New Release: Ultralytics v8.3.131

Discussion
,
1

:rocket: Ultralytics v8.3.131 Release Announcement

:glowing_star: Summary

We’re excited to announce the release of Ultralytics v8.3.131! This update introduces powerful enhancements for grayscale image support, robust handling of large and transparent images, a production-ready YOLO11 C++ Triton Inference example, and several important workflow and licensing updates. These features help make the platform even more flexible, reliable, and ready for complex, real-world use cases!

:pushpin: Quick Overview

  • Seamless grayscale image inference and visualization
  • No more size limits for extremely large images
  • Smart auto-conversion of transparent images
  • YOLO11 deployment on NVIDIA Triton Inference Server in C++ (Supports FP16)
  • Licensing and contributing workflow clarified

:tada: New Features

Grayscale Image Support

  • What’s new? Easily work with grayscale (single-channel) images for inference and visualization.
  • Details:
    • Added support for the channels parameter in data loaders
    • Enhanced image reading and plotting functions for graceful handling of grayscale images
  • See implementation by @Laughing-q

YOLO11 Triton C++ Example

  • What’s new? A complete C++ example for deploying YOLO11 models on NVIDIA Triton Inference Server, offering efficient FP16 inference and built-in result visualization.
  • Contributed by @asdemirel (First-time contributor—thank you!)

:framed_picture: Improvements

Large & Transparent Image Handling

  • No Limits on Image Size: The default pixel limit in image processing (PIL) has been removed—process huge images with ease!
  • Transparent Images Made Simple: Images with transparency (e.g. PNG + alpha) are now auto-converted to RGB before saving, preventing errors.

:hammer_and_wrench: Licensing and Workflow Tweaks

:busts_in_silhouette: New Contributors

A warm welcome to @asdemirel for their first contribution—your help bringing the YOLO11 Triton C++ example makes a difference!

:link: More Details & Full Changelog

:raising_hands: Get Involved

We encourage you to upgrade to v8.3.131, try the new features—especially if you work with grayscale or large/transparent images—and experiment with the C++ Triton deployment for YOLO11. Your feedback helps us shape a tool that serves the whole YOLO community!

  • Find a bug? Got a suggestion? Let us know in Ultralytics Discussions.
  • Thanks to our contributors and community for driving these improvements forward!

Happy building and detecting,
— Glenn & the Ultralytics Team