t=ultralytics/ultralytics:latest-jetson
sudo docker pull $t
sudo docker run -it --network host --runtime nvidia -v /home/agx/docker:/home/share --name train_yolov8 --shm-size=32g ultralytics/ultralytics:latest-jetson
–shm-size可以视情况更改
yolo segment train data=data.yaml model=yolov8n-seg.pt epochs=100 batch=64