YOLO

#29 v Multimodální modely

Ultralytics · 2× · naposledy 02. 9. 2026

19

Momentum

YOLO (You Only Look Once) is a family of real-time computer vision models developed by Ultralytics for object detection, instance segmentation, pose estimation, and other vision tasks. The current flagship generation is YOLO26, released in January 2026, featuring native end-to-end Inference without separate NMS post-processing and optimized performance for edge devices. The model is distributed as an Open Source Python package under AGPL-3.0 license as well as through a commercial Enterprise license; it is primarily a visual (image/video) recognition model, not a classical multimodal text-image LLM with Token-based API billing.

Zdroje (2)