release change): Plotting, saving, displaying, moving, and prompting segmentation masks can now work with masks stored inside their bounding boxes instead of building a full image-sized mask stack. Outputs and defaults remain unchanged; requesting
masks.datastill creates the dense masks. - 🏷️ More reliable automatic annotations: Class labels stay matched to their detections when SAM filters out some masks.
- 🎥 More complete video predictions: Video loading now reads frames until decoding ends, rather than stopping at a possibly inaccurate frame count in the file metadata.
- 🧭 More dependable tracking fallback: If ECC camera-motion estimation fails, the tracker now retains the identity transform instead of potentially using a partially modified estimate.
- 📌 Accurate checkpoint provenance: Checkpoints record the Git commit present when the run starts, even if the checkout changes later.
- 🧠 Updated Brain Tumor dataset: Now contains 1,115 brain MRI images, one
tumorclass, and a patient-based validation split. Documentation clarifies that it is for research and education, not clinical diagnosis. - 🛠️ Improved CI resilience and documentation: CI retries a failed asset-cache restore once, and several documentation page titles are clearer and more distinct.
🎯 Purpose & Impact
- 🚀 Segmentation workflows can avoid the memory and processing cost of a dense stack when they only need masks within detected boxes. This can help with larger images or many detections while preserving existing outputs.
- ✅ Annotation, video, and tracking fixes reduce the chance of incorrect labels, silently skipped frames, or unreliable camera-motion compensation.
- 🔍 Checkpoint metadata makes it easier to identify the exact code used for a training run.
- 📚 The revised dataset split and guidance make the Brain Tumor dataset easier to use appropriately, while its limitations remain important: it has no tumor-free examples and is not suitable for clinical diagnosis.
- 🔧 No new model architecture or change to prediction defaults is introduced in this release.
What's Changed
- Record the launch commit in checkpoint git metadata by @fcakyon in https://github.com/ultralytics/ultralytics/pull/27213
- Fix ECC identity fallback after failed motion estimation by @AhmetMert01 in https://github.com/ultralytics/ultralytics/pull/27219
- Give the settings docs page a distinct title by @miles-deans-ultralytics in https://github.com/ultralytics/ultralytics/pull/26595
- Keep auto-annotation classes aligned when SAM drops masks by @aswanth-07 in https://github.com/ultralytics/ultralytics/pull/27214
- Shorten front-matter titles on 25 docs pages by @miles-deans-ultralytics in https://github.com/ultralytics/ultralytics/pull/26316
- Read video frames until capture ends by @Vaishnavi220506 in https://github.com/ultralytics/ultralytics/pull/27215
- fix: retry failed CI asset restores once by @fcakyon in https://github.com/ultralytics/ultralytics/pull/27220
- Update brain-tumor dataset to one tumor class with a patient-id validation split by @raimbekovm in https://github.com/ultralytics/ultralytics/pull/27218
ultralytics 8.4.215Plot, convert and prompt segmentation masks without the dense mask stack by @glenn-jocher in https://github.com/ultralytics/ultralytics/pull/27216
Full Changelog: https://github.com/ultralytics/ultralytics/compare/v8.4.214...v8.4.215
Notes
Files
ultralytics/ultralytics-v8.4.215.zip
Additional details
Related works
- Is supplement to
- Software: https://github.com/ultralytics/ultralytics/tree/v8.4.215 (URL)
Software
- Repository URL
- https://github.com/ultralytics/ultralytics