ultralytics v8.4.138

v8.4.138

v8.4.138 - Fix legacy checkpoint and YOLO-World loading (#26032)

Changed 4
  • Optimize SAM, SAM2, and SAM3 image embeddings to compute in inference mode, reducing memory overhead during repeated inference
  • Move classification inference preprocessing to the inference device and perform in batches to reduce CPU overhead and improve throughput
  • Cap BoT-SORT global motion compensation corner detection at 400 points instead of 1,000 to reduce optical-flow computation
  • Document that quantize parameter may select or request different runtime precisions depending on export format
Fixed 7
  • Restore compatibility with legacy checkpoints created before v8.4.95 by recognizing loss and assignment classes stored in detection, classification, pose, segmentation, rotated-box, and keypoint components
  • Fix YOLO-World model and checkpoint loading issues
  • Prevent MuSGD optimizer crash with channels_last layout by replacing incompatible tensor flattening with a layout-safe operation
  • Prevent DDP failures when a mini-batch contains no assigned targets
  • Ensure detection, pose, OBB, and related model branches remain connected to the training graph when there are no positive samples
  • Correct multi-dataset tuning to preserve dataset names and iteration order across distributed workers
  • Record failed datasets with zero metrics instead of incomplete results in tuning histories

From ultralytics

🌟 Summary

Ultralytics v8.4.138 is a stability-focused release that restores compatibility with older checkpoints, fixes YOLO-World loading, and improves training, inference, tracking, tuning, and documentation reliability. 🛠️

📊 Key Changes

  • Legacy checkpoint loading fixed 🎯
    Restricted checkpoint loading now recognizes loss and assignment classes stored in checkpoints created before 8.4.95, including detection, classification, pose, segmentation, rotated-box, and keypoint-related components.

  • YOLO-World loading fixed 🌍
    Corrects package loading issues affecting YOLO-World models and related checkpoints.

  • MuSGD with channels_last no longer crashes ⚡
    Replaces an incompatible tensor flattening operation with a layout-safe one, allowing CUDA training with the MuSGD optimizer and automatic channels_last support to run correctly.

  • SAM feature extraction optimized 🧠
    SAM, SAM2, and SAM3 image embeddings are computed in inference mode, preventing unnecessary autograd graphs from being retained and reducing memory overhead during repeated inference.

  • Distributed training made more robust 🔧

    • Prevents DDP failures when a mini-batch contains no assigned targets.
    • Ensures detection, pose, OBB, and related model branches remain connected to the training graph even when there are no positive samples.

  • Tuning results corrected 📈
    Multi-dataset tuning now preserves dataset names and iteration order across distributed workers. Failed datasets are recorded with zero metrics instead of incomplete results, keeping tuning histories and fitness plots consistent.

  • Classification inference preprocessing improved 🚀
    More preprocessing work is moved to the inference device and performed in batches, which can reduce CPU overhead and improve classification throughput.

  • BoT-SORT tracking made faster 🏃
    Global motion compensation now caps corner detection at 400 points instead of 1,000, reducing optical-flow computation while retaining sufficient information for motion estimation.

  • Precision and quantization documentation clarified 📚
    Documentation now explains that quantize may select or request different runtime precisions depending on the export format. This avoids implying that every backend supports the same FP16, FP32, or INT8 behavior.

  • Documentation quality updates ✨
    Markdown tables were consistently formatted, the OBB navigation label was cleaned up, the TrackZone video was updated, and the YOLO26 CPU speed comparison now clearly identifies its YOLO26n-versus-YOLO11n ONNX baseline and hardware.

🎯 Purpose & Impact
  • More users can load existing models without retraining, especially those using checkpoints created with older Ultralytics versions. ✅
  • YOLO-World workflows become more dependable for users loading supported models and checkpoints.
  • Platform and cloud GPU training jobs are less likely to fail, particularly MuSGD runs using CUDA and channels_last.
  • Distributed training and hyperparameter tuning produce more trustworthy results, even with empty-target batches or failed datasets.
  • SAM inference can use less memory, which is helpful for long-running applications and repeated image embedding extraction.
  • Tracking pipelines may run with lower latency, especially in applications where BoT-SORT global motion compensation is a significant cost.
  • Users should review the updated quantization guidance before assuming that quantize=16 or quantize=32 produces the same computation precision across PyTorch, ONNX, OpenVINO, NCNN, MNN, Triton, and other backends. 🔍
What's Changed

Full Changelog: https://github.com/ultralytics/ultralytics/compare/v8.4.137...v8.4.138

View original

Upgraded? How did it go?

Discussion