Ultralytics YOLO

The Ultralytics package is the home of the modern YOLO family, bundling training, validation, inference, and export for real-time computer vision into one Python library and a single yolo command-line interface. It covers object detection, instance and semantic segmentation, image classification, pose estimation, oriented bounding boxes, and multi-object tracking, with pretrained weights spanning YOLOv5 through YOLO11 and newer generations in five sizes from nano to extra large. Models train on custom datasets with a few lines of Python and export to deployment formats such as ONNX, TensorRT, CoreML, and TFLite for edge and server inference on GPU or CPU. The code is open source under AGPL-3.0, with a separate enterprise license available for closed-source commercial products. It is one of the most widely used detection stacks in both industry and research, adopted by roboticists, ML engineers, and students who need accurate real-time vision without building a pipeline from scratch.