YOLOv5
Ultralytics YOLOv5 is a PyTorch-based computer-vision model for object detection, instance segmentation and image classification, mature and widely deployed after years of production use.
It can run inference through PyTorch Hub or a bundled detection script that downloads the latest release automatically, and supports training on a custom dataset with multi-GPU acceleration to speed up longer runs. Trained models export to formats including ONNX, CoreML, TensorRT and TFLite for deployment, including on NVIDIA Jetson devices, and documentation covers custom-data training tips, multi-GPU training, model ensembling and test-time augmentation.
Ultralytics recommends its newer, actively developed ultralytics package for projects that need newer architectures or additional tasks such as pose estimation and oriented object detection, while continuing to support YOLOv5 for existing users who rely on it.
Python, AGPL-3.0 licensed, with a separate enterprise license available on request.