pypi ultralytics 8.3.135
v8.3.135 - `ultralytics 8.3.135` TF.js fix with `onnx>=1.12.0,<1.18.0` pin (#20640)

latest releases: 8.3.194, 8.3.193, 8.3.192...
3 months ago

🌟 Summary

This release improves model export reliability, speeds up font handling, enhances pose documentation, and streamlines video processing for a smoother user experience. πŸš€

πŸ“Š Key Changes

  • πŸ›‘οΈ Export Stability: ONNX version is now limited to >=1.12.0 and <1.18.0 for TensorFlow SavedModel exports, preventing compatibility issues.
  • ⚑ Faster Predictions: Font checking is now cached, making repeated predictions much fasterβ€”especially when using yolo predict.
  • 🎬 Smarter Video Processing: The "quit" option for video processing only appears when results are displayed, improving efficiency for headless or automated tasks.
  • πŸ“ Clearer Pose Metrics: Pose task documentation now includes detailed explanations of new evaluation metrics.
  • πŸ€– YOLOE Prompt Handling: YOLOE now reuses prompts when a reference image is provided, simplifying multi-image predictions.

🎯 Purpose & Impact

  • πŸš€ More Reliable Model Exports: Users exporting models to TensorFlow or TFLite will experience fewer errors and smoother conversions.
  • ⚑ Faster Workflows: Predicting with YOLO models is now significantly quicker, especially for repeated or batch operations.
  • πŸ–₯️ Better Automation: Video processing is more efficient and less confusing for users running scripts without a display.
  • πŸ“ˆ Improved Understanding: Enhanced documentation helps users interpret pose detection results more easily.
  • 🀝 Simplified Usage: YOLOE’s improved prompt logic makes it easier to work with multiple images and reference prompts.

Overall, this update delivers a faster, more stable, and more user-friendly experience for Ultralytics users! 🌟

What's Changed

New Contributors

Full Changelog: v8.3.134...v8.3.135

Don't miss a new ultralytics release

NewReleases is sending notifications on new releases.