New Release: Ultralytics v8.3.4

[glenn-jocher]

Announcing Ultralytics v8.3.4 Release!

Summary

We’re excited to announce the release of Ultralytics v8.3.4! This update brings flexible enhancements to project dependencies, improved usability, and performance optimizations. Dive into the details below to see what’s new and how it can benefit your projects.

Key Changes

Numpy Dependency Update

• Improvement: Removed the upper version limit on numpy, now supporting any version above 1.23.0.
• Impact: Enhances compatibility with newer versions, simplifying dependency management.

CI Schedule Adjustment

• Change: Daily build and test schedule moved to 08:00 UTC.
• Impact: Better synchronization with daily resource availability, optimizing workflow efficiency.

Improved Example Documentation

• Addition: Detailed example for YOLOv8 using OpenVINO in C++.
• Impact: Potentially enhances inference speed and supports various hardware profiles.

Enhanced Error Handling in Export Formats

• Feature: Auto-correction mechanism for export format inputs.
• Impact: Provides guidance on available formats, reducing confusion and errors.

Device Selection Robustness

• Improvement: Enhanced parsing logic for device identifiers, including multiple GPUs.
• Impact: Ensures more reliable device selection.

Documentation and Video Guides

• Update: Added video tutorials on MobileSAM and SAM 2 models, with updated image links.
• Impact: Aids both new and experienced users in leveraging advanced model functionalities.

Purpose & Impact

• Compatibility & Flexibility: Unpinning numpy boosts compatibility and reduces conflicts.
• Optimized Workflow: CI schedule change aligns with developer workflows.
• User Experience: Improved error handling enhances user guidance.
• Accessibility & Education: Video tutorials foster a wider user base.
• Performance Optimization: OpenVINO example supports Intel platforms.

What’s Changed

• Fix Raspberry Pi CI by @glenn-jocher
• Update scheduled CI to 08:00 UTC by @glenn-jocher
• Add YouTube link to docs by @RizwanMunawar
• Fix overridden train_args by @Laughing-q
• Add contributor name for OpenVINO example by @rlggyp
• Auto-correct for export formats by @RizwanMunawar
• Robust device selection by @glenn-jocher
• Improve scikit-learn support for exports by @glenn-jocher
• Fixing predict docs by @ambitious-octopus
• SAM: Fix labels not being used in predict mode by @Y-T-G
• Minor fix in docs for python admonition by @jk4e
• Update image URLs in mobile-sam.md and sam-2.md by @RizwanMunawar
• Replace YOLO11-seg table with macro by @jk4e
• Unpin numpy<2.0.0 by @glenn-jocher

Full Changelog: v8.3.3…v8.3.4

Try It Out!

We encourage you to explore the new features and improvements in v8.3.4. Your feedback is invaluable to us, so please share your thoughts and experiences. Happy experimenting!

Release URL: Ultralytics v8.3.4